0% found this document useful (0 votes)

378 views176 pages

GS Users Manual

Uploaded by

Andrei Segal

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

378 views176 pages

GS Users Manual

Uploaded by

Andrei Segal

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 176

GS User's Manual

Copyright © 2002 Sony Computer Entertainment Inc.

© 2002 Sony Computer Entertainment Inc.

Publication date: April 2002

Sony Computer Entertainment Inc.

1-1, Akasaka 7-chome, Minato-ku
Tokyo 107-0052 Japan

Sony Computer Entertainment America

919 East Hillsdale Blvd.
Foster City, CA 94404, U.S.A.

Sony Computer Entertainment Europe

30 Golden Square
London W1F 9LD, U.K.

The GS User’s Manual is supplied pursuant to and subject to the terms of the Sony Computer Entertainment
PlayStation® license agreements.

The GS User’s Manual is intended for distribution to and use by only Sony Computer Entertainment licensed
Developers and Publishers in accordance with the PlayStation® license agreements.

Unauthorized reproduction, distribution, lending, rental or disclosure to any third party, in whole or in part, of
this book is expressly prohibited by law and by the terms of the Sony Computer Entertainment PlayStation®
license agreements.

Ownership of the physical property of the book is retained by and reserved by Sony Computer Entertainment.
Alteration to or deletion, in whole or in part, of the book, its presentation, or its contents is prohibited.

The information in the GS User’s Manual is subject to change without notice. The content of this book is
Confidential Information of Sony Computer Entertainment.

® and PlayStation® are registered trademarks, and GRAPHICS SYNTHESIZERTM and

TM
EMOTION ENGINE are trademarks of Sony Computer Entertainment Inc. All other trademarks are property
of their respective owners and/or their licensors.

About This Manual

The "GS User's Manual" describes the functional specifications and command set of the Graphics Synthesizer.
For information on how to access the Graphics Synthesizer from the Emotion Engine, refer to the "EE User's
Manual".
- Chapter 1 "Overview" describes the features, block configuration, and performance characteristics of the
Graphics Synthesizer.
- Chapter 2 "Local Memory" describes the Graphics Synthesizer's local memory and the formats for data
stored in memory.
- Chapter 3 "Drawing Function" describes the drawing primitives, called GS primitives, and the drawing
environment. It also describes the stages of pixel generation based on the GS primitives and drawing
environment, and explains texture mapping, fogging, antialiasing, pixel test, alpha-blending, and dithering.
- Chapter 4 "Image Data Transmission" describes the functions of data transmission between the Graphics
Synthesizer and the Emotion Engine, and within local memory.
- Chapter 5 "CRTC" describes the video signal output functions and feedback input functions.
- Chapter 6 "Signal" describes the signal generation function.
- Chapter 7 "Registers" describes the registers accessible from the Emotion Engine.
- Chapter 8 "Details of GS Local Memory" provides detailed descriptions of local memory addressing.

Changes Since Release of 5th Edition

Since release of the 5th Edition of the GS User's Manual, the following changes have been made.
Note that each of these changes is indicated by a revision bar in the margin of the affected page.

Ch. 1: Overview
• A correction has been made to “Local Memory” in the GS functional block description on page 17.
• A correction has been made to the “Shading” column of the Processing Performance in Sprite table on
page 18.
• A correction has been made to the “Shading” column of the Processing Performance in Point table on
page 18.

Ch. 3: Drawing Function

• Figure 3-1 Block Diagram for Drawing Processing, on page 34, has been revised.
• A correction has been made to “Pixel Test” in the process descriptions on page 35.
• Corrections have been made to section 3.4.1. Outline of Texture Mapping Process, on page 50. The
texture page buffer was also added to Figure 3-9 Texture Mapping Process Flow.
• A new section, “Texture Page Buffer”, has been added to section 3.4.2. Texture Mapping Process Flow,
on page 51. This section provides details about the texture page buffer.
• Information about the texture coordinate range was added to section 3.4.4. Specification of Texture
Coordinate values, on page 52.
• A correction has been made to section 3.7. Pixel Test, on page 66. Further clarification about the
position of scissoring in the pixel pipeline was also added.
• A correction has been made to section 3.7.2. Alpha Test, on page 67.
• A correction has been made to section 3.7.3. Destination Alpha Test, on page 68.
• A correction has been made to section 3.7.4. Depth Test, on page 68. A note on hardware bug relating
to Z test modes was also added.

Ch. 4: Image Data Transmission

• Corrections have been made to the introduction for section 4. Image Date Transmission, on page 73.
• The procedure in section 4.2.2. Transmission from Local Buffer to Host, on page 77, has been revised.
• Corrections have been made to section 4.2.3. Transmission from Local Buffer to Local Buffer, on page
77.
• Section 4.2.4. Restrictions on Drawing after Transmission, has been added to page 78. This section
contains information about a hardware bug related to texture transmission and drawing on the same
page.

Ch. 5: CRTC
• Corrections have been made to the “NTSC” and “PAL” rows in the Corresponding Video Signal table
on page 84.

Ch. 6: Signal
• Corrections have been made and information has been added to section 6.2.1. SIGNAL Register, on
page 95.
• Corrections have been made and information has been added to section 6.2.2. FINSH Register, on page
95.

Ch. 7: Registers
• A reference note (*3) has been added to the BITBLTBUF register FIELD table on page 101.
• A correction has been made to the “Contents” column in the FBA_1/FBA_2 register FIELD table on
page 106.
• Information has been added to the notes for the RGBAQ register FIELD table on page 119.
• Information has been added to the “S” and “T” rows in the ST register FIELD table on page 123.
• A correction has been made to the “ZTE” row in the TEST_1/TEST_2 register FIELD table on page
124.
• Information has been added to the description of the BUSDIR (w) register on page 144.
• A correction has been made to the note for the CSR (r/w) register FIELD table on page 146.

Ch. 8: Details of GS Local Memory

• Corrections have been made and information added to “Column” in section 8.1. Memory Access Units,
on page 162.
• Information has been added to section 8.3.1. PSMCT32/PSMCT24/PSMZ32/PSMZ24, on page 164.
• A correction has been made to section 8.6. Pixel Storage Format Conversion, on page 172.
• A correction has been made to section 8.6.3. PSMT16, on page 175.

Glossary
Term Definition
EE Emotion Engine. CPU of the PlayStation 2.
EE Core Generalized computation and control unit of EE. Core of the CPU.
COP0 EE Core system control coprocessor.
COP1 EE Core floating-point operation coprocessor. Also referred to as FPU.
COP2 Vector operation unit coupled as a coprocessor of EE Core. VPU0.
GS Graphics Synthesizer.
Graphics processor connected to EE.
GIF EE Interface unit to GS.
IOP Processor connected to EE for controlling input/output devices.
SBUS Bus connecting EE to IOP.
VPU (VPU0/VPU1) Vector operation unit.
EE contains 2 VPUs: VPU0 and VPU1.
VU (VU0/VU1) VPU core operation unit.
VIF (VIF0/VIF1) VPU data decompression unit.
VIFcode Instruction code for VIF.
SPR Quick-access data memory built into EE Core (Scratchpad memory).
IPU EE Image processor unit.
word Unit of data length: 32 bits
qword Unit of data length: 128 bits
Slice Physical unit of DMA transfer: 8 qwords or less
Packet Data to be handled as a logical unit for transfer processing.
Transfer list A group of packets transferred in serial DMA transfer processing.
Tag Additional data indicating data size and other attributes of packets.
DMAtag Tag positioned first in DMA packet to indicate address/size of data and address
of the following packet.
GS primitive Data to indicate image elements such as point and triangle.
Context A set of drawing information (e.g. texture, distant fog color, and dither matrix)
applied to two or more primitives uniformly. Also referred to as the drawing
environment.
GIFtag Additional data to indicate attributes of GS primitives.
Display list A group of GS primitives to indicate batches of images.

(This page is left blank intentionally)

Contents

1. Overview................................................................................................................................................................................. 13
1.1. GS Features .................................................................................................................................................................... 14
1.1.1. Drawing Process..................................................................................................................................................... 14
1.1.2. Local Memory......................................................................................................................................................... 15
1.1.3. Video I/O ............................................................................................................................................................... 15
1.2. Block Configuration ...................................................................................................................................................... 16
1.2.1. Whole Block Diagram ........................................................................................................................................... 16
1.3. Maximum Performance................................................................................................................................................. 18
2. Local Memory ........................................................................................................................................................................ 19
2.1. Data Stored in Local Memory...................................................................................................................................... 20
2.2. Addressing ...................................................................................................................................................................... 21
2.2.1. Address Value ......................................................................................................................................................... 21
2.2.2. Starting Address of Buffer .................................................................................................................................... 21
2.2.3. Address in Buffer ................................................................................................................................................... 21
2.3. Data Formats.................................................................................................................................................................. 22
2.3.1. Color Data Format (Pixel/Texel Format)........................................................................................................... 22
2.3.2. Z Value Format ...................................................................................................................................................... 22
2.4. Frame Buffer .................................................................................................................................................................. 23
2.4.1. Size of Frame Buffer.............................................................................................................................................. 23
2.4.2. Starting Address of Frame Buffer ........................................................................................................................ 23
2.4.3. Coordinate Systems................................................................................................................................................ 23
2.4.4. Pixel Correspondence ............................................................................................................................................ 24
2.4.5. Pixel Storage Format.............................................................................................................................................. 25
2.5. Z Buffer........................................................................................................................................................................... 26
2.5.1. Starting Address of Z Buffer ................................................................................................................................ 26
2.5.2. Coordinate System of Z Buffer ............................................................................................................................ 26
2.5.3. Z Value Storage Format ........................................................................................................................................ 26
2.5.4. Combinations of Frame Buffer and Z Buffer Formats .................................................................................... 26
2.6. Texture Buffer ................................................................................................................................................................ 27
2.6.1. Starting Address of Texture Buffer...................................................................................................................... 27
2.6.2. Coordinate System of Texture Buffer ................................................................................................................. 27
2.6.3. Texel Storage Format............................................................................................................................................. 29
2.7. CLUT Buffer .................................................................................................................................................................. 30
2.7.1. CLUT Configuration ............................................................................................................................................. 30
2.7.2. Starting Address...................................................................................................................................................... 30
2.7.3. CLUT Storage Mode.............................................................................................................................................. 30
2.7.4. CLUT Storage Format........................................................................................................................................... 31
3. Drawing Function ................................................................................................................................................................. 33
3.1. Outline of Drawing Function....................................................................................................................................... 34
3.2. GS Primitive ................................................................................................................................................................... 36

3.2.1. Types of GS Primitives ..........................................................................................................................................36

3.2.2. Drawing Attributes .................................................................................................................................................37
3.2.3. Outline of Drawing Procedure .............................................................................................................................38
3.2.4. Example of the Drawing Procedure.....................................................................................................................39
3.2.5. Drawing Control .....................................................................................................................................................42
3.2.6. Change of Drawing Attributes..............................................................................................................................43
3.2.7. Register Setting Order and Drawing Order ........................................................................................................43
3.2.8. Enabled Vertex Information .................................................................................................................................43
3.2.9. Primitive Drawing Rule..........................................................................................................................................44
3.3. Drawing Environment...................................................................................................................................................47
3.3.1. Drawing Environment Setting Registers .............................................................................................................47
3.3.2. Two Context Drawing Environment...................................................................................................................47
3.4. Texture Mapping ............................................................................................................................................................50
3.4.1. Outline of Texture Mapping Process...................................................................................................................50
3.4.2. Texture Mapping Process Flow ............................................................................................................................50
3.4.3. Texture Information Setting..................................................................................................................................51
3.4.4. Specification of Texture Coordinate Values .......................................................................................................52
3.4.5. Texture Wrap Modes..............................................................................................................................................52
3.4.6. Format Conversion.................................................................................................................................................54
3.4.7. CLUT Buffer Control ............................................................................................................................................55
3.4.8. Texel Sampling ........................................................................................................................................................58
3.4.9. Texture Function ....................................................................................................................................................59
3.4.10. Perspective Correction.........................................................................................................................................60
3.4.11. MIPMAP................................................................................................................................................................61
3.4.12. LOD Setting ..........................................................................................................................................................62
3.4.13. Texture Mapping Procedure ...............................................................................................................................63
3.5. Fog Effect........................................................................................................................................................................64
3.6. Antialiasing ......................................................................................................................................................................65
3.6.1. Principle of Antialiasing .........................................................................................................................................65
3.6.2. Antialiasing Control................................................................................................................................................65
3.7. Pixel Test .........................................................................................................................................................................66
3.7.1. Scissoring Test.........................................................................................................................................................66
3.7.2. Alpha Test................................................................................................................................................................67
3.7.3. Destination Alpha Test ..........................................................................................................................................68
3.7.4. Depth Test...............................................................................................................................................................68
3.8. Alpha Blending ...............................................................................................................................................................69
3.8.1. Blending Setting ......................................................................................................................................................69
3.8.2. PABE Flag ...............................................................................................................................................................69
3.9. Writing to the Frame Buffer .........................................................................................................................................70
3.9.1. Dithering ..................................................................................................................................................................70
3.9.2. Color Clamp ............................................................................................................................................................71
3.9.3. Alpha Value Correction .........................................................................................................................................71
3.9.4. Format Conversion.................................................................................................................................................71
3.9.5. Masking ....................................................................................................................................................................72
4. Image Data Transmission .....................................................................................................................................................73
4.1. Transmission Parameters...............................................................................................................................................74

4.1.1. Destination Buffer / Source Buffer (BITBLTBUF Register) .......................................................................... 74

4.1.2. Offset in Buffer (TRXPOS Register) .................................................................................................................. 74
4.1.3. Width and Height of Transmission Area (TRXREG Register) ....................................................................... 75
4.1.4. Transmission Direction and Transmission Start (TRXDIR Register) ............................................................ 75
4.1.5. Limitations on Transmission Start Coordinates and Width ............................................................................. 76
4.2. Transmission Process .................................................................................................................................................... 77
4.2.1. Transmission from Host to Local Buffer ........................................................................................................... 77
4.2.2. Transmission from Local Buffer to Host ........................................................................................................... 77
4.2.3. Transmission from Local Buffer to Local Buffer .............................................................................................. 77
4.2.4. Restrictions on Drawing after Transmission ...................................................................................................... 78
4.3. Transmission Data Format........................................................................................................................................... 79
5. CRTC ...................................................................................................................................................................................... 81
5.1. CRTC Function.............................................................................................................................................................. 82
5.1.1. PCRTC Features..................................................................................................................................................... 82
5.1.2. Corresponding Video Signal ................................................................................................................................. 84
5.2. Video Output.................................................................................................................................................................. 85
5.2.1. Video Clock............................................................................................................................................................. 85
5.2.2. Rectangular Area Read Output Circuit................................................................................................................ 85
5.2.3. Merge Circuit .......................................................................................................................................................... 87
5.3. Feedback Write............................................................................................................................................................... 90
5.3.1. Rectangular Area Write Input Circuit.................................................................................................................. 90
6. Signal ....................................................................................................................................................................................... 93
6.1. Outline of Signal Processing......................................................................................................................................... 94
6.2. Signal Register ................................................................................................................................................................ 95
6.2.1. SIGNAL Register................................................................................................................................................... 95
6.2.2. FINISH Register .................................................................................................................................................... 95
6.2.3. LABEL Register ..................................................................................................................................................... 95
7. Registers.................................................................................................................................................................................. 97
7.1. General Purpose Registers............................................................................................................................................ 98
ALPHA_1 / ALPHA_2 : Alpha Blending Setting.....................................................................................................100
BITBLTBUF : Setting for Transmission between Buffers .......................................................................................101
CLAMP_1 / CLAMP_2 : Texture Wrap Mode .........................................................................................................102
COLCLAMP : Color Clamp Control...........................................................................................................................103
DIMX : Dither Matrix Setting ......................................................................................................................................104
DTHE : Dither Control.................................................................................................................................................105
FBA_1 / FBA_2 : Alpha Correction Value ................................................................................................................106
FINISH : FINISH Event Occurrence Request..........................................................................................................107
FOG : Vertex Fog Value Setting ..................................................................................................................................108
FOGCOL : Distant Fog Color Setting ........................................................................................................................109
FRAME_1 / FRAME_2 : Frame Buffer Setting .......................................................................................................110
HWREG : Data Port for Transmission between Buffers .........................................................................................111
LABEL : LABEL Event Occurrence Request ...........................................................................................................112
MIPTBP1_1 / MIPTBP1_2 : MIPMAP Information Setting (Level 1 to 3).........................................................113
MIPTBP2_1 / MIPTBP2_2 : MIPMAP Information Setting (Level 4 to 6).........................................................114
PABE : Alpha Blending Control in Units of Pixels ...................................................................................................115
PRIM : Drawing Primitive Setting................................................................................................................................116

PRMODE : Setting for Attributes of Drawing Primitives ....................................................................................... 117

PRMODECONT : Specification of Primitive Attribute Setting Method .............................................................. 118
RGBAQ : Vertex Color Setting.................................................................................................................................... 119
SCANMSK : Raster Address Mask Setting ................................................................................................................ 120
SCISSOR_1 / SCISSOR_2 : Setting for Scissoring Area......................................................................................... 121
SIGNAL : SIGNAL Event Occurrence Request ...................................................................................................... 122
ST : Specification of Vertex Texture Coordinates ..................................................................................................... 123
TEST_1 / TEST_2 : Pixel Test Control..................................................................................................................... 124
TEX0_1 / TEX0_2 : Texture Information Setting................................................................................................... 125
TEX1_1 / TEX1_2 : Texture Information Setting................................................................................................... 127
TEX2_1 / TEX2_2 : Texture Information Setting................................................................................................... 128
TEXA : Texture Alpha Value Setting.......................................................................................................................... 129
TEXCLUT : CLUT Position Specification ................................................................................................................ 130
TEXFLUSH : Texture Page Buffer Disabling ........................................................................................................... 131
TRXDIR : Activation of Transmission between Buffers ......................................................................................... 132
TRXPOS : Specification of Transmission Areas in Buffers..................................................................................... 133
TRXREG : Specification of Transmission Areas in Buffers.................................................................................... 134
UV : Specification of Vertex Texture Coordinates.................................................................................................... 135
XYOFFSET_1 / XYOFFSET_2 : Offset Value Setting ......................................................................................... 136
XYZ2 : Setting for Vertex Coordinate Values ........................................................................................................... 137
XYZ3 : Setting for Vertex Coordinate Values (without Drawing Kick) ................................................................ 138
XYZF2 : Setting for Vertex Coordinate Values......................................................................................................... 139
XYZF3 : Setting for Vertex Coordinate Values (without Drawing Kick).............................................................. 140
ZBUF_1 / ZBUF_2 : Z Buffer Setting ...................................................................................................................... 141
7.2. Privileged Registers ..................................................................................................................................................... 142
BGCOLOR (w) : Background Color Setting ............................................................................................................. 143
BUSDIR (w) : Host Interface Bus Switching ............................................................................................................. 144
CSR (r/w) : System Status............................................................................................................................................. 145
DISPFB1 (w) : Setting for Rectangular Area Read Output Circuit 1...................................................................... 147
DISPFB2 (w) : Setting for Rectangular Area Read Output Circuit 2...................................................................... 148
DISPLAY1 (w) : Setting for Rectangular Area Read Output Circuit 1 .................................................................. 149
DISPLAY2 (w) : Setting for Rectangular Area Read Output Circuit 2 .................................................................. 150
EXTBUF (w) : Setting for Feedback Write Buffer.................................................................................................... 151
EXTDATA (w) : Feedback Write Setting .................................................................................................................. 152
EXTWRITE (w) : Feedback Write Function Control .............................................................................................. 153
IMR (w) : Interrupt Mask Control ............................................................................................................................... 154
PMODE (w) : PCRTC Mode Setting.......................................................................................................................... 155
SIGLBLID (r/w) : Signal ID Value Read................................................................................................................... 156
SMODE2 (w) : Setting for Modes Related to Video Synchronization ................................................................... 157
7.3. Register List in Address Order .................................................................................................................................. 158
8. Details of GS Local Memory............................................................................................................................................. 161
8.1. Memory Access Units ................................................................................................................................................. 162
8.2. Page Arrangement in Buffer ...................................................................................................................................... 163
8.3. Data Structure in a Page ............................................................................................................................................. 164
8.3.1. PSMCT32/PSMCT24/PSMZ32/PSMZ24 ..................................................................................................... 164
8.3.2. PSMCT16/PSMCT16S....................................................................................................................................... 165

8.3.3. PSMT8 ...................................................................................................................................................................166

8.3.4. PSMT4 ...................................................................................................................................................................168
8.4. Occupied Memory Area..............................................................................................................................................170
8.5. Pointing within a Page.................................................................................................................................................171
8.6. Pixel Storage Format Conversion..............................................................................................................................172
8.6.1. PSMT4 ...................................................................................................................................................................172
8.6.2. PSMT8 ...................................................................................................................................................................174
8.6.3. PSMT16 .................................................................................................................................................................175

(This page is left blank intentionally)

1. Overview

1.1. GS Features
The GS is a high-performance graphics processor. The following sections summarize the features of the
drawing process, local memory and video input/output.
1.1.1. Drawing Process
The GS draws primitives, such as Polygon or Line, to the frame buffer in local memory, based on drawing
requests from the host processor. The main features are:

Full color internal operation

32-bit calculation precision (8 bits each for RGB + 8 bits for Alpha value)

A wide variety of drawing primitives

Point, Line, LineStrip, Triangle, TriangleStrip, TriangleFan, and Sprite

Shading
Flat shading and Gouraud shading

Texture mapping
Perspective correction
Bilinear and trilinear texture mapping
MIPMAP
Tiled textures specified by wrap mode

Texture mode
16/24/32 bits (without CLUT), 4/8 bits (with CLUT)

Z buffer
High-speed Z buffer processing without speed decrease
Z value format of 3 types: 16 bits, 24 bits and 32 bits

Alpha blending
High-speed blending without speed decrease

Edge antialiasing
Removing the edge aliasing of Lines and Triangles

Fogging
Fog effects in pixel units enabled for all primitives

High-efficiency scissoring
High-efficiency scissoring to negate the need for scissoring in geometry calculation

Registers with 2 contexts

Makes saving and restoring the drawing environment by a context switch unnecessary.

1.1.2. Local Memory

The chip contains 4 MB of DRAM as local memory to save pixel information (frame buffer and Z buffer)
generated by the drawing function and texture information.

Unified configuration
Contains the frame buffer, Z buffer, textures and CLUTs in the same memory space.

2 port configuration
In this configuration, the frame + Z buffer port is independent of the texture port. They each have a page
buffer of 8KB, and are concurrently accessible.

Wide bandwidth
Bandwidth of frame buffer: 38.4 GBytes/sec (1024 bits x 150 MHz x 2)
Texture bandwidth: 9.6 GBytes/sec (512 bits x 150 MHz)
Transmission from DRAM to page buffer: 8192 bits @ 1cycle at maximum

High-speed transmission between buffers

Host -> Local, Local -> Local and Local -> Host transmissions
Maximum transmission speed: 1.2 GB/sec

1.1.3. Video I/O

This function reads image information from memory in the scanning order of the television signal, performs
D/A conversion and outputs the video signal.

Various output formats

NTSC/PAL/VESA
Interlace/non-interlace can be switched.

1.2. Block Configuration

1.2.1. Whole Block Diagram
The whole GS configuration is shown in the following figure:

Host Bus
Local/Host
Host Interface
Host/Local
Drawing Function Vertex Info/Drawing Environment
Block

Setup/Rasterizing

XYZ/RGBA/STQ etc.

Pixel Pipeline x 16 (Max)

#0 #1 #2 #3 #4 #14 #15

Texel x 16
RGBA/Z

Memory Interface

Display Function RGBA + Z Texture

Block Memory Block (1024 bits x 2) (512 bits)

PCRTC
Frame Page Buffer Texture Page Buffer

Local Memory
(4 MB)
Video Out

Figure 1-1 GS Whole Block Diagram

The functional blocks in the figure perform as follows:

Host Interface
This interface transfers data with the host (CPU). Drawing data and buffer transfer data from the host pass
through this interface.

Setup/Rasterizing (preprocessing)
This block develops the graphics to draw to the pixels based on vertex information received from the host,
and calculates information such as RGBA value, Z value, texture value, and fog value for each pixel.

Pixel Pipeline
This block performs processes such as texture mapping, Fogging, and Alpha-blending, and determines the
final drawing color based on pixel information calculated in the Rasterizing block. It can process a
maximum of 16 pixels concurrently.

Memory Interface
This block reads data from and writes data to the GS local memory. It writes to memory the drawing pixel
values (RGBA, Z) at the end of a pixel operation, reads from memory the pixel values of the frame buffer
(used for the pixel test or Alpha-blending), and reads from memory the RGBA values for the display image.

Local Memory
The GS has a 32-Mbit built-in local memory, containing the frame buffer, Z buffer, texture and CLUT. This
memory has a 1024-bit read port and a 1024-bit write port for drawing and accessing the frame buffer and Z
buffer and a 512-bit port for texture reading.

PCRTC
PCRTC displays the contents of the frame memory in the specified output format.

1.3. Maximum Performance

Maximum pixel writing rate
The GS processes 16 pixels per cycle when not performing texture mapping and 8 pixels per cycle when
performing texture mapping. The maximum achievable pixel writing rates are:
Texture Mapping Max Pixel Writing Rate
Without Texture Mapping 2.4 Gpixel/sec
With Texture Mapping 1.2 Gpixel/sec
(Each case has 32-bit pixels, Z-buffering ON and Alpha-Blending ON.)

Maximum setup performance

Setup performance means how many polygons the preprocessor can process per second; the upper limit
occurs when pixel writing does not create a processing bottleneck. It changes depending on the kind of
primitives to be drawn and the drawing conditions, as shown below.

Processing Performance in Triangle Drawing

Texture Shading Fogging Anti-aliasing Necessary Performance
cycle count (triangle/sec)
OFF Flat ON/OFF OFF 2 75M
ON Flat/Gouraud OFF OFF 4 37.5M
ON Flat/Gouraud ON OFF 5 30M
ON/OFF Gouraud OFF OFF 4 37.5M
ON/OFF Gouraud ON OFF 5 30M
OFF Flat ON/OFF ON 6 25M
ON Flat/Gouraud OFF ON 8 18M
ON Flat/Gouraud ON ON 9 16M
ON/OFF Gouraud OFF ON 8 18M
ON/OFF Gouraud ON ON 9 16M

Processing Performance in Line Drawing

Texture Shading Fogging Anti-aliasing Necessary Performance
cycle count (line/sec)
ON Flat/Gouraud OFF OFF 4 37.5M
ON Flat/Gouraud ON OFF 5 30M
OFF Flat ON/OFF ON 6 25M
ON Flat/Gouraud OFF ON 7 21M

Processing Performance in Sprite

Texture Shading Fogging Anti-aliasing Necessary Performance
cycle count (sprite/sec)
ON/OFF Flat/Gouraud ON/OFF OFF 3 50M

Processing Performance in Point

Texture Shading Fogging Anti-aliasing Necessary Performance
cycle count (point/sec)
ON/OFF Flat/Gouraud ON/OFF OFF 1 150M

When a difference in Z coordinates between the vertices composing one drawing primitive exceeds 216, a penalty
occurs and the cycle count required increases.

2. Local Memory

2.1. Data Stored in Local Memory

The GS uses four kinds of data in local memory for drawing. The user must secure data areas (buffers) in local
memory before starting drawing.

Frame buffer
Area for drawing. Stores pixels (RGBA) of the drawing result.

Z buffer
Area for drawing. Stores Z value of the drawing result.

Texture buffer
Stores texture image data.

CLUT buffer
Stores the Color Look up Table (CLUT) used when the texture is an index color.

These buffers can be freely arranged in local memory.

2.2. Addressing
2.2.1. Address Value
Local memory addressing uses linear addresses in 32-bit word units.
32 bits

(0x00000)

Frame
FBP: Frame Buffer Starting Address

ZBP: Z Buffer Starting Address

Address

TBP: Texture Buffer Starting Address

Texture
CBP: CLUT Buffer Starting Address
Texture

CLUT
(0xfffff)

Figure 2-1 Local Memory of GS

2.2.2. Starting Address of Buffer

The different buffer types require different starting address alignments, as follows:

Buffer Type Alignment Size

Frame buffer 2K words
Z buffer 2K words
Texture buffer 64 words
CLUT buffer 64 words
The starting address of a buffer is specified, not as a 32-bit address, but in units of the address divided by the
above alignment size.

2.2.3. Address in Buffer

Two-dimensional coordinates are used to specify the data position in each buffer. Given the starting address
and the buffer width (maximum X), the XY coordinate values in the buffer are converted to one-dimensional
real memory addresses. When converting from two-dimensional coordinates to a memory address, the GS
performs special processing to improve drawing efficiency. (For details of the conversion, see "8. Details of GS
Local Memory".)

2.3. Data Formats

The formats for data stored in the buffer (pixels, Z values, etc.) are shown below.

2.3.1. Color Data Format (Pixel/Texel Format)

The GS supports the following five color data formats.

RGBA32
31 24 23 16 15 8 7 0
A (8 bits) B (8 bits) G (8 bits) R (8 bits)

RGB24
23 16 15 8 7 0
B (8 bits) G (8 bits) R (8 bits)

RGBA16
15 14 10 9 5 4 0
A B (5 bits) G (5 bits) R (5 bits)

IDTEX8
7 0
index (8 bits)

IDTEX4
3 0
index
(4 bits)
Color formats IDTEX8 and IDTEX4 can be used only in the texture buffer. They become index values for
the Color Look up Table (CLUT). For details of the interpretation of the index value, see "2.7. CLUT
Buffer".
2.3.2. Z Value Format
Data in the following three formats can be stored in the Z buffer.

Z32
31 0
Z (32 bits)

Z24
23 0
Z (24 bits)

Z16
15 0
Z (16 bits)

2.4. Frame Buffer

The frame buffer is the area where image data of drawing results are stored. Settings related to the frame buffer
are made in the FRAME register (FRAME_1 or FRAME_2).
2.4.1. Size of Frame Buffer
The frame buffer width must be a multiple of 64 pixels. This value is specified in the FBW field of the FRAME
register.
The frame buffer height is arbitrary, and not specially set by the register. The height can be limited by limiting
the rectangular area where drawing is performed with the Scissoring function (described later in this document).

2.4.2. Starting Address of Frame Buffer

The starting address of the frame buffer in local memory is specified in the FBP field of the FRAME register.
Since the alignment of the frame buffer is 2K words, the FBP field is set as the address divided by 2048.

2.4.3. Coordinate Systems

There are two kinds of coordinate systems that point to the pixels in the frame buffer.

Primitive Coordinate System

This is the coordinate system of the drawing space. It is used to set vertex coordinate values during the
drawing process. Both X and Y coordinates are fixed 16-bit decimal values (12-bit integer and 4-bit decimal)
in the range of 0 to 4095.9375. The rectangular area in the frame buffer where drawing is actually performed
is defined in this space.

Window Coordinate System

This coordinate system uses the upper left point of the rectangular area of the frame buffer as the origin.
The calculation of memory addresses is based on the coordinate values.
Mapping to the Primitive coordinate system is done by using the offset value. The offset is stored in the
XYOFFSET register.
Assuming that the Primitive coordinate values are (Px,Py) and the Offset values are (Offx,Offy), the Window
coordinate values (Wx,Wy) are obtained from the following formulas:
Wx = Px – Offx
Wy = Py – Offy
The following figure shows the relation between the Primitive Coordinate System and the Window
Coordinate System:

(0,0) 4095.9375
x

Origin Point Px,Py

in Primitive Offset
Coordinate System

(0,0) x
Wx,Wy
Origin Point
in Window
Coordinate System

4095.9375
y
y

Figure 2-2 Coordinate Systems in Frame Buffer

2.4.4. Pixel Correspondence

The pixels in the frame buffer are centered on positions where the fractional parts of the Window coordinate
values are 0 (the intersection of the grid in the figure).

Wx
1.0 2.0 3.0
Wy Frame
Buffer

1.0

Center of Pixel

2.0

3.0

Figure 2-3 Frame Buffer Coordinates and Pixel Position

2.4.5. Pixel Storage Format

The pixel storage format defines how the pixels are arranged in each 32-bit word of local memory. The eight
formats are shown below:

PSMCT32
31 0
RGBA32

PSMCT24
31 24 23 0
Not used RGB24

PSMCT16, PSMCT16S
31 16 15 0
RGBA16 (Upper) RGBA16 (Lower)

PSMZ32
31 0
Z32

PSMZ24
31 24 23 0
Not used Z24

PSMZ16, PSMZ16S
31 16 15 0
Z16 Z16

The pixel storage format used is specified in the PSM field (Pixel Storage Mode) of the FRAME register.

2.5. Z Buffer
The Z buffer is an area where the Z values of the pixels are stored as a result of drawing. The size (width and
height) of the Z buffer is the same as that of the frame buffer.
2.5.1. Starting Address of Z Buffer
The starting address of the Z buffer in local memory is specified in the ZBP field of the ZBUF register. The
alignment of the Z buffer is 2048 words, so the ZBP field is set to the address divided by 2048.
2.5.2. Coordinate System of Z Buffer
The coordinate system of the Z buffer is the same as that of the frame buffer.

2.5.3. Z Value Storage Format

The data storage format of the Z buffer is as follows:

PSMZ32
31 0
Z32

PSMZ24
31 24 23 0
Not used Z24

PSMZ16, PSMZ16S
31 16 15 0
Z16 Z16

2.5.4. Combinations of Frame Buffer and Z Buffer Formats

To use the pixel storage format for the frame buffer and the Z value storage format for the Z buffer in
combination, both the formats must belong to the same group (Group1 or 2 below).

Group 1 Group 2
PSMCT32
PSMCT24
PSMCT16S PSMCT16
PSMZ32
PSMZ24
PSMZ16S PSMZ16

The following are some examples of applicable and inapplicable combinations.

Pixel Storage Format for Z Value Storage Format for Z Combination

Frame Buffer Buffer
PSMCT32 PSMZ24 Yes
PSMCT32 PSMZ16 No
PSMCT32 PSMZ16S Yes
PSMCT16 PSMZ24 No
PSMCT16S PSMZ24 Yes
PSMCT16 PSMZ16 Yes

2.6. Texture Buffer

The texture buffer stores texture image data for drawing textured polygons. The TEX0 register stores settings
related to the texture buffer.
The height and width of textures in the buffer are specifiable separately, and independently of the width of the
buffer. They must be a power of 2, with a maximum size of 1024 texels.
The width of the buffer is a multiple of 64 texels, but it must be a multiple of 128 texels when PSMT8 or
PSMT4 is specified as the buffer storage format.

2.6.1. Starting Address of Texture Buffer

The starting address of the texture buffer in local memory is called the texture base pointer. It is specified in the
TBP0 field in the TEX0 register, in units of 64 words, that is, the starting address divided by 64.
When MIPMAP is performed, it is necessary to set the starting addresses and buffer widths for the textures of
two or more MIP levels. Texture settings of Level 0 (the maximum resolution) are stored in the TEX0 register
(as well as the case where MIPMAP is not used). Texture settings of Level 1 to 6 are specified in the MIPTBP1
and MIPBP2 registers.

2.6.2. Coordinate System of Texture Buffer

Coordinates in the texture buffer are specified as either texture coordinates (STQ) or texel coordinates (UV).
When performing perspective correction, texture coordinates must be used.

Texture coordinates
The GS registers use texture coordinates specified in a two-dimensional homogeneous coordinate system (S,
T, Q) to achieve perspective correction.
The coordinate values stored in the texture buffer are indicated by the normalized texture coordinates (s, t)
showing the upper left point as (0.0, 0.0) and the lower right point as (1.0, 1.0) in a single-precision floating-
point representation.

0.0 1.0 s

A
0.0

1.0

1.0 1.0
t
Figure 2-4 Normalized Texture Coordinates
The texture coordinates (S, T, Q) use the results of the conversion of (s, t). For the calculation, see "3.4.10.
Perspective Correction".

Texel coordinates
Texel coordinates are two-dimensional coordinate values indicated by (U, V). The UV value is a fixed 16-bit
decimal value, where 1 texel is 1.0. The upper 12 bits are the unsigned integer part, and the lower 4 bits are
the fractional part.

0.0 16.0 U

A
0.0

16.0

16.0 16.0

V
Figure 2-5 Texture Coordinates (in case Texture Size is 16x16)
Texels are centered on the position where the fractional parts of the texel coordinate values are 0.5 (the
center of the square in the grid in the figure).

U 1.0 2.0 3.0

V
1.0

2.0

Center of Texel
3.0

Figure 2-6 Texel Coordinate System and Center of Texel

2.6.3. Texel Storage Format

There are 8 forms of Texel Storage Format for the texture buffer as shown below:

PSMCT32
31 0
RGBA32

PSMCT24
31 24 23 0
Not used RGB24

PSMCT16
31 16 15 0
RGBA16 RGBA16

PSMT8
31 24 23 16 15 8 7 0
IDTEX8 IDTEX8 IDTEX8 IDTEX8

PSMT8H
31 24 23 0
IDTEX8 Not used

PSMT4
31 24 23 16 15 8 7 0
IDTEX4 IDTEX4 IDTEX4 IDTEX4 IDTEX4 IDTEX4 IDTEX4 IDTEX4

PSMT4HH
31 28 27 0
IDTEX4 Not used

PSMT4HL
31 28 27 24 23 0
Not used IDTEX4 Not used

Since there is no influence on the "not used" bits, it is possible to combine textures of two different modes (e.g.
PSMCT24 and PSMT8H) in the same area.
The PSM field of the TEX0 register specifies which texel storage format is used.

2.7. CLUT Buffer

A CLUT (Color Look up Table) converts a texel value from an index to RGBA color data, when the color
format of the texel is IDTEX8 or IDTEX4. The CLUT buffer is the area where the CLUT is stored. Settings
related to the CLUT are stored in the TEX0 register or the TEX2 register.
The TEX2 register is a subset of the TEX0 register. Among the values set by the TEX0 register, it is possible to
change only CLUT-related information by changing the values in the TEX2 register.
2.7.1. CLUT Configuration
A CLUT is an array of two or more elements (CLUT entries) that describe color information, arranged in a
prescribed order. The number of CLUT entries depends on the texel color format; it is 16 for IDTEX4 and 256
for IDTEX8. The CLUT data format in local memory is the same as for pixels. For instance, CLUT entries
arranged as 16x16 in width and height are the same as 16x16 image data.

2.7.2. Starting Address

The starting address of the CLUT buffer in local memory is called the CLUT base pointer, and is specified in the
CBP field of the TEX0 or TEX2 register. The CLUT base pointer is in units of 64 words; that is, the value of
the address divided by 64 is stored in the CBP field.
2.7.3. CLUT Storage Mode
The CLUT buffer is a two-dimensional space like the other buffers. The CLUT storage mode determines how
CLUT entries are arranged in the two-dimensional space of the CLUT buffer. There are two CLUT storage
modes, CSM1 and CSM2. They are specified in the CSM field of the TEX0 register.

CSM1
Data can be more efficiently read from local memory in CSM1 than in CSM2.
The CLUT entries are arranged in order in 16x16 rectangles in IDTEX8 (8 bits) and 8x2 in IDTEX4 (4 bits)
as shown below. The numbers in the figures show the corresponding index values (in hexadecimal).

Arrangement of CLUT in IDTEX4

X
Y 0 1 2 3 4 5 6 7

8 9 a b c d e f

Arrangement of CLUT in IDTEX8

X
Y 00 01 02 03 04 05 06 07 10 11 12 13 14 15 16 17

08 09 0a 0b 0c 0d 0e 0f 18 19 1a 1b 1c 1d 1e 1f

20 21 22 23 24 25 26 27 30 31 32 33 34 35 36 37

28 29 2a 2b 2c 2d 2e 2f 38 39 3a 3b 3c 3d 3e 3f

e0 e1 e2 e3 e4 e5 e6 e7 f0 f1 f2 f3 f4 f5 f6 f7

e8 e9 ea eb ec ed ee ef f8 f9 fa fb fc fd fe ff

In CSM1, arrangement should be made from the start of the buffer (the position where X and Y are each 0).

CSM2
In CSM2, two or more CLUT sets can be arranged at free positions in one big CLUT buffer. However, the
CLUT entries must be in PSMCT16, and the efficiency of transmission to the temporary buffer decreases.
The position in the buffer is specified in the TEXCLUT register.
The CLUT entries are arranged in order as shown below. The CLUT entries are arranged in a width of 256
or 16 and a height of 1. The numbers in the figures show the corresponding index values (in hexadecimal).

Arrangement of CLUT in IDTEX4

X 0 1 2 3 4 5 6 7 8 9 a b c d e f
Y

Arrangement of CLUT in IDTEX8

X 00 01 02 03 04 05 06 07 f9 fa fb fc fd fe ff
Y

2.7.4. CLUT Storage Format

The storage formats of individual CLUT entries are the same as those of the pixels in the frame buffer. When
the CLUT storage mode is CSM1, PSMCT32, PSMCT16 or PSMCT16S can be specified. When it is CSM2,
only PSMCT16 can be specified. They are specified in the CPSM field of the TEX0 register.

(This page is left blank intentionally)

3. Drawing Function

3.1. Outline of Drawing Function

The GS receives the information necessary for drawing (vertex and drawing environment information) from the
host processor via the Host interface. It draws drawing primitives called GS primitives as processing units.
The drawing flow in the GS is shown below.

Host Interface
Drawing Environment Vertex Information

Drawing Setup (preprocessing)

Environment
Register DDA Set Value (Initial Value / Gradient)

Rasterizing (DDA)
Scissoring

XY Z F RGBA STQ/UV Coverage XYZRGBA....

To Each Process

Texture Texture Mapping

CLUT Texture Function
RGB A

Fogging Antialiasing
RGB A

Pixel Test
Alpha Test
A, Z Destination Alpha Test
Depth Test

RGB
Alpha Blending

Formatting
XY Z RGB A

Memory Interface

Figure 3-2 Block Diagram for Drawing Processing

The processes shown in the figure are:

Setup (Preprocessing)
The gradient (e.g. shading coefficient) and the initial value of DDA necessary for drawing primitives are
calculated based on the vertex information received from the host.
If the primitive is a Triangle, the gradient of the RGBA value, Z value, texture value, and Fog value on the
three sides and on the scan line are calculated.

Rasterizing (DDA)
The pixels of a primitive are generated by DDA (Digital Differential Algorithm). 8 or 16 pixels are generated
concurrently. The RGBA value, Z value, texture value and Fog value for each pixel are calculated from the
gradient obtained in the preprocessing stage, and are given to each of the parallel pixel pipelines.

Texture Mapping
Textures are mapped to pixels. The pixel color is determined by applying the texture function to the Texture
CLUT RGBA value read from the memory block and the RGBA value calculated by DDA.

Antialiasing
The alpha value is replaced with the pixel coverage calculated by DDA (proportion of the pixel occupied by
the theoretical edge of the primitive.) The edges of the primitive are smoothed when Alpha-blending is
implemented with this alpha value.

Fogging
The RGB value and Fog color value output from the texture mapping block are blended according to the
Fog value of the pixel calculated by DDA.

Pixel Test
Whether to draw a pixel is based on its XYZ and RGBA values. Three kinds of tests—alpha test,
destination alpha test, and depth test—are performed one by one.

Alpha-blending
Blending the RGB value of a pixel and the RGB value in the frame memory is implemented according to the
alpha value of the pixel or the alpha value in the frame memory.

Formatting
The pixel value for drawing is converted into the data format of the frame buffer. Dithering and color
clamping are applied if necessary.

Memory Interface
Read/write is performed to local memory in the chip. The operations are: writing drawing pixel values
(RGBA, Z) to the memory after a pixel operation, reading pixel values into the frame buffer from the
memory (used for pixel test and alpha-blending), and reading RGBA values for display from memory.

Environment Registers
These registers store various parameters necessary for drawing. These values are referred to in each process
of drawing.

3.2. GS Primitive
3.2.1. Types of GS Primitives
The GS uses the following seven primitive types:

Point
A Point is an independent point that is drawn with 1 piece of vertex information.

1 3
2

Line
A Line is an independent line that is drawn with 2 pieces of vertex information.

1
3
2
4

LineStrip
A LineStrip is a series of continuous lines that share endpoints. The first line is drawn with 2 pieces of
vertex information. The succeeding lines are drawn with 1 piece of vertex information.

3
1
5
2
4

Triangle
A Triangle is an independent triangle that is drawn with 3 pieces of vertex information.

1 4

3 6
2 5

TriangleStrip
A TriangleStrip is a series of triangles that share sides. The first triangle is drawn with 3 pieces of vertex
information, and the succeeding ones are drawn whenever 1 piece of vertex information is added.

3 7
1 5

2
4 6

TriangleFan
A TriangleFan is a series of triangles sharing one vertex. The first triangle is drawn with 3 pieces of vertex
information, and the succeeding triangles are drawn whenever 1 piece of vertex information is added.

4
5
3
6

1 7

Sprite
A Sprite is an independent rectangle, drawn with 2 pieces of vertex information showing the endpoints of a
diagonal line.

4
1
3

3.2.2. Drawing Attributes

Besides their types, GS primitives have the following drawing attributes:

Drawing Contents Setting

Attribute
IIP Shading Method Flat/Gouraud
TME Texture Mapping ON/OFF
FGE Fogging ON/OFF
ABE Alpha-Blending ON/OFF
AA1 Antialiasing ON/OFF
FST Texture Coordinates STQ/UV
CTXT Context 1/2
FIX Interpolation FIX ON/OFF

Some drawing attributes are fixed for specific GS primitive types:

Point: Shading Method flat and Antialiasing Off
Sprite: Shading Method flat and Antialiasing Off
3.2.3. Outline of Drawing Procedure
The general drawing procedure is:
1) Set GS primitive type/drawing attribute (PRIM register)
2) Set vertex information (Vertex info setting registers)
3) Vertex Kick (XYZF2, XYZF3 registers, etc.)
4) Start drawing (XYZF2 register etc.)

1) Set GS Primitive Type and Drawing Attribute

The GS primitive type is specified by writing to the PRIM register. The condition of the vertex queue is also
initialized. The drawing attributes of the GS primitive can be set, and it is also possible to set GS primitive
drawing attributes in the PRMODE register. (See "3.2.6. Change of Drawing Attributes".)

2) Set Vertex Information

Vertex information includes the drawing coordinates, vertex color, texture coordinates, and Fog coefficient.
These values are set in the vertex information setting registers shown below.

Setting Item Register Vertex Drawing

Name Kick Kick
Coordinate values XYZ2 Yes Yes
Coordinate values and Fog coefficient XYZF2 Yes Yes
Coordinate values XYZ3 Yes No
Coordinate values and Fog coefficient XYZF3 Yes No
Color info and Q value of texture coordinates RGBAQ No No
ST of texture coordinates ST No No
UV of texel coordinates UV No No
Fog coefficient FOG No No

3) Vertex Kick
If the vertex information setting registers that have a Vertex Kick function are written to, the vertex
information set up to that point is placed in the vertex queue, and the queue goes one step forward. This
operation is called the Vertex Kick. The following figure shows the image of operation.

Vertex Information Setting Resisters

XYZ RGBA STQ/UV F

Copy
Vertex Queue
1st Vertex
XYZ, RGBA, STQ/UV, F

Move
2nd Vertex
XYZ, RGBA, STQ/UV, F Differs depending
on the primitive
Move type.
3rd Vertex
XYZ, RGBA, STQ/UV, F

Figure 3-3 Operation Image of Vertex Kick

4) Start Drawing (Drawing Kick)
When the necessary vertex information is arranged in the vertex queue, drawing begins.
3.2.4. Example of the Drawing Procedure
The following examples show the order of register setting for the drawing procedures for the GS primitive
types. The examples show the drawing sequences for non-textured GS primitives, and the RGBA value and the
coordinate values are given to each vertex.

Point Drawing Procedure

[Register Setting Order]
PRIM
RGBAQ Vertex1 Setting
XYZ2 Vertex Kick & Drawing Kick (Vertex1)
RGBAQ Vertex2 Setting
XYZ2 Vertex Kick & Drawing Kick (Vertex2)
:

Line Drawing Procedure

[Register Setting Order]
PRIM
RGBAQ Vertex1 Setting
XYZ2 Vertex Kick
RGBAQ Vertex2 Setting
XYZ2 Vertex Kick & Drawing Kick (Vertex1 & Vertex2)

LineStrip Drawing Procedure

[Register Setting Order]
PRIM
RGBAQ Vertex1 Setting
XYZ2 Vertex Kick
RGBAQ Vertex2 Setting
XYZ2 Vertex Kick & Drawing Kick(Vertex1 & Vertex2)
RGBAQ Vertex3 Setting
XYZ2 Vertex Kick & Drawing Kick (Vertex2 & Vertex3)

RGBAQ Vertex4 Setting

XYZ2 Vertex Kick & Drawing Kick (Vertex3 & Vertex4)

Triangle Drawing Procedure

RGBAQ Vertex6 Setting

Vertex Kick & Drawing Kick
XYZ2 (Vertex4 & Vertex5 & Vertex6)
:

TriangleStrip Drawing Procedure

[Register Setting Order]
PRIM
RGBAQ Vertex1 Setting
XYZ2 Vertex Kick
RGBAQ Vertex2 Setting
XYZ2 Vertex Kick
RGBAQ Vertex3 Setting
XYZ2 Vertex Kick & Drawing Kick
(Vertex1 & Vertex2 & Vertex3)
RGBAQ Vertex4 Setting
XYZ2 Vertex Kick & Drawing Kick
(Vertex2 & Vertex3 & Vertex4)
RGBAQ
Vertex5 Setting
XYZ2 Vertex Kick & Drawing Kick
(Vertex3 & Vertex4 & Vertex5)
RGBAQ Vertex6 Setting
XYZ2 Vertex Kick & Drawing Kick
(Vertex4 & Vertex5 & Vertex6)
:

TriangleFan Drawing Procedure

[Register Setting Order]
PRIM
RGBAQ Vertex1 Setting
XYZ2 Vertex Kick
RGBAQ Vertex2 Setting
XYZ2 Vertex Kick
RGBAQ Vertex3 Setting
XYZ2 Vertex Kick & Drawing Kick
(Vertex1 & Vertex2 & Vertex3)
RGBAQ Vertex4 Setting
XYZ2 Vertex Kick & Drawing Kick
(Vertex1 & Vertex3 & Vertex4)
RGBAQ
Vertex5 Setting
XYZ2 Vertex Kick & Drawing Kick
(Vertex1 & Vertex4 & Vertex5)
RGBAQ Vertex6 Setting
Vertex Kick & Drawing Kick
XYZ2 (Vertex1 & Vertex5 & Vertex6)
:

Sprite Drawing Procedure

[Register Setting Order]
PRIM
RGBAQ Vertex1 Setting
XYZ2 Vertex Kick
RGBAQ Vertex2 Setting
XYZ2 Vertex Kick & Drawing Kick
(Vertex1 & Vertex2)
RGBAQ Vertex3 Setting
XYZ2 Vertex Kick

RGBAQ Vertex4 Setting

Vertex Kick & Drawing Kick
XYZ2
(Vertex3 & Vertex4)
:

3.2.5. Drawing Control

When drawing a GS primitive consisting of two or more polygons, such as TriangleStrip, it is possible to control
a polygon that is being drawn without changing the vertex string order. As a result, operations like normal clips
become easy.
To control drawing in this way, the vertex information is written to the XYZ3 register instead of the XYZ2
register and to the XYZF3 register instead of the XYZF2 register. Because of this, the Drawing Kick is not
performed at the Vertex Kick, and only the vertex queue is advanced. For this reason, the polygon is not drawn.
The figure below is an example of such register settings. In this example, only the 2nd triangle is not drawn.

[Register Setting Order]

PRIM
RGBAQ Vertex1 Setting
XYZ2 Vertex Kick
RGBAQ Vertex2 Setting
XYZ2 Vertex Kick
RGBAQ Vertex3 Setting
XYZ2 Vertex Kick & Drawing Kick
(Vertex1 & Vertex2 & Vertex3)
RGBAQ
Vertex4 Setting
XYZ3 Vertex Kick (No Drawing Kick)
RGBAQ
Vertex5 Setting
XYZ2 Vertex Kick & Drawing Kick
(Vertex3 & Vertex4 & Vertex5)
RGBAQ Vertex6 Setting
Vertex Kick & Drawing Kick
XYZ2
(Vertex4 & Vertex5 & Vertex6)
:

3.2.6. Change of Drawing Attributes

The drawing attributes of a GS primitive are specified in the PRIM register. However, if the PRMODE register
is used, it is possible to change drawing attributes while drawing a primitive consisting of two or more polygons
such as TriangleStrip.
The following is an example of performing Flat shading to only one of the triangles in the process of drawing a
TriangleStrip, to which Gouraud shading is specified.

[Register Setting Order]

PRMODECONT Set to AC=1
PRIM Specify types of primitive (The attribute is ignored)
PRMODE Set to IIP=1(Gouraud)
RGBAQ Vertex1 Setting
XYZ2 Vertex Kick
RGBAQ Vertex2 Setting
XYZ2 Vertex Kick
RGBAQ Vertex3 Setting
XYZ2 Vertex Kick & Drawing Kick
(Vertex1 & Vertex2 & Vertex3)
PRMODE Reset to IIP=0(Flat)
RGBAQ
Vertex4 Setting
XYZ2 Vertex Kick & Drawing Kick
(Vertex2 & Vertex3 & Vertex4)
RGBAQ Vertex5 Setting
XYZ2 Vertex Kick & Drawing Kick
(Vertex3 & Vertex4 & Vertex5)
RGBAQ Vertex6 Setting
Vertex Kick & Drawing Kick
XYZ2 (Vertex4 & Vertex5 & Vertex6)
:

3.2.7. Register Setting Order and Drawing Order

Since the pixel pipeline, where drawing is performed, is designed to maintain the input register setting order,
dependence between register setting and drawing is in register setting order. This means register setting will not
override drawing. Moreover, the order of texture data transmission (Host/Local transmission) and drawing
operation will not be reversed. However, when using the data converted in Host/Local transmission or
Local/Local transmission as texture or CLUT for the first time, the texture buffer must be disabled with the
TEXFLUSH register.
3.2.8. Enabled Vertex Information
Note that there may be unusable vertex information in a register, depending on the primitive type and drawing
attribute.
For instance, a Sprite consists of two pieces of vertex information, but it only needs one piece of vertex
information for the Z coordinate value and Fog information, since depth is not provided. Therefore, the Z
coordinate and Fog coordinate of the first vertex are ignored, and the setting for the second one takes effect.
Also, when doing Flat shading, one piece of vertex color information is sufficient, so the vertex color
information set immediately before each drawing kick becomes effective. By this method, it is possible to
perform proper drawing by specifying Flat shading even in case of TriangleStrip.

3.2.9. Primitive Drawing Rule

The rules for how the DDA generates pixels are shown for each primitive.

Point
The pixel closest to the specified XY coordinates is drawn.

Closest Pixel

×
Specified XY coordinates

Figure 3-4 Pixel Drawing Rule for Point

Line, LineStrip
By defining the area for each pixel as shown with small gray squares in Figure 3-5, the pixels located in the
areas where a line goes through are drawn. However, a pixel located at the endpoint of a line is not drawn.
Any line can be drawn without being broken. Even when continued lines that share endpoints are drawn,
they cannot be drawn doubly unless crossed.

End point
is not drawn.
Start point
is drawn.

Pixel Drawn

Figure 3-5 Pixel Drawing Rule for Line

Drawn
only once

Pixel Drawn

Figure 3-6 Pixel Drawing Rule for LineStrip

Triangle, TriangleStrip, TriangleFan

Draws the pixels in the three sides specified by the three vertices. When a side passes the center of a pixel,
drawing is performed if it is the left side, and is not performed if it is the right side. When a side parallel to
the X axis passes the center of a pixel, drawing is performed if it is the top side, and is not performed if it is
the bottom side.
Even when drawing triangles that share sides with the same vertices, neither a gap nor a double line can be
created. (Figure 3-7)

Drawn on
the Left Side.
Not Drawn on
the Right Side.

Drawn on
the Top Side.

Not Drawn on
the Bottom Side.

Figure 3-7 Pixel Drawing Rule for Triangle

Drawn
only once.

Figure 3-8 Pixel Drawing Rule for Side-Sharing Triangles

Sprite
Draws pixels in a rectangular area with two vertices specified as diagonal points. When a side passes the
center of a pixel, drawing is performed if it is the top or left side and not performed if it is the bottom or
right side.
1st Diagonal Point

2nd Diagonal Point

1st Diagonal Point Drawn on the Top Side.

Drawn on
the Left Side. Not Drawn on the Right Side.

2nd Diagonal Point

Not Drawn on the Bottom Side.

Figure 3-9 Pixel Drawing Rule for Sprite

3.3. Drawing Environment

There are various parameters used for GS primitive drawing, such as texture information, in addition to the
drawing attributes set by the PRIM register. These are called the drawing environment. Once the drawing
environment is set, it remains in effect for multiple GS primitives until it is reset. Settings can be changed even
in the process of drawing a GS primitive consisting of two or more polygons, such as TriangleStrip.

3.3.1. Drawing Environment Setting Registers

The following are the drawing environment setting registers.

Register Name Contents

XYOFFSET_1/2 Offset value of vertex coordinates
PRMODECONT PRIM attributes enabled/disabled
TEX0_1/2 Attributes of texture buffer and texture mapping
TEX1_1/2 Attributes of texture mapping
TEX2_1/2 CLUT entry
CLAMP_1/2 Wrap mode of texture mapping
TEXCLUT CLUT setting
SCANMSK Drawing control with Y coordinate of pixel
MIPTBP1_1/2 Base pointer for MIPMAP on each level
MIPTBP2_1/2 Base pointer for MIPMAP on each level
TEXA Reference value when expanding Alpha value of TEX16 and
TEX24
FOGCOL Fogging distant color
SCISSOR_1/2 Scissoring area
ALPHA_1/2 Alpha-blending attributes
DIMX Dither matrix
DTHE Dithering enabled/disabled
COLCLAMP Color clamp/mask
TEST_1/2 Pixel operation
PABE Alpha-blending in pixel units enabled/disabled
FBA_1/2 Alpha correction value
FRAME_1/2 Frame buffer setting
ZBUF_1/2 Z buffer setting

3.3.2. Two Context Drawing Environment

Some of the drawing environment registers have two registers for the same function, such as XYOFFSET_1
and XYOFFSET_2. The selection of the first or second can be made by the primitive attribute CTXT. Even in
drawing command strings containing drawing commands from two kinds of contexts intermixed in GS primitive
units, drawing can be performed without repeating register save/load.

The registers with two contexts are listed below.

Register Name Contents
XYOFFSET_1/2 Offset value of Vertex coordinates
TEX0_1/2 Attributes of texture buffer and texture mapping
TEX1_1/2 Attributes of texture mapping
TEX2_1/2 CLUT entry
CLAMP_1/2 Wrap mode of texture mapping
MIPTBP1_1/2 Base pointer for MIPMAP on each level
MIPTBP2_1/2 Base pointer for MIPMAP on each level
SCISSOR_1/2 Scissoring area
ALPHA_1/2 Alpha-blending attributes
TEST_1/2 Pixel operation
FBA_1/2 Alpha correction value
FRAME_1/2 Frame buffer setting
ZBUF_1/2 Z buffer setting

The following is an example of switching the contexts by the primitive.

[Register Setting Order]
TEX1_1 MMAG=0 Point Sampling
TEX1_2 MMAG=1 Bilinear Sampling
PRIM Triangle, CTXT=0
ST
RGBAQ
XYZF2
ST
Drawing Triangle with
RGBAQ Context 1 (Point Sampling)
XTZF2
ST

RGBAQ
XYZF2
PRIM Triangle, CTXT=1

RGBAQ
XYZF2

ST
Drawing Triangle with
RGBAQ Context 2 (Bilinear Sampling)
XYZF2
ST
RGBAQ

XYZF2

Moreover, the contexts can be switched in the process of drawing a primitive.

[Register Setting Order]
TEX1_1 MMAG=0
Point Sampling
TEX1_2 MMAG=1 Bilinear Sampling

PRMODECONT AC=0
PRIM TriangleStrip
PRMODE CTXT=0
ST
RGBAQ
XYZF2 Drawing 1st Triangle with Context 1
ST (Point Sampling)
RGBAQ
XYZF2
ST

RGBAQ
XYZF2
ST

RGBAQ Drawing 2nd Triangle with Context 2

PRMODE CTXT=1 (Bilinear Sampling)

XYZF2
PRMODE CTXT=0
ST
Drawing 3rd Triangle with Context 1
RGBAQ (Point Sampling)

XYZF2

3.4. Texture Mapping

3.4.1. Outline of Texture Mapping Process
The GS can do texture mapping for all drawing primitives.
Assuming TME=1 in the primitive drawing information, texture mapping is based on the coordinate values of
the texture corresponding to each vertex.
Texture data must be stored in the texture buffer in local memory before drawing. Texture data is normally
transmitted from the host processor by the transmitting function between buffers. To obtain special effects,
however, images created via the drawing process can be used as textures.

3.4.2. Texture Mapping Process Flow

The texture mapping flow is shown in Figure 3-9.

DDA
RGBA U,V S,T Q

Normalization Texel
Coordinate Calculation LOD
s=S/Q Calculation
t=T/Q

Texture Texel Coordinate Calculation

Width / Height u = s x USIZE
v = t x VSIZE

STQ/UV Selection

Wrap Mode
Memory Address Calculation

Address
Texture Texture
Bit Expansion Page Buffer Buffer
Texel Value

Filtering
Address (Fractional Part)
Point Sampling
Bilinear
Fragment Color Texel Color

Texture Function Color Component

MODULATE/DECAL/HIGHLIGHT

Pixel Color

Figure 3-11 Texture Mapping Process Flow

The address in the texture buffer is calculated by using the texture coordinate (STQ or UV) values of the pixel
output from the DDA, and then the texel is read from the texture buffer. Bit expansion of RGB values and
Alpha value, and reference to the CLUT, are made according to the texture information, and bilinear filter etc.
are applied according to the filter mode. The texel color from this result and the fragment color output from the
DDA are calculated according to the texture function mode, and become the pixel color.

Texture Page Buffer

When texture mapping is performed, reference to local memory is made through the texture page buffer. To
appropriately reflect the changes made to local memory in drawing, it is necessary to invalidate the texture
page buffer when:
• Starting to use the data transmitted from the host (or within local memory) as texture data
• Starting to use the CLUT data stored in the CLUT buffer (or loading it into the temporary buffer)
• Starting to use the images created via the drawing process as textures
To invalidate the texture page buffer, write an arbitrary value to the TEXFLUSH register.
3.4.3. Texture Information Setting
Texture information is set in the TEX0_1 or TEX0_2 registers. The following values can be set; they should
correspond to the format of the texture data stored in the texture buffer.

Texture Size (TW, TH)

The texture size can be 1 x 1 texel (min.) to 1024 x 1024 texels (max.). Width (TW) and height (TH) are
specifiable independently as powers of 2. When performing bilinear or trilinear sampling, the texture size
must be 8 x 8 texels.
Assume USIZE and VSIZE represent the texture height and width in texel units. Then
USIZE = 2TW (texels)
VSIZE = 2TH (texels)
The TW and TH values are specified in the TW and TH fields in the register.

Texel Storage Mode (PSM)

Can be one of the following: RGBA32, RGB24, RGBA16, IDTEX8, IDTEX4

Starting Address (TBP0: Texture Base Pointer)

The TBP0 field of the register represents the address in units of 64 words.
Address = TBP0 x 64 (words)

Texture Buffer Width (TBW)

The TBW field of the register represents the buffer width in units of 64 texels. It is used for address
conversion in the texture buffer, and it is the same value as the buffer width set when the data is transferred
from host to local buffer. It differs from the texture size TW.
Width = TBW x 64 (words)

Color Component (TCC)

The TCC field determines whether to use the A value of the texture during the texture function execution.

Texture Function (TFX)

The TFX field represents the operation mode of the texture function.

When the texel storage format is IDTEX8 or IDTEX4, settings related to the CLUT buffer should be made in
addition to the above settings. See "3.4.7. CLUT Buffer Control".

3.4.4. Specification of Texture Coordinate Values

There are two types of texture coordinate values: the texture coordinate system (S, T, Q), and the texel
coordinate system (U, V). The FST field of the primitive drawing attribute determines which coordinate system
to use. The features of the two types of coordinate systems are:

Coordinate System Texture Coordinate System Texel Coordinate System

Type Two Dimensional Homogeneous Two Dimensional Coordinate System
Coordinate System
Features When s and t (normalized texel When the width in u direction is USIZE
coordinates) obtained from the and the width in v direction is VSIZE in
division of S and T by Q are in the the texture, and U is [0.0, USIZE] and V
[0.0, 1.0] range, they show the whole is [0.0, VSIZE], they show the whole
texture. texture.
Perspective Possible by setting Q to the Impossible
Correction appropriate value. (See "3.4.10.
Perspective Correction".)
MIPMAP Possible Possible
(LOD Value (Fixed value in each primitive) (Fixed value in each primitive)
Specification)
MIPMAP Possible Impossible
(Q Linear
Interpolation)

The relationship between texture coordinates (S, T, Q) and texel coordinates (U, V) is as follows:
u = USIZE x S/Q
v = VSIZE x T/Q
u and v values that the GS can process are in the range of -2047 to +2047. Values exceeding this range are
clamped. This may cause irregularities in texture drawing when s and t values become large.
The GS performs processing by rounding down the lower 8 bits of the mantissa of 32-bit floating-point values
specified to S, T, and Q. This may cause irregularities to the texture on the screen when expanded.
3.4.5. Texture Wrap Modes
When trying to refer to the outside of the valid range of the texture, a processing method (wrap mode) can be
selected. This method enables tiling, the repetition of a small texture. Using antialiasing and bilinear filter in
inappropriately set wrap mode may cause a defect when drawing the boundary lines of the texture, since they
refer to the texels outside the primitive edges.
Assuming the texture width and height are USIZE and VSIZE, when the texel coordinate values (u, v) of a texel
to be processed are outside the range of (0, 0) - (USIZE, VSIZE), the texel coordinate values (u', v') are
calculated by a method specific to each wrap mode.
There are four wrap modes (REPEAT, CLAMP, REGION_CLAMP, and REGION_REPEAT), and they can
be set independently for the horizontal and vertical directions. Settings are made in the WMS and WMT fields
of the CLAMP register.

REPEAT Mode
The original image is mapped repeatedly.
(Formulas)
u' = u % USIZE
v' = v % VSIZE

CLAMP Mode
The outermost color of the texel is enlarged.
(Formulas)
u < 0.0: u' = 0.0
v < 0.0: v' = 0.0
u > USIZE: u' = USIZE
v > VSIZE: v' = VSIZE

REGION_CLAMP Mode
The data of the rectangular area specified in the MINU, MINV, MAXU and MAXV areas in the CLAMP
register becomes effective, and the outermost texel color is enlarged outside the area in the same way as the
CLAMP mode.
(Formulas)
u < MINU: u' = MINU
v < MINV: v' = MINV
u > MAXU: u' = MAXU
v > MAXV: v' = MAXV
When MIPMAP is performed, on the levels other than MIPMAP0, clamping is performed within the range
MINU >>n, MINV >>n, MAXU >>n, and MAXV >>n in the texel coordinate system, assuming the level
value to be n.

REGION_REPEAT Mode
The following operations are applied to the integer parts (uint, vint) of the texel coordinates, and the texel
coordinate values are calculated.
(Formulas)
u' = (uint & UMSK) | UFIX
v' = (vint & VMSK) | VFIX
When using the REGION_REPEAT mode, the pattern, by which a part of the texture area is repeated, can
be used, and a mosaic effect can be achieved.

Example of Original Texture Pattern (64 x 64)

UMSK=0x00f UFIX=0x020 UMSK=0x3f0 UFIX=0x000

VMSK=0x00f VFIX=0x010 VMSK=0x3f0 VFIX=0x000

Figure 3-12 REGION_REPEAT Mode Effect

UMSK, VMSK, UFIX, and VFIX are specified in the CLAMP_1 or CLAMP_2 register. They are the same
bits as the MINU, MINV, MAXU, and MAXV fields respectively, but are processed differently, according to
the wrap mode.
However, the following conditions exist to perform drawing correctly when the bilinear filter is used as a
filter for reading texels.
In the mask pattern of UMSK and VMSK, the bit position where the value is 1 should take 0 as the
corresponding bit values of UFIX and VFIX. This is shown in the formulas below.

UFIX & (1 << (iu + 1) - 1) = 0

VFIX & (1 << (iv + 1) - 1) = 0
(iu/iv shows the position of the most significant bit out of the bits with UMSK/VMSK equal to 1.)

3.4.6. Format Conversion

The GS performs calculations of 8 bits each for RGBA. When the color format of the texel value read from the
texture buffer is not RGBA32, the color format is converted as follows.

CLUT Conversion
When the texel storage formats are IDTEX8 and IDTEX4, the actual color value is obtained via the Color
Look-up Table (CLUT) based on the texel value.

CLUT
Color Value 0
Color Value 1
Texel Value Color Value 2 Format
:
(IDTEX4)
: Conversion
:
Color Value 15

Figure 3-13 Conversion from Texel Value (IDTEX4) to Color Value via CLUT
The CLUT is a color value table of 256 entries with IDTEX8 and 16 entries with IDTEX4. This table is
read temporarily from the local buffer to a temporary buffer and used. (This temporary buffer is different
from the texture page buffer, so the contents of the texture page buffer are not destroyed by using the
CLUT.)
Data load control from the CLUT to a temporary buffer is described in detail in "3.4.7.
CLUT Buffer Control".

RGB Bit Expansion

When the texel value or the color format of the color value converted via the CLUT is RGBA16, 5 bits each
for RGB are expanded into 8 bits as follows:

Input Luminance
Value (5bits) E D C B A

Output
Luminance
Value (8bits) E D C B A 0 0 0

Alpha Value Bit Expansion

When the color formats are RGBA16 and RGB24, the Alpha value of 8 bits is obtained as shown in the table
below based on the TEXA register.

RGBA16
A=0 A=1
AEM = 0 TA0 TA1
AEM = 1 R = G = B = 0 -> 0 TA1
R | G | B is not equal to 0 -> TA0

RGB24
AEM = 0 TA0
AEM = 1 R = G = B = 0 -> 0
R | G | B is not equal to 0 -> TA0

3.4.7. CLUT Buffer Control

The Color Look-up Table (CLUT) for texture mapping is used to convert the texture's index value to a color
value after copying from the CLUT buffer in local memory to the temporary buffer.

Temporary Buffer
Local Memory .
.
CLUT Buffer CSA .
Color Value 0 .
Color Value 1 Color Value 0
Color Value 2 Color Value 1
. Color Value 2 Format
. .
. Load Conversion
.
Color Value 15 .
.
Color Value 15
.
.
.
.

CLUT2

CLUT3
.
.
.
.

Figure 3-14 Loading from CLUT Buffer to Temporary Buffer

When loading to the CLUT for the first time after CLUT data is stored in local memory, it is necessary to
invalidate the texture buffer by accessing the TEXFLUSH register.

Temporary Buffer for CLUT

The temporary buffer is a cache-like memory inside the GS. It has a 1 KB capacity, which is equivalent to the
entries of 256 colors in a 32-bit CLUT Entry Storage Format (CPSM) and 512 colors in a 16-bit CPSM. CSA
specifies which part of the temporary buffer is used.
Entries in the temporary buffer are arranged as follows:

CPSM=PSMCT16,PSMCT16S:(0 to 31 can be specified in CSA)

16 bits
0
:
CSA=0 :
15
16
:
CSA=1 :
31
32
:
:
:
CSA=31
511

CPSM=PSMCT32: (0 to 15 can be specified in CSA)

32 bits
256 0
: :
CSA=0 : :
271 15
272 16
: :
CSA=1 : :
303 31
304 32
: :
: :
: :
CSA=15
511 255

Load Operation
CLUT data is loaded from the CLUT buffer to the temporary buffer when the TEX0 or TEX2 register is
accessed. Only one set of CLUT data is loaded, and values in other areas of the temporary buffer are saved.

CLUT Data Specification

The CLUT data in the source is specified with CBP, CPSM or PSM of the TEX0 or TEX2 register or may
occasionally be specified with TEXCLUT. When the CLUT storage mode in the CLUT buffer is CSM1, the
CLUT must be located at the buffer starting address (the upper left point). When it is CSM2, the CLUT
location in the buffer is specified with TEXCLUT.

Loading Position in Temporary Buffer

CLUT data in the temporary buffer is loaded to the entry address (CSA multiplied by 16) with CSA in the
TEX0 or TEX2 register.

Loading Condition
The condition of loading to the temporary buffer is specified with CLD.

CLD Flag Transmission Control

000 CLUT data is not loaded. (Data in the temporary buffer is stored.)
001 CLUT data is always loaded.
010 CLUT data is always loaded, and internal register CBP0 is rewritten by the value of CBP.
011 CLUT data is always loaded, and internal register CBP1 is rewritten by the value of CBP.
100 CBP0 is compared with CBP, and if different, CLUT data is loaded.
101 CBP1 is compared with CBP, and if different, CLUT data is loaded.
The internal registers CBP0 and CBP1 control loading. Since they remember which CLUT buffer the data
has been loaded from, the temporary buffer can be used as cache memory (only in CSM1).

3.4.8. Texel Sampling

Point sampling or bilinear sampling can be selected when performing filter processing during texel sampling.
The sampling method is specified in the drawing environment register, TEX1.

Point Sampling
Texel coordinates (iu, iv) are calculated from the normalization texel coordinates (s, t) according to the
following formulas, and the color value of the texel is assumed to be the texture value.

iu = Integer Part of (s x USIZE) iv = Integer Part of (t x VSIZE)

Operations such as repetition and clamping are performed to iu and iv depending on the wrap mode.
U 1.0 2.0 3.0
V
1.0

2.0

3.0

Figure 3-15 Texel Selection with Point Sampling Filter

Bilinear Sampling
The texture value is calculated by applying linear interpolation to four texel colors as below. First, texel
coordinates (iu0, iv0), (iu1, iv0), (iu0, iv1), and (iu1, iv1) are calculated from the normalization texel
coordinates (s, t) by the following formulas.

iu0 = Integer Part of (s x USIZE – 0.5) iu1 = iu0 + 1

iv0 = Integer Part of (t x VSIZE – 0.5) iv1 = iv0 + 1

The texture value is calculated by linear interpolation from the following formulas, assuming that the color
values of these four texels are (Ra, Ga, Ba, Aa), (Rb, Gb, Bb, Ab), (Rc, Gc, Bc, Ac), and (Rd, Gd, Bd, Ad)
respectively and that the decimal parts of (s x USIZE - 0.5) and (t x VSIZE - 0.5) are Alpha and Beta
respectively.

R = (1.0- α)(1.0-β)Ra + α (1.0-β)Rb + (1.0-α)β Rc + αβ Rd

G = (1.0- α)(1.0-β)Ga + α (1.0-β)Gb + (1.0-α)β Gc + αβ Gd
B = (1.0- α)(1.0-β)Ba + α (1.0-β)Bb + (1.0-α)β Bc + αβ Bd
A = (1.0- α)(1.0-β)Aa + α (1.0-β)Ab + (1.0-α)β Ac + αβ Ad

Operations such as repetition and clamping are performed to iu0, iv0, iu1, and iv1 depending on the
wrapping mode.

1.0 2.0 3.0

U
V
1.0

a b

2.0

c d

3.0

Figure 3-16 Texel Selection with Bilinear Filter

3.4.9. Texture Function

The colors (Rt, Gt, Bt, At) obtained from the texture are blended with the colors (Rf, Gf, Bf, Af) of the
fragment obtained from the primitive by the DDA and converted into output colors (Rv, Gv, Bv, Av). At this
time, according to the texture function specified by the TFX flag of the TEX0 register and the mode specified
by the TCC flag, the conversion operation in the following table is applied. When the TCC flag is RGBA and
the texel format is TEX 24 or TEX16, the value set in the TEXA register is used as the Alpha value.

Function TCC=RGB TCC=RGBA Description

MODULATE Rv = Rt * Rf Rv = Rt * Rf Adjusts the brightness of the texture
Gv = Gt * Gf Gv = Gt * Gf color according to the fragment color.
Bv = Bt * Bf Bv = Bt * Bf When the value of the fragment color
Av = Af Av = At * Af is 0x80, the brightness of the output
color is the same as that of the original
texture.
DECAL Rv = Rt Rv = Rt Outputs the texture color as is.
Gv = Gt Gv = Gt
Bv = Bt Bv = Bt
Av = Af Av = At
HIGHLIGHT Rv = Rt * Rf + Af Rv = Rt * Rf + Af Adjusts the brightness of the texture
Gv = Gt * Gf + Af Gv = Gt * Gf + Af color as in the case of MODULATE,
Bv = Bt * Bf + Af Bv = Bt * Bf + Af then adds a highlight in pure white
Av = Af Av = At + Af based on the alpha value of the
fragment color. Suitable for
highlighting translucent polygons.
HIGHLIGHT2 Rv = Rt * Rf + Af Rv = Rt * Rf + Af Similar to HIGHLIGHT, but stores
Gv = Gt * Gf + Af Gv = Gt * Gf + Af the alpha value of the texture color.
Bv = Bt * Bf + Af Bv = Bt * Bf + Af Suitable for highlighting opaque
Av = Af Av = At polygons.

However, the operator * means A*B = (A x B) >> 7, and the calculation result is clamped between 0 and 0xff.
(When the value of the fragment color is 0x80, the brightness of the texture corresponds to the brightness of the
output color.)

3.4.10. Perspective Correction

The GS has the ability to correct texture perspective distortion. This is done by giving the value, to which
appropriate preprocessing is performed in the texture coordinate system (S, T, Q), to the GS as the vertex
coordinate values in the texture space.
When drawing a three-dimensional primitive, the point (Xv, Yv, Zv) in the view coordinate system usually
shown in the rectangular parallelepiped is converted into the point (Xp, Yp, Zp) in the primitive coordinate
system shown in the truncated quadrangular-pyramid by transparency perspective conversion. In the primitive
coordinate system, a place close to the viewpoint is expanded, and a place far from the viewpoint is reduced and
distorted.

Z Z

X X
Y Y

View Coordinate System Primitive Coordinate System

Figure 3-17 Coordinate System in Transparency Perspective Conversion

This distortion is caused by transparency perspective conversion. Calculations for linear interpolation for DDA
etc. are made when drawing a primitive to the frame buffer. This is performed in the primitive coordinate
system. However, since the relation between texture and primitive is defined in the view coordinate system, a
positional deviation from the primitive is caused if the texture coordinate values are linear-interpolated as they
are when drawing the primitive. This is the so-called perspective distortion of the texture.
To avoid this, equivalent conversion is performed to the texture coordinates as well. Let the texture coordinates
of the primitive vertex in the view coordinate system (the world coordinate system or local coordinate system is
also OK) be (sv, tv) and the divisor in the transparency perspective conversion be W. The texture coordinates
(S, T, Q) in the primitive coordinate system are gained from the formulas below.

S = sv / W T = tv / W Q=1/W
Proper perspective correction can be achieved by giving these S, T, and Q to the GS. (After linear-interpolating
S, T, and Q, the GS calculates the normalization texel coordinate values from s = S/Q and t = T/Q.)
Moreover, perspective distortion is not generated for Sprites, where each fragment always takes the same Z
value. In such a case, texel coordinates for the vertex of each primitive can be specified directly. (Note that they
are not the normalization texel coordinates.)
Whether the homogeneous texture coordinates (S, T, Q) or texel coordinates (U, V) are used to specify the
texture coordinates is determined by the FST flag of the PRIM register or PRMODE register. Regarding the
primitive vertex, S and T values are set in the ST register and Q value is set in the Q area of the RGBAQ
register. Moreover, U and V values are set in the UV register.

3.4.11. MIPMAP
MIPMAP textures can be set in seven stages from level 0 to level 6. The texture height and width should each
be a power of 2. When MIPMAP level 0 texture is 2m x 2n in sizing, levels 1 and 2 are set to 2m-1 x 2n-1 and 2m-2
x 2n-2 respectively in texture sizing. Sizing is set in succession by reducing the height and width by half, until
either the width or height reaches 1 for point sampling or 8 for bilinear sampling.
When MIPMAP is not performed, level 0 is used.

Calculation of Texture Base Pointer

The user can set the base pointer of the MIPMAP texture at each level to the desired position in the same
way as texture level 0. However, because the page size changes according to the pixel storage format of the
texture, it is necessary to set the base pointer so that the textures do not overlap.
In the example below, the MIPMAP is composed assuming a texture of 512 x 512 texels is level 0 when the
format is PSMCT16.

MIPMAP Base Pointer Buffer Texture Width Texture Height

Level Width
0 7200 8 9 9
1 9248 4 Not specified Not specified
2 9760 2 Not specified Not specified
3 9888 1 Not specified Not specified
4 9920 1 Not specified Not specified
5 9928 1 Not specified Not specified
6 9930 1 Not specified Not specified

The minimum value at base pointer intervals by which the textures of different MIPMAP levels are stored
changes depending on the texture format, texture size and ratio of height and width.

Automatic Calculation of Base Pointer

The base pointer for texture levels 1 to 3 can be calculated automatically on the condition that texture format
and size must meet the following requirements.

• PSMCT32/PSMCT24/PSMT8H/PSMT4HL/PSMT4HH
Width = Height = {32, 64, 128, 256, 512}
• PSMCT16/PSMT8/PSMT4
Width = Height = {32, 64, 128, 256, 512, 1024}

With automatic calculation, textures up to level 3 are stored in a continuous memory area, as in the examples
above. When the texture is loaded, it is necessary to store the texture at the position of the base pointer
calculated automatically.
The automatic calculation is executed when the TEX0 register is set after the MTBA field of the TEX1
register is set to 1.
For the texture of MIPMAP levels 1 to 3, the base pointer is specified in the fields TBP1 to TBP3 of the
MIPTBP1 register and the buffer width is specified in the field TBW1 to TBW3. As for the texture of
MIPMAP levels 4 to 6, the base pointer is specified in the fields TBP4 to TBP6 of the MIPTBP2 register and
the buffer width is specified in the field TBW4 to TBW6. Moreover, the maximum MIPMAP level used is
set to the MXL flag in the TEX1 register.

3.4.12. LOD Setting

The texture level and filter used for mapping are controlled by the value of the LOD (Level of Detail). There
are two methods of calculating the LOD, and they are specified in the LCM fields of the TEX1 register. One
method is to obtain the LOD from the Q value of each primitive vertex and the L and K fields of the TEX1
register according to the formula below, and the other is to specify the LOD in the K field only. When LCM is
set to 0, LOD is calculated from the Q value even if the texture coordinates are given by UV.

LCM = 0 : LOD = ( log2(1/Q) << L) + K

LCM = 1 : LOD = K

When the calculated LOD value is 0.0 or less, the level 0 texture is used, and the filter specified in the MMAG
field of the TEX1 register is used.
When the LOD value is more than 0.0, the level 0 to 6 textures are used depending on the LOD value, and the
filter specified by the MMIN flag of the TEX1 register is used.
The following table shows the relation between the MMIN flag and the filter used.

MMIN Flag Level m Filer Level m+1 Filter Processing between Levels
0 Disabled Disabled Level 0 point sampling
1 Disabled Disabled Level 0 bilinear
2 Point sampling Point sampling Selection of m or m+1 by LOD
(round-off)
3 Point sampling Point sampling Linear interpolation by LOD
4 Bilinear Bilinear Selection of m or m+1 by LOD
(round-off)
5 Bilinear Bilinear Linear interpolation by LOD
(trilinear)
*m is the integer part of the LOD value.

The bilinear filter in each MIPMAP imposes almost no speed penalty, but the linear interpolation filter between
levels affects the drawing time.
Moreover, when the number of pixels for the height or width of the texture is four or less, correct images
cannot be obtained with bilinear and trilinear filters.

Specifications of L and K
The weight of the MIPMAP filter is specified by L, and the position of MIPMAP level 0 is set by K.
Assuming the fragment position, where the ratio of the texel to the frame buffer pixel is 1 to 1, to be Z0 in
texture mapping, K is set by the following formula.

K = - ( log2(Z0 / h ) << L )

However, h is assumed to be the distance between the viewpoint and screen, and Q of each vertex of the
primitive is assumed to be h/Z. The relation between Z and LOD is shown in the figure below.

Z
Viewpoint

LOD
0 1 2

Figure 3-18 LOD when L = 0 and K = - log2( Z0 / h )

When mapping the texture to the primitive parallel to the screen, it is appropriate to set L = 0, in which the
reduction on the display corresponds to the increase of LOD. However, when the primitive is inclined to the
screen, a value more than 0 is set to L.
Because the effect of MIPMAP changes with the texture pattern, appropriate L and K values should be set
according to the texture used. To control the MIPMAP levels individually in primitives units, the value of Q can
be directly controlled.

3.4.13. Texture Mapping Procedure

The following is an example of a register setting procedure for texture mapping. (The registers necessary for
general drawing should already be set.)

1. Transmit the texture to the local buffer.

2. Transmit the CLUT to the local buffer.
3. Set the TEXCLUT register (if necessary).
4. Access the TEXFLUSH register (only once at first).
5. Set the TEXA register.
6. Set the TEX1_1 | TEX1_2 register.
7. Set the TEX0_1 | TEX0_2 register.
8. Set the MIPTBP1_1 | MIPTBP1_2 register.
9. Set the MIPTBP2_1 | MIPTBP2_2 register.
(Repeat the following.)
10. Set the PRIM register.
11. Set the ST or UV register.
12. Set the RGBAQ register.
13. Set the XYZF register.

3.5. Fog Effect

The Fog effect is achieved by blending the color of each texel output from the texture function and the Fog
color. Let the output colors from the texture function be (Rv, Gv, Bv, Av); let the Fog coefficient, which has
been obtained by linear-interpolating the Fog value set for each vertex of the primitive, be F; and the Fog colors
written to the FOGCOL register be (Rfc, Gfc, Bfc). The blending operation is shown in the formulas below.

R = F * Rv + ( 0xff – F )* Rfc
G = F * Gv + ( 0xff – F )* Gfc
B = F * Bv + ( 0xff – F )* Bfc
A = Av

However, A*B = ( A x B) >> 8.

The Fog coefficient is stored in the F field of the XYZF or XYZF2 register. The Fog color is stored in the
FOGCOL register. Whether or not Fogging is performed is specified with the FGE flag of the PRIM register.

3.6. Antialiasing
In the GS, Antialiasing can be performed for Line, LineStrip, Triangle, TriangleStrip, and TriangleFan. This
process is performed for each line or triangle.
3.6.1. Principle of Antialiasing
Antialiasing is achieved by calculating the coverage Cov (ratio of the area that covers the pixel) to each pixel at
the edge of the GS primitive, assuming this to be the Alpha value, and performing Alpha-blending to the
destination color (color distant from the primitive) and the primitive color.
The effect of Antialiasing is different depending on the combination of the AA1 flag to set Antialiasing, ABE
flag to set Alpha-blending, and the Alpha value calculated for the pixel. This is shown in the table below. Cov
shows the value of the coverage of the pixel. When the coverage is 100%, Cov = 0 x 80 (128 in decimal) is
obtained.

AA1 ABE Alpha Value to be Output Antialiasing Blending

Flag Flag Effect Effect
0 0 A NO NO
0 1 A NO YES
1 0 Cov YES NO
1 1 Cov (Pixel with A equal to 0x80) YES NO
1 1 A (Pixel with A not equal to 0x80) NO YES

3.6.2. Antialiasing Control

Drawing Order
To perform Antialiasing appropriately, the part distant from the primitive should be drawn before drawing
the primitive. It is necessary to sort all the drawing primitives in positional order (far to close) to achieve
ideal Antialiasing.

Write to Z Buffer
Since Antialiasing is performed to each line or triangle, how to deal with the parts shared by lines and
triangles becomes a problem. In the GS, to guarantee the connection of the image in the boundary part,
writing to the Z buffer is not performed for pixels with coverage Cov less than 0x80.
As for the line, Cov is always less than 0x80. Therefore, the Z buffer is not written when the AA1 flag is 1.
Antialiasing is controlled by the AA1 flag of the PRIM or PRMODE register. Moreover, it is necessary to
set the flags of the ALPHA register for blending, as shown below.

A = Cs (Source Color)
B = Cd (Destination Color)
C = As (Source Alpha)
D = Cd (Destination Color)

3.7. Pixel Test

The pixel test is the process of inspecting whether or not a pixel to be drawn meets specified conditions. It also
controls drawing processing for pixels that failed. The pixel test does not change the pixel value.
The pixel test consists of the following test items, which are executed in this order.

Pixel Test Test Contents

Scissoring Test Test by the position where the pixel is drawn.
Alpha Test Comparison test between the pixel's alpha value and standard value.
Destination Test by the pixel's alpha value at the drawing destination in the frame buffer.
Alpha Test
Depth Test Comparison test between the pixel's Z value and the corresponding Z value in
Z buffer.

From the point of view of GS internal processing, the scissoring test is carried out in the Rasterizing (DDA)
stage and not in the pixel test stage. For the convenience of the user, however, it is described in this section.

Processes succeeding the pixel test such as Alpha-blending are performed even to a pixel that failed the alpha
test, destination alpha test, or depth test, and the pixel is finally controlled in drawing when the pixel value is
written in the frame buffer and the Z buffer. It may appear as if the pixel test were performed after Alpha-
blending.
3.7.1. Scissoring Test
This test checks whether the coordinate values of the pixel to be drawn are in the rectangular (scissoring) area
specified in the window coordinate system. Pixels judged to be outside the scissoring area are not processed
further.

Set Register
SCISSOR_1, SCISSOR_2

Test Contents
The pixel is judged by the following formulas, assuming the pixel coordinates are (x, y). A pixel located on
the boundary of the scissoring area passes.

scissor_test(int x, y)
{
if (x < SCISSOR.SCAX0 || x > SCISSOR.SCAX1) return(FAIL);
if (y < SCISSOR.SCAY0 || y > SCISSOR.SCAY1) return(FAIL);
return(PASS);
}

The scissoring test cannot be omitted.

3.7.2. Alpha Test

The Alpha value of the drawing pixel and the preset standard alpha value are compared. Processing is controlled
if the pixel does not meet the comparison condition that was set.

Set Register
TEST_1, TEST_2 (ATE, ATST, AREF, and AFAIL fields)

Test Contents
ATE ATST Test Contents
0 - No testing. (All pixels pass.)
1 NEVER All pixels fail.
ALWAYS All pixels pass.
LESS Pixels less than AREF pass.
LEQUAL Pixels less than or equal to AREF pass.
EQUAL Pixels equal to AREF pass.
GEQUAL Pixels greater than or equal to AREF pass.
GREATER Pixels greater than AREF pass.
NOTEQUAL Pixels not equal to AREF pass.

A pixel that failed the alpha test is not controlled completely in drawing, but can be specified in the AFAIL
field in the TEST register so that the RGB, A, and Z values are controlled individually. Details are as
follows.

Alpha Test Result AFAIL Setting RGB A Z

Failed KEEP Controlled Controlled Controlled
FB_ONLY Effective Effective Controlled
ZB_ONLY Controlled Controlled Effective
RGB_ONLY* Effective Controlled Controlled
*RGB_ONLY is effective only when the color format is RGBA32.
In other formats, operation is made with FB_ONLY.

Controlled values do not become effective even if they pass in other pixel tests after this.

3.7.3. Destination Alpha Test

This test checks the Alpha value of the pixel for drawing in the frame buffer (Destination Alpha Value). A
failed pixel is controlled in drawing.

Set Register
TEST_1, TEST_2 (DATE and DATM fields)

Test Contents
The contents of the Destination Alpha Test depend on the pixel storage mode of the frame buffer.

Frame Buffer: PSMCT32 (RGBA32)

DATE DATM Test Contents
0 - No testing. (All pixels pass.)
1 0 Pixels with bit 7 of A set to 0 pass.
1 1 Pixels with bit 7 of A set to 1 pass.

Frame Buffer: PSMCT16 (RGBA16)

DATE DATM Test Contents
0 - No testing. (All pixels pass.)
1 0 Pixels with A set to 0 pass.
1 1 Pixels with A set to 1 pass.

Frame Buffer is PSMCT24 (RGB24)

Because the frame buffer does not have A value, all pixels pass regardless of the setting of DATE and
DATM.
3.7.4. Depth Test
This test compares the Z value of the drawing pixel and the corresponding Z value in the Z buffer. A failed
pixel is controlled in drawing.

Set Register
TEST_1, TEST_2 (ZTE and ZTST fields)

Test Contents
ZTE ZTST Test Contents
0 - (Disabled)
1 NEVER All pixels fail.
ALWAYS All pixels pass.
GEQUAL Pixels greater than or equal to Z buffer value pass.
GREATER Pixels greater than Z buffer value pass.
Setting ZTE to 0 is prohibited since it may cause a malfunction. To omit the depth test, set ZTE to 1 and
ZTST to ALWAYS, and the ZMSK field of the ZBUF register to 1 respectively. Due to these settings, the Z
buffer is neither accessed nor updated for any pixels. As a result, the operation becomes the same as the one
without the depth test.

3.8. Alpha Blending

Blending calculation can be made between the depth test output color (Source Color) Cs and the frame buffer
pixel color (Destination Color) Cd.
Calculation of RGB values is performed. With Alpha value A, the source Alpha value (As) is written in the
frame buffer as is. However, when the FBA flag of the FBA register is set, a prescribed operation is performed.
Alpha-blending is set ON/OFF by the ABE flag of the PRIM or PRMODE register. When performing
antialiasing (setting AA1 to 1), the blending function operates automatically. Therefore, the ABE flag must
always be set to 0 (OFF).

3.8.1. Blending Setting

The formula for blending is shown below. A, B, C, and D in the formula can be set to take the value in the
following table with the flag of the ALPHA register. FIX is the fixed Alpha value set in the ALPHA register.

Output Color = (A - B) * C + D
However, X * Y = (X x Y) >> 7

Flag Selectable Color

A Cs, Cd, 0
B Cs, Cd, 0
C As, Ad, FIX
D Cs, Cd, 0

When the Alpha value is 0x80, the multiplier to the source color becomes 1.0. An example of flag setting
applied to normal blending is shown as follows. When Antialiasing AA1 is performed, the same setting is made.

A B C D Equivalent Formula
Cs Cd As Cd Cs*As + Cd*(0x80 - As)

Additionally, various settings such as subtraction and brightness addition/decrease are possible.

A B C D Equivalent Formula
Cd Cs 0x80 0 Cd - Cs
Cd 0 As Cd Cd*(0x80 + As)
0 Cd As Cd Cd*(0x80 - As)

In the RGBA16 mode, where Destination Alpha value is 1 bit only, Alpha is treated as 0 when A is 0 and as
0x80 when A is 1. In the RGB24 mode, Alpha is treated as 0x80.
The result of Alpha-blending is passed to the following Dithering processing as is, without being clamped.
3.8.2. PABE Flag
When the PABE flag is 1, the value of MSB of the Source Alpha determines whether to perform Alpha-blending
per pixel. Alpha-blending is ON when MSB is 1 and OFF when MSB is 0.

3.9. Writing to the Frame Buffer

The following processing is performed according to the pixel storage mode before the pixel value, the output
from the pixel pipeline, is written in the frame buffer.

RGB A Z

A Value
Dithering
Correction

Clamping

Format Conversion

Frame Buffer Z Buffer

Figure 3-19 Write Process to Frame Buffer

3.9.1. Dithering
If the frame buffer color format is RGBA16, the RGB luminance values are each converted from 8 bits to 5 bits.
At this time, it is possible to decrease the Mach band by dithering.
Dithering is set ON/OFF by the DTHE register.
Dithering is performed by adding the offset value selected according to the coordinates of the pixel from the
dither matrix of 4 x 4 set in the DIMX register to each luminance value of the RGB of the pixel. It is shown in
the following formulas.

Rout = Rin + DIMX[Y%4][X%4]

Gout = Gin + DIMX[Y%4][X%4]
Bout = Bin + DIMX[Y%4][X%4]

X,Y : Window Coordinate Values of Pixel in Frame Buffer

Rin, Gin, Bin : Input Luminance Value (8 bits)
Rout, Gout, Bout : Output Luminance Value (8 bits)

Dither Matrix Setting Example

-4 2 -3 3
0 -2 1 -1
-3 3 -4 2
1 -1 0 -2

When the color format of the frame buffer is RGBA32 or RGB24, the result of dithering is not guaranteed.

3.9.2. Color Clamp

Since the RGB value of a pixel occasionally exceeds the range of 0-255 after operations such as Alpha-blending,
the result is stored with 9 bits for each RGB value. It is possible to select clamping to set this value within the
range of 0-255, or to extract the lower 8 bits. Whether or not to perform clamping is determined by the
COLCLAMP register.

3.9.3. Alpha Value Correction

It is possible to correct the pixel alpha value with the value previously set in the FBA register. This is done
when reusing the image data in the frame buffer as a texture.

RGBA32
FBA

bit 7 OR
bit 6

Input A Value Output A Value

bit 0

RGBA16
FBA

Input A Value bit 7 OR Output A Value

Figure 3-20 Alpha Value Correction

3.9.4. Format Conversion

Pixel color values are packed into a fixed number of bits, according to the pixel storage mode set by the
FRAME_1 or FRAME_2 register, and written in the frame buffer.

RGBA32 (PSMCT32) Mode

A B G R
7 0 7 0 7 0 7 0

A B G R
31 24 23 16 15 8 7 0

RGB24 (PSMCT24) Mode

A B G R
7 0 7 0 7 0 7 0

Not Used
B G R
23 16 15 8 7 0

RGBA16 (PSMCT16,PSMCT16S) Mode

A B G R
7 0 7 3 0 7 3 0 7 3 0

A B G R
R
15 14 10 9 5 4 0

3.9.5. Masking
The mask can be set so that only the prescribed bits are written when writing data to the frame buffer.
The value of the mask is specified in the FBMSK field of the FRAME_1 or FRAME_2 register. Since the bit
position of the FBMSK corresponds to the pixel value before format conversion, special care is needed when
the pixel format of the frame buffer is RGBA16. The relation between FBMSK and the frame buffer in the case
of RGBA16 is shown as follows.

31 24 23 19 15 11 7 3 0
FBMSK

Frame Buffer A B G R
15 14 10 9 5 4 0

Regarding the Z buffer, it is possible to control only whether or not to write all the bits.
This is specified with the ZMSK of ZBUF_1 or ZBUF_2 register.

4. Image Data Transmission

The GS supports three modes of image data transmission, according to the direction of transmission:

• Transmission from host to local buffer

• Transmission from local buffer to host
• Transmission from local buffer to local buffer

The local buffer is the local memory in the GS, and the host is the external device connected to the GS.
When transmitting data, the arrangement of the pixels in local memory, or the pixel storage format, is converted
according to the specification. At this time, however, no processes involving changes to the pixel value (e.g.
color reduction, Alpha value operation, and masking) are performed. When transferring between local buffers,
transmission is supported only between buffers that have the same number of bits per pixel.

4.1. Transmission Parameters

When transmitting data, it is necessary to set the following parameters.

4.1.1. Destination Buffer / Source Buffer (BITBLTBUF Register)

For each buffer from/to which data is transmitted, the base pointer, buffer width, and pixel storage format are
set in the BITBLTBUF register.
When transmitting from host to local buffer, only the destination buffer is set. When transmitting from local
buffer to host, only the source buffer is set. When transmitting from local buffer to local buffer, both the
destination and source buffers are set.
The buffer base pointer is specified as a 14-bit unsigned integer in units of 64 words. The buffer width is
specified by a 6-bit unsigned integer in units of 64 pixels.
All pixel storage formats can be used, except that when transmitting from local buffer to host, pixel storage
formats PSMT4, PSMT4HL and PSMT4HH cannot be used. When transmitting from local buffer to local
buffer, the bit length of the pixels to be stored respectively in the source and destination buffer should be the
same.

4.1.2. Offset in Buffer (TRXPOS Register)

The TRXPOS register specifies the offset coordinates at the upper left point of the transmission area in the
transmission buffer and the pixel transmission direction. When transmitting from host to local buffer, only the
settings for the destination buffer are made. When transmitting from local buffer to host, only the settings for
the source buffer are made. And when transmitting from local buffer to local buffer, settings are made for both
the source and destination buffers.
The four figures below show the direction of the pixel transmission of the transfer data within the transmission
destination area. This setting is enabled only when transmitting from local buffer to local buffer, and when
transmitting from host to local buffer and local buffer to host, data is arranged in the direction from left to right
and top to bottom. The transmission starting point in the rectangular area varies according to the pixel
transmission direction.

Start Start

0. Left to Right 1. Left to Right 2. Right to Left 3. Right to Left

Top to Bottom Bottom to Top Top to Bottom Bottom to Top

Figure 4-1 Pixel Transmission Direction

4.1.3. Width and Height of Transmission Area (TRXREG Register)

The TRXREG register specifies the width and height of the transmission area.
When transmitting from host to local buffer, settings for the destination buffer are made. When transmitting
from local buffer to host, settings for the source buffer are made. And when transmitting from local buffer to
local buffer, settings are made for the destination buffer. (The width and height of the transmission area in the
destination and the source should be the same.)
When the coordinate value of the transmission area is 2048 or more, it wraps around at 2048. That is, if the
transmission start coordinates are (SAX, SAY), the coordinates of the pixel to be transmitted actually are (X, Y)
and the offset coordinates from the transmission start coordinates of the pixel are (RRX, RRY), then:

X = (SAX + RRX) % 2048

Y = (SAY + RRY) % 2048

The figure below shows the relation between transmission area parameters when the pixel transmission direction
is left to right and top to bottom.
Buffer Width
(SBW/DBW)
Address
(SBP/DBP) (SSAY/DSAY)

(SSAX/
DSAX)
RRH

RRW

Figure 4-2 Parameters to Show Transmission Area

4.1.4. Transmission Direction and Transmission Start (TRXDIR Register)

This register specifies the directions of transmission. The following three can be specified.
• Host to local buffer
• Local buffer to host
• Local buffer to local buffer
The GS starts transmission immediately after the TRXDIR register is accessed.

4.1.5. Limitations on Transmission Start Coordinates and Width

The table below shows the limitations on the start coordinates and transmission area width according to the
pixel storage format. These limitations do not apply to the transmission from local buffer to host.

Pixel Storage Limitation on Limitation on

Format Start X Coordinate Width
PSMCT32 No limitation Multiples of 2
PSMZ32
PSMCT24 No limitation Multiples of 8
PSMZ24
PSMCT16 No limitation Multiples of 4
PSMCT16S
PSMZ16
PSMZ16S
PSMT8 Multiples of 2 Multiples of 8
PSMT8H
PSMT4 Multiples of 4 Multiples of 8
PSMT4HL
PSMT4HH

4.2. Transmission Process

4.2.1. Transmission from Host to Local Buffer
Transmission of image data from host to local buffer is performed by writing data to the transmission data
register (HWREG) after setting the parameters described above.
Data is written to HWREG in 64-bit units. During data transmission, it is possible to access other registers and
to issue drawing primitives. It is guaranteed that in the GS, transmission data will not overtake a drawing
primitive that was issued before setting the transmission direction setting register (TRXDIR).
If the TRXDIR has been accessed before transmission is ended (before a fixed amount of data are written to
HWREG), transmission is ended, and a new transmission is started using the current values of BITBLTBUF,
TRXPOS and TRXREG.
To perform drawing using the transmitted data as a texture immediately after data transmission, it is necessary to
access the TEXFLUSH register before primitive drawing because the order of writing to local memory is not
guaranteed.

4.2.2. Transmission from Local Buffer to Host

When transmitting image data from local buffer to host, it is necessary to switch the transmission directions of
the bus between the GS and host after checking the GS input FIFO is empty. The process is as follows:

1. Set the transmission parameters. (See "4.1. Transmission Parameters".)

2. Access the FINISH register. (FINISH event occurs when data is input to the GS.)
3. Wait until the FINISH field of the privileged port CSR register becomes 1.
4. Set the FINISH field of the CSR register to 0 and clear the event.
5. Write 1 to the privileged port BUSDIR register.
(The transmission direction between the GS and host is reversed.)
6. Read data from Host interface.
7. Write 0 to the BUSDIR register after reading all data.
(The transmission direction is changed to the original one.)

The host interface has a built-in FIFO for access to the general-purpose port register, and is usually put in the
direction from the host to the inside of the GS. It is necessary to use this FIFO in the reverse direction only in
the image transmission from the local buffer to the host. The general-purpose port register cannot be accessed
while the FIFO is in the reverse direction. However, the privileged port register can be accessed.
The Host FIFO requires the total data size of Local-Host transmission to be multiples of 128 bytes during DMA
transmission and multiples of 16 bytes during IO transmission.

4.2.3. Transmission from Local Buffer to Local Buffer

Image data transmission from local buffer to local buffer starts when the above-mentioned parameters are set
and the TRXDIR register is accessed.
The color formats of the source and destination buffers must have the same number of bits per pixel (pixel
depth). If the pixel depth is the same, conversion between pixel storage formats with different pixel
arrangement (e.g. PSMT4 and PSMT4HH) is possible.

4.2.4. Restrictions on Drawing after Transmission

A problem may occur when performing drawing immediately after performing an image data transmission to the
local buffer. If both the pixel to be drawn and the pixel most recently written during the image data
transmission are on the same page, and no page break has occurred, the z value will be written to an invalid
address (equivalent to ZBP = 0).
This phenomenon should present no problems under normal processing patterns. However, it should be
considered in situations such as transmitting image data to an area in the frame buffer, or when drawing a
primitive (using the Z buffer) to a texture area.
The following sequence is a workaround to this problem:
Transmit image data
Force a page break via a dummy data transmission to an address
Begin drawing
An alternative workaround would be to force a page break while preventing the write operation to the Z buffer
via t he alpha test, by drawing a dummy primitive (such as a LINE primitive).
In the case of local-to-local transmissions, changing pixel transmission directions may correct the problem.

4.3. Transmission Data Format

In transmitting the image data, the data is packed according to the following data format. Even if it is a
transmission to a buffer with a storage format that leaves blanks such as PSMCT24 and PSMT8H, padding is
unnecessary and the data packed as below can be transmitted.
Packed data is read after the transmission from local buffer to host.

RGBA32/TEX32/Z32
63 55 47 39 31 23 15 7 0
Texel 1 Texel 0
A1 B1 G1 R1 A0 B0 G0 R0
Texel 3 Texel 2
A3 B3 G3 R3 A2 B2 G2 R2

RGB32/TEX24/Z24
63 55 47 39 31 23 15 7 0
Texel 2 Texel 1 Texel 0
G2 R2 B1 G1 R1 B0 G0 R0
Texel 4 Texel 3
R5 B4 G4 R4 B3 G3 R3 B2

RGBA16/TEX16/Z16
63 47 31 15 0
Texel 3 Texel 2 Texel 1 Texel 0
A B3 G3 R3 A B2 G2 R2 A B1 G1 R1 A B0 G0 R0
Texel 7 Texel 6 Texel 5 Texel 4
A B7 G7 R7 A B6 G6 R6 A B5 G5 R5 A B4 G4 R4

IDTEX8
63 55 47 39 31 23 15 7 0

Texel 7 Texel 6 Texel 5 Texel 4 Texel 3 Texel 2 Texel 1 Texel 0

Texel 15 Texel 14 Texel 13 Texel 12 Texel 11 Texel 10 Texel 9 Texel 8

IDTEX4
63 59 55 51 ..... 15 11 7 3
T15 T14 T13 T12 ..... T3 T2 T1 T0

T31 T30 T29 T28 ..... T19 T18 T17 T16

(This page is left blank intentionally)

5. CRTC

5.1. CRTC Function

5.1.1. PCRTC Features
The GS's video output is performed in the PCRTC block as shown in Figure 5-1. The PCRTC has the
following functions:

• Image signals can be output with video clocks conforming to VESA standard (135MHz at maximum),
NTSC, and PAL by setting the standard video clock and the CRT mode.
• There are two independent rectangular area read output circuits. Input area, input pixel format, output
resolution, output size, output screen position, etc. can be set independently. Also, two output images can
be alpha-blended with BGColor and output using the built-in merge circuit (see Figure 5-2).
• There is a rectangular area write input circuit allowing a rectangular area in the output feedback image to be
written to an area in the frame or texture buffer. Also, RGB -> YCbCr conversion can be performed at
that time.

Clock NTSC/PAL

Memory Sync Generator PCRTC

Controller
PLL

Rectangular Area Read Output DAC

Circuit 1

Rectangular Area Read Output

Frame Circuit 2
Buffer Memory
(ASIC DRAM) Controller

Input Circuit

M
Rectangular Area Write Input
Circuit
U
X

Figure 5-1 PCRTC Block Diagram

The following is an example of PCRTC data flow.

Sync Gen.

Frame Buffer NTSC/PAL/VESA

Rectangular Area Read Output Merge Circuit

Circuit 1

out1 CRT1
1 Output
0 1
Rectangular Area Read Output
Circuit 2

out2
1 1

Rectangular Area Write Input

Circuit

Figure 5-2 PCRTC Merge Circuit

As shown in the above figure, two rectangular area read output circuits read rectangular data in different sizes
and different pixel formats located in different areas of the frame buffer, arrange the data to arbitrary positions
with different resolutions, and output to the merge circuit. In the merge circuit, two output images are blended
at the set alpha value and are output as the CRT1 image.
In the above figure, the rectangular area read output circuit 1 controls the high-resolution text image in a single
buffer, and rectangular area read output circuit 2 controls the three-dimensional polygon image with a Z buffer
by RGBA 32-bit dual buffering. The two images are blended in the merge circuit, and a blended image of the
high-resolution text image and 24-bit full color three-dimensional image can be seen.
The rectangular write circuit can write a specified rectangular area in the CRT output image to a specified area in
the frame buffer at an arbitrary sampling rate (feedback write).

5.1.2. Corresponding Video Signal

The following are the display frequencies and typical resolutions that the GS supports.
Mode Resolution fv(Hz) fH(kHz) Remarks
NTSC* 256 x 448(224) 59.940 15.734 ( ): Resolution in non-
320 x 448(224) (59.82) interlaced scanning
384 x 448(224)
512 x 448(224)
640 x 448(224)
PAL* 256 x 512(256) 50.000 15.625 ( ): Resolution in non-
320 x 512(256) (49.76) interlaced scanning
384 x 512(256)
512 x 512(256)
640 x 512(256)
VESA 640 x 480 59.940 31.469 Frequencies may vary
72.809 37.861 within VESA standard
75.000 37.500 (+/- 0.5%).
85.008 43.269
800 x 600 56.250 35.156 Frequencies may vary
60.317 37.879 within VESA standard
72.188 48.077 (+/- 0.5%).
75.000 46.875
85.061 53.674
1024 x 768 60.004 48.363 Frequencies may vary
70.069 56.476 within VESA standard
75.029 60.023 (+/- 0.5%).
84.997 68.677
1280 x 1024 60.020 63.981Frequencies may vary
75.025 79.976within VESA standard
(+/- 0.5%).
DTV 720 x 480 59.94 31.469 480 P
1920 x 1080 60.00 33.750 1080 I
*Resolutions in NTSC and PAL modes are recommended to realize the appropriate screen size.

5.2. Video Output

5.2.1. Video Clock
Besides the system clock, the GS internally generates the standard video clock VCK, which defines the PCRTC
operation timing. The main settings related to the display area on the screen are done in VCK units. The VCK
frequency predetermined for each video mode is set by the kernel service.

5.2.2. Rectangular Area Read Output Circuit

This circuit reads the specified rectangular area in the frame buffer adjusting to CRT scanning, and outputs the
converted data to the display coordinates. It can display the contents of the read rectangular area magnified by
the specified coefficients (MAGH/MAGV).
The figure below shows the operation of this circuit.

Local Memory

Display
FBP Frame Memory DX
Coordinates
DBX
DY DW+1
DBY

X MAGH
DH+1
X MAGV

FBW

Figure 5-3 Read from Frame Buffer to PCRTC

The PCRTC is equipped with two independent rectangular area read output circuits. Each of them
independently has a DISPFB1/2 register (FBP, FBW, DBX, DBY, PSM) and DISPLAY1/2 register (DH, DW,
DX, DY, MAGH, MAGV), where the parameters in the above figure and the input pixel format can be set. The
two circuits function almost the same, but differ in the merging order and some of the input pixel format
settings in the merge circuit (described later in this document).
The unit of horizontal coefficients, DX, DW, and MAGH, which are related to the display coordinates, is VCK
(sub-pixel).

MAGH/MAGV
MAGH and MAGV decide the magnification of width and height respectively. MAGV actually decides the
magnification to the scanning line, and MAGH decides the magnification in units of sub-pixels (VCK).
The values of MAGH to obtain typical resolutions in NTSC/PAL mode are shown below.

Horizontal Resolution (Pixel) MAGH Set Value

256 9
320 7
384 6
512 4
640 3

Other free resolutions can be set according to the value of MAGH.

The relation between the display pixels, DX and MAGH is as follows.

MAGH
VCK

DY
DX

MAGV

Figure 5-4 Relation between Display Pixels, DX and MAGH

Read Format
The format of data read from the frame buffer by the rectangular area read output circuit can be selected
from the following four kinds, depending on the PSM field of the DISPFB register.

PSMCT16/16S
15 10 8 0
31 26 21 16
A B G R
PSMCT24
24 16 8 0
B G R
PSMCT32
24 16 8 0
A B G R
PS-GPU24
24 16 8 0
R1 B0 G0 R0
G2 R2 B1 G1
B3 G3 R3 B2
*Rectangular area read output circuit 2 does not support the PS-GPU24 format.
*Rectangular area setting of the PS-GPU24 should be the same as that of the PSMCT16.

Interlace Read
In interlace mode, the reading method in the vertical direction differs depending on the value of the FFMD
field in the SMODE 2 register as follows.

0: FIELD Mode Read every other line from the start (+0,+2,+4…/+1,+3,+5...)
1: FRAME Mode Read every line from the start (+0,+1,+2,+3,+4,+5….)

5.2.3. Merge Circuit

The merge circuit synthesizes two image outputs of rectangular area read output circuits 1/2 and BGColor
(background color) and obtains two different output images, as shown in Figure 5-5.
MMOD AMOD
ALP

M α
Merge Circuit U
Rectangular X CRT 1
out1 D (Analog)
Area xα
Output + A
NO.1 C
x (1 -α)

enable Alpha-Blending Circuit

Rectangular out2
Area M
Output U
NO.2 M
X U
BGColor X

SLBG Rectangular Area Write Circuit

Figure 5-5 PCRTC Merge Circuit

OUT 2 Output
The output image of rectangular area read circuit 2 is output to OUT2. The BGColor register value is output
outside the rectangular area or when rectangular area read circuit 2 is in the enable-off state.
The alpha value from rectangular area read circuit 2 is output within the rectangular area and 0x00 is output
outside the area.

OUT 1 Output
An alpha-blended image of the output image of rectangular area read circuit 1 and OUT2 or BGColor is
output to OUT1. OUT2 and BGColor is selected by the value of the SLBG field of the PMODE register.

SLBG=0: OUT2
SLBG=1: BGColor

The alpha used for blending is selected according to the value of the MMOD field of the PMODE register,
as follows.

MMOD=0: Alpha value from rectangular area read circuit 1*

MMOD=1: Value of ALP field in PMODE register
* When the alpha value from rectangular area read circuit 1 is selected, the alpha in the frame buffer is set
based on 0x80 =1.0. The value doubled first and then clamped with 0xff is used for blending.

The alpha output to OUT1 is selected according to the value of the AMOD field of the PMODE register, as
follows.

AMOD=0: Alpha value from rectangular area read circuit 1

AMOD=1: Alpha value from rectangular area read circuit 2

Assuming that the output of rectangular area read circuit 1 is Rs, Gs, and Bs, the output of rectangular area
read circuit 2/BGColor is Rd, Gd, and Bd, and the alpha value is A, the output (Ro,Go,Bo) of OUT1 is
calculated as follows:

Ro = Rs x A + Rd x (0xff – A)
Go = Rs x A + Gd x (0xff – A)
Bo = Rs x A + Bd x (0xff – A)
The above results are multiplied by 256(0x100) / 255(0xff).

Data Format and Alpha Value

The Alpha value is handled as follows, based on the format of the data read from the frame buffer.
PSMCT32: 8-bit alpha value in 32-bit color is used.
PSMCT24: 0x80 is used.
PSMCT16: 0x80 is used if 1-bit alpha value in 16-bit color is 1.
0x00 is used if 1-bit alpha value in 16-bit color is 0.
PS-GPU24: 0x80 is used.

Color Processing in PSMCT16(S) Mode

Since only 5 bits each are given to color data R, G and B in the PSMCT16(S) mode, the color is expanded to
8 bits.
bit4 bit3 bit2 bit1 bit0
Input Luminance
E D C B A
Value (5bits)

Output Luminance
Value (8bits) E D C B A 0 0 0
bit7 bit6 bit5 bit4 bit3 bit2 bit1 bit0

5.3. Feedback Write

5.3.1. Rectangular Area Write Input Circuit
This circuit can write a specified rectangular area in the output feedback image to an arbitrary position in a local
buffer at an arbitrary sampling rate, as shown in the figure below.
The writing format is fixed to RGBA 32-bit (PSMCT32).
When inputting data, RGB→YCbCr conversion can be made.

Local Memory

SX Feedback Input Image

EXBP
Frame Buffer SY (WW+1)*SMPH
WDX

WDY WW+1
(WH+1)
*SMPV

WH+1 1 / SMPH
1 / SMPV

EXBW

Captured/SMPH
SX
SY

Captured/SMPV

Figure 5-6 Write from PCRTC to Frame Buffer

The rectangular area write input circuit can set the parameters shown in the figure above by means of the
EXTBUF register (EXBP, EXBW, WDX, WDY, FBIN, WFFMD, EMODA, EMODC) and the EXTDATA
register (SX, SY, WH, WW, SMPH, SMPV).

The unit of horizontal coefficients SX and SMPH, which relate to the input image, is VCK (sub-pixel).

Selection of Input Image

The input image of the rectangular area write input circuit is selected according to the value of the FBIN
field in the EXTBUF register, as follows:

FBIN=0: OUT1 (Synthetic Image of Rectangular Area Read Circuits 1 and 2)

FBIN=1: OUT2 (Image of Rectangular Area Read Circuit 2)

Processing at Data Write

The input Alpha data and color data perform the following processing according to the value of the
EMODA/EMODC field in the EXTBUF register.

Alpha Value Write Mode Setting (EMODA)

EMODA =0: A (Input Alpha Value)
EMODA =1: Y (Conversion Value of Input RGB)
EMODA =2: Y/2 (Conversion Value of Input RGB)
EMODA =3: 0

Color Value Write Mode Setting (EMODC)

EMODC =0: (R,G,B) (Direct Input of Input RGB)
EMODC =1: (Y,Y,Y) (Conversion Value of Input RGB: Y for All RGB)
EMODC =2: (Y,Cb,Cr) (YCbCr Conversion Value of Input RGB)
EMODC =3: (A, A, A) (Input Alpha Value: Alpha for All RGB)

Calculation of RGB->YCbCr conversion is as below:

Ey = 0.587 x Eg + 0.114 x Eb + 0.299 x Er
Ecb = -0.311 x Eg + 0.500 x Eb - 0.169 x Er
Ecr = -0.419 x Eg - 0.081 x Eb + 0.500 x Er

Er,Eg,Eb = R,G,B(0~0xff) / 255(0xff)

Y = (DBh) x Ey + (0x10)
Cr = (E0h) x Ecb + (0x80)
Cb = (E0h) x Ecr + (0x80)

Write Processing in Interlace Mode

For the input image of the rectangular area write input circuit, the writing method is decided depending on
the value of the WFFMD field of the EXTBUF register as follows.

WFFMD =0:
FIELD: Write to every other line from the start (+0,+2,+4.. / +1,+3,+5..)
WFFMD =1:
FRAME: Write to every line from the start (+0,+1,+2,+3,+4,+5….)

Write Command
The register EXTWRITE to execute write commands is provided in addition to the registers to set addresses
and coordinates. Write processing of the feedback image is begun by writing 1 to this register and ended by
writing 0.
If write processing of the specified rectangular area ends, an interrupt (EXTWINT field in the CSR register)
occurs.
When the system is reset and FLUSH is caused during write processing, write processing is suspended.

(This page is left blank intentionally)

6. Signal

6.1. Outline of Signal Processing

The GS can generate a signal (an interrupt) to the host, by writing data to the signal register. By using this
function in the drawing command string, it is possible to synchronize with the progress of drawing in the host.

The GS has three kinds of signal registers. Although it asserts interrupt signals to the host when signals are
generated, it can control them individually with the Interrupt Mask Register, IMR.

6.2. Signal Register

The three kinds of signal registers of the GS are described below.

6.2.1. SIGNAL Register

If data is written to the SIGNAL register, an exception is generated immediately regardless of the ongoing
drawing process. The value of the SIGID field of the SIGLBLID register is updated, and the value of the
SIGNAL flag of the CSR register becomes 1. At this time, if the SIGMSK flag of the IMR register is 0, an
interrupt is generated to the host.

The interrupt state is cleared when the SIGNAL flag of the CSR register is cleared.

When the second write operation is performed to the SIGNAL register without clearing the interrupt state, an
interrupt is not generated to the host. The GS stops the drawing process and will not honor write operations to
the general-purpose registers. The GS will resume the drawing process and will honor write operations to the
general-purpose registers when the SIGNAL flag of the CSR register is cleared. However, the interrupt state
remains not cleared. At this time, the interrupt corresponding to the second write operation to the SIGNAL
register occurs by setting the SIGMSK flag of the IMR register to 1 and then to 0. The interrupt state is cleared
by clearing the SIGNAL flag of the CSR register again.

6.2.2. FINISH Register

If data is written to the FINISH register, an exception is generated after all the ongoing drawing and host ->
local transmission processes, and the value of the FINISH flag of the CSR register becomes0. At this time, if
the FINISHMSK flag of the IMR register is 0, an interrupt is generated to the host. By adding a write operation
to the FINISH register at the end of the drawing command, the host can check the end of drawing.

Unlike the SIGNAL register, the FINISH register does not stop the GS from drawing when it is written
successively.
6.2.3. LABEL Register
If data is written to the LABEL register, the value of the LABELID field of the SIGLBLID register is updated.
An interrupt is not generated.

The drawing process does not stop by writing to the LABEL register.

(This page is left blank intentionally)

7. Registers

This chapter describes the functions of the registers, which are accessible from the host, and the meanings of the
fields in the registers.

7.1. General Purpose Registers

The general purpose registers are mainly used to set vertex information, drawing environment and transmission
between buffers for the drawing primitives. All of them are write-only registers.

Among the registers that set the drawing environment, the registers with names in the format "XXX_1,
XXX_2" are designed to have two contexts. They consist of two sets of registers with the same functions.
Registers with two contexts are as shown below:

FRAME (FRAME_1, FRAME_2)

ZBUF (ZBUF_1, ZBUF_2)
TEX0 (TEX0_1, TEX0_2)
TEX1 (TEX1_1, TEX1_2)
TEX2 (TEX2_1, TEX2_2)
MIPTBP1 (MIPTBP1_1, MIPTBP1_2)
MIPTBP2 (MIPTBP2_1, MIPTBP2_2)
CLAMP (CLAMP_1, CLAMP_2)
TEST (TEST_1, TEST_2)
ALPHA (ALPHA_1, ALPHA_2)
XYOFFSET (XYOFFSET_1, XYOFFSET_2)
SCISSOR (SCISSOR_1, SCISSOR_2)
FBA (FBA_1, FBA_2)

Explanatory Notes

Register Name Outline Address

FOGCOL : Fog Color

0x3d
This register sets fogging ........ Details of
Function

BIT ASSIGN
6 2 1 1 0 0 0
3 3 6 5 8 7 0

FCB FCG FCR

FIELD
Name Pos. Format Contents
FCR 7:0 int0:8:0 Fog Color (RED)
FCG 15:8 int0:8:0 Fog Color (GREEN)
FCB 23:16 int0:8:0 Fog Color (BLUE)

Field Bit Position Field Data Field Name in

Name in the Field Format the Register

Notation of the field bit position

63:48 Bit 63 to bit 48
15 Bit 15 only

Notation of the field data format

float 32 IEEE754-based 32-bit single-precision floating-point value.
int *:*:* Integer or fixed decimal value.
"*" indicates number of sign bits, number of integer bits, and number of decimal bits, left to
right.
(Notation example of "int")
int 1:11:4 16-bit fixed decimal value with 1 bit for sign, 11 bits for integer part and 4 bits for decimal
part.
int 0:8:0 8-bit integer with no sign, 8 bits for integer part and 0 bit for decimal part.
int 0:2:0 Bit field of 2-bit width with no sign.

ALPHA_1 / ALPHA_2 : Alpha Blending Setting

0x42 / 0x43
These registers define the blend function of Alpha Blending. ALPHA_1 sets Context 1 and ALPHA_2 sets
Context 2.

The basic form of the blending function is as follows:

Cv = (A-B)*C>>7 + D
(Cv = Output Color Value)
(A,B,D = Input Color Value)
(C = Input Alpha Value)
Each item of A/B/C/D in the function above is set by this register.

BIT ASSIGN
3 3 0 0 0 0 0 0 0 0
9 2 7 6 5 4 3 2 1 0

FIX D C B A

FIELD
Name Pos. Format Contents
A 1:0 int 0:2:0 Specification of Input Color Value A
00 Cs RGB value of the source is used.
01 Cd RGB value in the frame buffer is used.
10 0
11 Reserved
B 3:2 int 0:2:0 Specification of Input Color Value B
00 Cs RGB value of the source is used.
01 Cd RGB value in the frame buffer is used.
10 0
11 Reserved
C 5:4 int 0:2:0 Specification of Input Alpha Value C
00 As Alpha of the source is used.
01 Ad Alpha in the frame buffer is used.
10 FIX FIX-field value is used as Alpha.
11 Reserved
D 7:6 int 0:2:0 Specification of Input Color Value D
00 Cs RGB value of the source is used.
01 Cd RGB value in the frame buffer is used.
10 0
11 Reserved
FIX 39:32 int 0:8:0 Fixed Alpha Value
Referred to when the value of C above is FIX.
(The range is 0 - 255, and at the 128 position, 1.0 is indicated.)

BITBLTBUF : Setting for Transmission between Buffers

0x50
This register stores buffer-related settings for transmission source and destination during transmission between
buffers.

BIT ASSIGN
6 5 5 4 4 3 2 2 2 1 1 0
1 6 3 8 5 2 9 4 1 6 3 0

DPSM DBW DBP SPSM SBW SBP

FIELD
Name Pos. Format Contents
SBP 13:0 int 0:14:0 Source Buffer Base Pointer (SBP = Word Address/64)
SBW 21:16 int 0:6:0 Source Buffer Width (SBW = Width in Units of Pixels/64) (*1)
SPSM 29:24 int 0:6:0 Source Pixel Storage Format (*2, *3)
000000 PSMCT32
000001 PSMCT24
000010 PSMCT16
001010 PCMCT16S
010011 PSMT8
010100 PSMT4
011011 PSMT8H
100100 PSMT4HL
101100 PSMT4HH
110000 PSMZ32
110001 PSMZ24
110010 PSMZ16
111010 PSMZ16S
DBP 45:32 int 0:14:0 Destination Buffer Base Pointer (DBP = Word Address/64)
DBW 53:48 int 0:6:0 Destination Buffer Width (DBW = Width in Units of Pixels/64) (*1)
DPSM 61:56 int 0:6:0 Destination Pixel Storage Format (*3)
(Same as Source Pixel Storage Format)
*1 Values must be in the range of 1 to 32.
*2 Operation is undefined when values other than those in the table are set.
*3 In local-local transmission, the formats specified in the SPSM and DPSM fields must have the same number
of bits per pixel (pixel depth).

CLAMP_1 / CLAMP_2 : Texture Wrap Mode

0x08 / 0x09
These registers set the texture's wrap mode (repeating or clamping) for both S and T. CLAMP_1 sets Context 1
and CLAMP_2 sets Context 2.

BIT ASSIGN
4 3 3 2 2 1 1 0 0 0 0 0
3 4 3 4 3 4 3 4 3 2 1 0
W W
MAXV MINV MAXU MINU M M
T S

FIELD
Name Pos. Format Contents
WMS 1:0 int 0:2:0 Wrap Mode in Horizontal (S) Direction
00 REPEAT
01 CLAMP
10 REGION_CLAMP
11 REGION_REPEAT
WMT 3:2 int 0:2:0 Wrap Mode in Vertical (T) Direction
00 REPEAT
01 CLAMP
10 REGION_CLAMP
11 REGION_REPEAT
MINU 13:4 int 0:10:0 Clamp Parameter in U Direction (*1)
(In REGION_CLAMP Mode)
Clamp Value of Lower Limit
(In REGION_REPEAT Mode)
UMSK Value
MAXU 23:14 int 0:10:0 Clamp Parameter in U Direction (*1)
(In REGION_CLAMP Mode)
Clamp Value of Upper Limit
(In REGION_REPEAT Mode)
UFIX Value
MINV 33:24 int 0:10:0 Clamp Parameter in V Direction (*2)
(In REGION_CLAMP Mode)
Clamp Value of Lower Limit
(In REGION_REPEAT Mode)
VMSK Value
MAXV 43:34 int 0:10:0 Clamp Parameter in V Direction (*2)
(In REGION_CLAMP Mode)
Clamp Value of Upper Limit
(In REGION_REPEAT Mode)
VFIX Value
*1 Fields MINU and MAXU are not used when the value of WMS, the wrap mode in horizontal direction, is
REPEAT or CLAMP. (Don't Care.)
*2 Fields MINV and MAXV are not used when the value of WMT, the wrap mode in vertical direction, is
REPEAT or CLAMP. (Don't Care.)

COLCLAMP : Color Clamp Control

0x46
This register stores settings as to whether clamping for the RGB value of the pixel is performed.

BIT ASSIGN
0
0
C
L
A
M
P

FIELD
Name Pos. Format Contents
CLAMP 0 int 0:1:0 Color Clamping Method
0 MASK Lower 8 bits are enabled(wraps around.)
1 CLAMP Clamped in [0,255] range.

DIMX : Dither Matrix Setting

0x44
This register sets the value of the dither matrix.

BIT ASSIGN
6 6 5 5 5 5 5 4 4 4 4 4 3 3 3 3
2 0 8 6 4 2 0 8 6 4 2 0 8 6 4 2

DM DM DM DM DM DM DM DM
33 32 31 30 23 22 21 20

3 2 2 2 2 2 1 1 1 1 1 0 0 0 0 0
0 8 6 4 2 0 8 6 4 2 0 8 6 4 2 0

DM DM DM DM DM DM DM DM
13 12 11 10 03 02 01 00

FIELD
Name Pos. Format Contents
DM00 2:0 int 1:2:0 Dither Matrix (0,0)
DM01 6:4 int 1:2:0 Dither Matrix (0,1)
DM02 10:8 int 1:2:0 Dither Matrix (0,2)
DM03 14:12 int 1:2:0 Dither Matrix (0,3)
DM10 18:16 int 1:2:0 Dither Matrix (1,0)
DM11 22:20 int 1:2:0 Dither Matrix (1,1)
DM12 26:24 int 1:2:0 Dither Matrix (1,2)
DM13 30:28 int 1:2:0 Dither Matrix (1,3)
DM20 34:32 int 1:2:0 Dither Matrix (2,0)
DM21 38:36 int 1:2:0 Dither Matrix (2,1)
DM22 42:40 int 1:2:0 Dither Matrix (2,2)
DM23 46:44 int 1:2:0 Dither Matrix (2,3)
DM30 50:48 int 1:2:0 Dither Matrix (3,0)
DM31 54:52 int 1:2:0 Dither Matrix (3,1)
DM32 58:56 int 1:2:0 Dither Matrix (3,2)
DM33 62:60 int 1:2:0 Dither Matrix (3,3)

DTHE : Dither Control

0x45
This register stores settings for dithering (performed/not performed).
If the pixel storage mode of the frame buffer is PSMCT32 or PSMCT24, dithering should be turned off.

BIT ASSIGN
0
0
D
T
H
E

FIELD
Name Pos. Format Contents
DTHE 0 int 0:1:0 Dithering Control
0 Dithering is not performed.
1 Dithering is performed.

FBA_1 / FBA_2 : Alpha Correction Value

0x4a / 0x4b
These registers set the fixed value of Alpha when writing to the frame buffer. FBA_1 sets Context 1 and FBA_2
sets Context 2.

The OR of the fixed value set by this register and the most significant bit of the Alpha value of the pixel is
written to the frame buffer.

BIT ASSIGN
0
0
F
B
A

FIELD
Name Pos. Format Contents
FBA 0 int 0:1:0 MSB of Alpha value for drawing to the frame buffer.
RGBA32 Mode: A = As | (FBA<<7)
RGBA16 Mode: A = As | FBA&0x01

FINISH : FINISH Event Occurrence Request

0x61
When the drawing process currently being performed is completed, this register makes a request for the
occurrence of the FINISH event. Any data can be written.

BIT ASSIGN

FOG : Vertex Fog Value Setting

0x0a
This register sets the fog value of the vertex. When the fog value is 0, the fog effect is at the maximum (the
drawing color becomes the same as the distant color). When the fog value is 255, the fog effect is at the
minimum.
To completely eliminate the fog effect, fogging should be set off with the FGE flag of the PRIM register.

BIT ASSIGN
6 5
3 6

FIELD
Name Pos. Format Contents
F 63:56 int0:8:0 Fog Value

FOGCOL : Distant Fog Color Setting

0x3d
This register sets the distant color with RGB value when performing fogging.

BIT ASSIGN
2 1 1 0 0 0
3 6 5 8 7 0

FCB FCG FCR

FIELD
Name Pos. Format Contents
FCR 7:0 int 0:8:0 Fog Color (R)
FCG 15:8 int 0:8:0 Fog Color (G)
FCB 23:16 int 0:8:0 Fog Color (B)

FRAME_1 / FRAME_2 : Frame Buffer Setting

0x4c / 0x4d
These registers store various settings related to the frame buffer. FRAME_1 sets Context 1 and FRAME_2 sets
Context 2.

BIT ASSIGN
6 3 2 2 2 1 0 0
3 2 9 4 1 6 8 0

FBMSK PSM FBW FBP

FIELD
Name Pos. Format Contents
FBP 8:0 int0:9:0 Frame Buffer Base Pointer
FBP = Word Address/2048
FBW 21:16 int0:6:0 Frame Buffer Width
FBW = Width in Units of Pixels/64
The effective range of set value is 1 to 32
PSM 29:24 int0:6:0 Frame Buffer Pixel Storage Format
000000 PSMCT32
000001 PSMCT24
000010 PSMCT16
001010 PSMCT16S
010011 reserved
010100 reserved
011011 reserved
100100 reserved
101100 reserved
110000 PSMZ32
110001 PSMZ24
110010 PSMZ16
111010 PSMZ16S
FBMSK 63:32 int0:32:0 Frame Buffer Drawing Mask (*)
0 The corresponding bit in the frame buffer is updated.
1 The corresponding bit in the frame buffer is not updated.
* See "3.9.5. Masking". When the pixel storage format of the frame buffer is PSMCT16, PSMCT16S or
PSMZ16, note the bit relation between the mask pattern and frame buffer.

HWREG : Data Port for Transmission between Buffers

0x54
This register writes transmission data when host -> local transmission between buffers is performed.
It is enabled only when the BITBLTBUF/TRXPOS/TRXREG/TRXDIR registers are set effectively in terms
of transmission, and the transmission between buffers is activated. When this register is written in other
conditions, no operation is made.

BIT ASSIGN
6 0
3 0

DATA

FIELD
Name Pos. Format Contents
DATA 63:0 int0:64:0 Host -> Local Transmission Data

LABEL : LABEL Event Occurrence Request

0x62
This register updates the value of LBLID field in the SIGLBLID register. The ID field is composed of 32 bits,
and SIGLBLID is updated only in the bit position where the bit value corresponding to IDMSK is 1.

BIT ASSIGN
6 3 3 0
3 2 1 0

IDMSK ID

FIELD
Name Pos. Format Contents
ID 31:0 int0:32:0 Value to be written to SIGLBLID register
IDMSK 63:32 int0:32:0 Whether or not corresponding SIGLBLID bit is updated.
0 Masked. (Not updated.)
1 Not masked. (Updated.)

MIPTBP1_1 / MIPTBP1_2 : MIPMAP Information Setting (Level 1 to 3)

0x34 / 0x35
These registers set the buffer pointer and buffer width of Level 1 to Level 3 textures when performing
MIPMAP. MIPTBP1_1 sets Context 1 and MIPTBP1_2 sets Context 2.

BIT ASSIGN
5 5 5 4 3 3 3 2 1 1 1 0
9 4 3 0 9 4 3 0 9 4 3 0

TBW3 TBP3 TBW2 TBP2 TBW1 TBP1

FIELD
Name Pos. Format Contents
TBP1 13:0 int0:14:0 MIPMAP Level 1 Texture Base Pointer (*1)
TBW1 19:14 int0:6:0 MIPMAP Level 1 Texture Buffer Width (*2)
TBP2 33:20 int0:14:0 MIPMAP Level 2 Texture Base Pointer (*1)
TBW2 39:34 int0:6:0 MIPMAP Level 2 Texture Buffer Width (*2)
TBP3 53:40 int0:14:0 MIPMAP Level 3 Texture Base Pointer (*1)
TBW3 59:54 int0:6:0 MIPMAP Level 3 Texture Buffer Width (*2)
*1 The texture base pointer is set to the value of the word address divided by 64.
*2 The buffer width is set to the value of the width (in units of texels) divided by 64. It is set to 1 if the width is
less than 64.

MIPTBP2_1 / MIPTBP2_2 : MIPMAP Information Setting (Level 4 to 6)

0x36 / 0x37
These registers set the buffer pointer and buffer width of Level 4 to Level 6 textures when performing
MIPMAP. MIPTBP2_1 sets Context 1 and MIPTBP2_2 sets Context 2.

BIT ASSIGN
5 5 5 4 3 3 3 2 1 1 1 0
9 4 3 0 9 4 3 0 9 4 3 0

TBW6 TBP6 TBW5 TBP5 TBW4 TBP4

FIELD
Name Pos. Format Contents
TBP4 13:0 int0:14:0 MIPMAP Level 4 Texture Base Pointer (*1)
TBW4 19:14 int0:6:0 MIPMAP Level 4 Texture Buffer Width (*2)
TBP5 33:20 int0:14:0 MIPMAP Level 5 Texture Base Pointer (*1)
TBW5 39:34 int0:6:0 MIPMAP Level 5 Texture Buffer Width (*2)
TBP6 53:40 int0:14:0 MIPMAP Level 6 Texture Base Pointer (*1)
TBW6 59:54 int0:6:0 MIPMAP Level 6 Texture Buffer Width (*2)
*1 The texture base pointer is set to the value of the word address divided by 64.
*2 The buffer width is set to the value of the width (in units of texels) divided by 64. It is set to 1 if the width is
less than 64.

PABE : Alpha Blending Control in Units of Pixels

0x49
This register sets whether Alpha Blending control in units of pixels is performed.
For details, see "3.8. Alpha Blending".

BIT ASSIGN
0
0
P
A
B
E

FIELD
Name Pos. Format Contents
PABE 0 int0:1:0 Alpha Blending Control in Units of Pixels
0 Not performed.
1 Performed. (Alpha blending is OFF for the pixel where the A value
is MSB=0 and ON for the pixel where the A value is MSB=1.)

PRIM : Drawing Primitive Setting

0x00
This register specifies the types of drawing primitives and various attributes, and initializes the contents of the
vertex queue.
Each of the IIP/TME/FGE/ABE/AA1/FST/CTXT/FIX fields is enabled only when AC=1 is set by the
PRMODECONT register.

BIT ASSIGN
1 0 0 0 0 0 0 0 0 0
0 9 8 7 6 5 4 3 2 0
C
F F A A F T I
T
I S A B G M I PRIM
X
X T 1 E E E P
T

FIELD
Name Pos. Format Contents
PRIM 2:0 int0:3:0 Types of Drawing Primitives
000 Point
001 Line
010 Line Strip
011 Triangle
100 Triangle Strip
101 Triangle Fan
110 Sprite
111 Specification Prohibited
IIP 3 int0:1:0 Shading Method
0 Flat Shading
1 Gouraud Shading
TME 4 int0:1:0 Texture Mapping
0 Texture Mapping OFF
1 Texture Mapping ON
FGE 5 int0:1:0 Fogging
0 Fogging OFF
1 Fogging ON
ABE 6 int0:1:0 Alpha Blending
0 Alpha Blending OFF
1 Alpha Blending ON
AA1 7 int0:1:0 1 Pass Antialiasing (*1)
0 1 Pass Antialiasing OFF
1 1 Pass Antialiasing ON
FST 8 int0:1:0 Method of Specifying Texture Coordinates (*2)
0 STQ value is used (ST/RGBAQ Register is referred
to.)
1 UV value is used (UV Register is referred to.)
CTXT 9 int0:1:0 Context
0 Environment of Context 1 is used.
1 Environment of Context 2 is used.
FIX 10 int0:1:0 Fragment Value Control (RGBAFSTQ Change by DDA)
0 Unfixed (Normal)
1 Fixed
*1 When AA1=1 is set, note Alpha Blending process. See "3.6. Antialiasing".
*2 When FST=1 is set, perspective correction is not performed.

PRMODE : Setting for Attributes of Drawing Primitives

0x1b
This register changes the attributes of drawing primitives. It is enabled only when AC=0 is set by the
PRMODECONT register.

BIT ASSIGN
1 0 0 0 0 0 0 0
0 9 8 7 6 5 4 3
C
F F A A F T I
T
I S A B G M I
X
X T 1 E E E P
T

FIELD
Name Pos. Format Contents
IIP 3 int0:1:0 Shading Method
0 Flat Shading
1 Gouraud Shading
TME 4 int0:1:0 Texture Mapping
0 Texture Mapping OFF
1 Texture Mapping ON
FGE 5 int0:1:0 Fogging
0 Fogging OFF
1 Fogging ON
ABE 6 int0:1:0 Alpha Blending
0 Alpha Blending OFF
1 Alpha Blending ON
AA1 7 int0:1:0 1 Pass Antialiasing (*1)
0 1 Pass Antialiasing OFF
1 1 Pass Antialiasing ON
FST 8 int0:1:0 Method of Specifying Texture Coordinates (*2)
0 STQ value (ST/RGBAQ register is referred to.)
1 UV value (UV register is referred to.)
CTXT 9 int0:1:0 Context
0 Environment of Context 1 is used.
1 Environment of Context 2 is used.
FIX 10 int0:1:0 Fragment Value Control (RGBAFSTQ Change by DDA)
0 Unfixed (Normal)
1 Fixed
*1 When AA1=1 is set, note Alpha Blending process. See "3.6. Antialiasing".
*2 When FST=1 is set, perspective correction is not performed.

PRMODECONT : Specification of Primitive Attribute Setting Method

0x1a
This register sets whether to use primitive attributes (IIP, TME, FGE, ABE, AA1, FST, CTXT, FIX) specified
by the PRMODE register or the PRIM register.

BIT ASSIGN
0
0

A
C

FIELD
Name Pos. Format Contents
AC 0 int0:1:0 Register that specifies primitive attributes
0 PRMODE register
1 PRIM register

RGBAQ : Vertex Color Setting

0x01
This register sets the RGBA value of the vertex and the Q value of the normalized texture coordinates.
The Q value is not only used for calculating texture coordinates but also as the parameter to decide the LOD.

BIT ASSIGN
6 3 3 2 2 1 1 0 0 0
3 2 1 4 3 6 5 8 7 0

Q A B G R

FIELD
Name Pos. Format Contents
R 7:0 int 0:8:0 Luminance value of R element of vertex color
G 15:8 int 0:8:0 Luminance value of G element of vertex color
B 23:16 int 0:8:0 Luminance value of B element of vertex color
A 31:24 int 0:8:0 Alpha value of vertex (0x80 = 1.0)
Q 63:32 float 32 Normalized texture coordinates (Q) (*)
* Negative values are acceptable. However, the vertices of one primitive must be the same. The lower 8 bits of
the mantissa are rounded down.

SCANMSK : Raster Address Mask Setting

0x22
This register sets whether drawing is performed only to odd or even row by the row address of the frame buffer.
When drawing is prohibited, drawing efficiency decreases (1/2 in the worst case.)
The mask is only effective for drawing primitives.

BIT ASSIGN
0 0
1 0
M
S
K

FIELD
Name Pos. Format Contents
MSK 1:0 int0:2:0 00 Normal drawing (not masked.)
01 Reserved.
10 Drawing of pixel with even Y coordinate is prohibited.
11 Drawing of pixel with odd Y coordinate is prohibited.

SCISSOR_1 / SCISSOR_2 : Setting for Scissoring Area

0x40 / 0x41
These registers specify the scissoring area. The coordinate values for the upper-left/lower-right points of the
enabled drawing area are specified by the window coordinate system. SCISSOR_1 sets Context 1 and
SCISSOR_2 sets Context 2.

BIT ASSIGN
5 4 4 3 2 1 1 0
8 8 2 2 6 6 0 0

SCAY1 SCAY0 SCAX1 SCAX0

FIELD
Name Pos. Format Contents
SCAX0 10:0 int 0:11:0 X-coordinate value for upper-left point of enabled drawing area
(Window coordinate system)
SCAX1 26:16 int 0:11:0 X-coordinate value for lower-right point of enabled drawing area
(Window coordinate system)
SCAY0 42:32 int 0:11:0 Y-coordinate value for upper-left point of enabled drawing area
(Window coordinate system)
SCAY1 58:48 int 0:11:0 Y-coordinate value for lower-right point of enabled drawing area
(Window coordinate system)

SIGNAL : SIGNAL Event Occurrence Request

0x60
This register generates the SIGNAL event and updates the value of the SIGLBLID register. The ID field is
composed of 32 bits, and SIGLBLID is updated only in the bit position where the bit value corresponding to
IDMSK is 1.

BIT ASSIGN
6 3 3 0
3 2 1 0

IDMSK ID

ST : Specification of Vertex Texture Coordinates

0x02
This register sets the S and T values of the vertex texture coordinates. The value Q is specified by the RGBAQ
register.
For the relation between the texture coordinate values S, T and Q and the texel, see "3.4.10. Perspective
Correction".

BIT ASSIGN
6 3 3 0
3 2 1 0

T S

FIELD
Name Pos. Format Contents
S 31:0 float 32 Texture Coordinate Value S (The lower 8 bits of the mantissa are
rounded down.)
T 63:32 float 32 Texture Coordinate Value T (The lower 8 bits of the mantissa are
rounded down.)

TEST_1 / TEST_2 : Pixel Test Control

0x47 / 0x48
These registers perform settings related to the pixel test. TEST_1 sets Context 1 and TEST_2 sets Context 2.
For details, see "3.7. Pixel Test".

BIT ASSIGN
1 1 1 1 1 1 1 1 0 0 0 0
8 7 6 5 4 3 2 1 4 3 1 0
Z D D A A
Z F A
T A A T
T A AREF T
S T T I S
E E
T M E L T

FIELD
Name Pos. Format Contents
ATE 0 int 0:1:0 Alpha Test
0 Alpha test OFF
1 Alpha test ON
ATST 3:1 int 0:3:0 Alpha Test Method
000 NEVER All pixels fail.
001 ALWAYS All pixels pass.
010 LESS Pixels with A less than AREF pass.
011 LEQUAL Pixels with A less than or equal to
AREF pass.
100 EQUAL Pixels with A equal to AREF pass.
101 GEQUAL Pixels with A greater than or equal to
AREF pass.
110 GREATER Pixels with A greater than AREF pass.
111 NOTEQUAL Pixels with A not equal to AREF pass.
AREF 11:4 int 0:8:0 Alpha Value to be Compared and Referred to
AFAIL 13:12 int 0:2:0 Processing Method when Failed in Alpha Test
00 KEEP Neither frame buffer nor Z buffer is updated.
01 FB_ONLY Only frame buffer is updated.
10 ZB_ONLY Only Z buffer is updated.
11 RGB_ONLY Only frame-buffer RGB is updated.
DATE 14 int 0:1:0 Destination Alpha Test
0 Destination Alpha Test OFF
1 Destination Alpha Test ON
DATM 15 int 0:1:0 Destination Alpha Test Mode
0 Pixels with destination alpha equal to 0 pass.
1 Pixels with destination alpha equal to 1 pass.
ZTE 16 int 0:1:0 Depth Test
0 Depth test OFF (not allowed)
1 Depth test ON
ZTST 18:17 int 0:2:0 Depth Test Method
00 NEVER All pixels fail.
01 ALWAYS All pixels pass.
10 GEQUAL Pixels with Z greater than or equal to
Z buffer value pass.
11 GREATER Pixels with Z greater than Z buffer
value pass.

TEX0_1 / TEX0_2 : Texture Information Setting

0x06 / 0x07
These registers set various kinds of information regarding the textures to be used. TEX0_1 sets Context 1 and
TEX0_2 sets Context 2.

BIT ASSIGN
6 6 6 5 5 5 5 5 3 3 3 3 3 3 2 2 2 2 1 1 1 0
3 1 0 6 5 4 1 0 7 6 5 4 3 0 9 6 5 0 9 4 3 0
C
C T T
P
CLD CSA S CBP F C TH TW PSM TBW TBP0
S
M X C
M

FIELD
Name Pos. Format Contents
TBP0 13:0 int0:14:0 Texture Base Pointer
TBP0 = Word Address/64
TBW 19:14 int0:6:0 Texture Buffer Width
TBW0 = Width in Units of Texels/64
PSM 25:20 int0:6:0 Texture Pixel Storage Format
000000 PSMCT32
000001 PSMCT24
000010 PSMCT16
001010 PSMCT16S
010011 PSMT8
010100 PSMT4
011011 PSMT8H
100100 PSMT4HL
101100 PSMT4HH
110000 PSMZ32
110001 PSMZ24
110010 PSMZ16
111010 PSMZ16S
TW 29:26 int0:4:0 Texture Width: Width = 2TW (max 210) (*1)
TH 33:30 int0:4:0 Texture Height: Height = 2TH (max 210) (*1)
TCC 34 int0:1:0 Texture Color Component
0 RGB
1 RGBA (TEXA register value is At in RGB24/RGBA16)
TFX 36:35 int0:2:0 Texture Function
00 MODULATE
01 DECAL
10 HIGHLIGHT
11 HIGHLIGHT2
CBP 50:37 int0:14:0 CLUT Buffer Base Pointer
CBP = Word Address/64
CPSM 54:51 int0:4:0 CLUT Pixel Storage Format
0000 PSMCT32
0010 PSMCT16
1010 PSMCT16S
CSM 55 int0:1:0 CLUT Storage Mode
0 CSM1
1 CSM2

Name Pos. Format Contents

CSA 60:56 int0:5:0 CLUT Entry Offset
CSA = Offset/16
In CSM2 mode, CSA=0 must be set.
CLD 63:61 int0:3:0 CLUT Buffer Load Control
000 Temporary buffer contents are not changed.
001 Load is performed to CSA position of buffer.
010 Load is performed to CSA position of buffer and
CBP is copied to CBP0. (*2)
011 Load is performed to CSA position of buffer and
CBP is copied to CBP1. (*2)
100 If CBP0 !=CBP, load is performed and
CBP is copied to CBP0. (*2)
101 If CBP1 != CBP, load is performed and
CBP is copied to CBP1. (*2)
*1 Texture size must be 8 x 8 texels or more when using bilinear or trilinear sampling.
*2 CBP0 and CBP1 are internal registers of the GS.

TEX1_1 / TEX1_2 : Texture Information Setting

0x14 / 0x15
These registers set information on the sampling method of the textures. TEX1_1 sets Context 1 and TEX1_2
sets Context 2.

BIT ASSIGN
4 3 2 1 0 0 0 0 0 0 0
3 2 0 9 9 8 6 5 4 2 0
M M
L
T M
K L MMIN MXL C
B A
M
A G

FIELD
Name Pos. Format Contents
LCM 0 int0:1:0 LOD Calculation Method
0 Due to the formula (LOD = (log2(1/|Q|)<<L)+K)
1 Fixed value (LOD = K)
MXL 4:2 int0:3:0 Maximum MIP Level (0-6)
MMAG 5 int0:1:0 Filter when Texture is Expanded (LOD < 0)
0 NEAREST
1 LINEAR
MMIN 8:6 int0:3:0 Filter when Texture is Reduced (LOD >= 0)
000 NEAREST
001 LINEAR
010 NEAREST_MIPMAP_NEAREST
011 NEAREST_MIPMAP_LINEAR
100 LINEAR_MIPMAP_NEAREST
101 LINEAR_MIPMAP_LINEAR
MTBA 9 int0:1:0 Base Address Specification of MIPMAP Texture of Level 1 or More
0 Value specified by MIPTBP1 and MIPTBP2 is used.
1 Base address of TBP1 - TBP3 is automatically set.
(See "3.4.11. MIPMAP".)
L 20:19 int0:2:0 LOD Parameter Value L (See the section for LCM.)
K 43:32 int1:7:4 LOD Parameter Value K (See the section for LCM.)

TEX2_1 / TEX2_2 : Texture Information Setting

0x16 / 0x17
These registers set texture information. They are subsets of the TEX0 register. TEX2_1 sets Context 1 and
TEX2_2 sets Context 2.

BIT ASSIGN
6 6 6 5 5 5 5 5 3 2 2
3 1 0 6 5 4 1 0 7 5 0
C
C
P
CLD CSA S CBP PSM
S
M
M

FIELD
Name Pos. FormatContents
PSM 25:20 int0:6:0
Texture Pixel Storage Format
000000 PSMCT32
000001 PSMCT24
000010 PSMCT16
001010 PCMCT16S
010011 PSMT8
010100 PSMT4
011011 PSMT8H
100100 PSMT4HL
101100 PSMT4HH
110000 PSMZ32
110001 PSMZ24
110010 PSMZ16
111010 PSMZ16S
CBP 50:37 int0:14:0 CLUT Buffer Base Pointer
CBP=Word Address/64
CPSM 54:51 int0:4:0 CLUT Entry Storage Format
0000 PSMCT32
0010 PSMCT16
1010 PSMCT16S
CSM 55 int0:1:0 CLUT Storage Mode
0 CSM1
1 CSM2
CSA 60:56 int0:5:0 CLUT Entry Offset
CSA = Offset/16
In CSM2 mode, CSA=0 must be set.
CLD 63:61 int0:3:0 CLUT Buffer Load Control
000 CLUT buffer contents are not changed.
001 Load is performed to CSA position of buffer.
010 Load is performed to CSA position of buffer and
CBP is copied to CBP0. (*)
011 Load is performed to CSA position of buffer and
CBP is copied to CBP1. (*)
100 If CBP0 != CBP, load is performed and
CBP is copied to CBP0. (*)
101 If CBP1 != CBP, load is performed and
CBP is copied to CBP1. (*)
* CBP0 and CBP1 are internal registers of the GS. For details, see "3.4.7. CLUT Buffer Control".

TEXA : Texture Alpha Value Setting

0x3b
This register sets the Alpha value to be referred to when the Alpha value of the texture is not an 8-bit value. See
"3.4.6. Format Conversion".

BIT ASSIGN
3 3 1 0 0
9 2 5 7 0
A
TA1 E TA0
M

FIELD
Name Pos. Format Contents
TA0 7:0 int0:8:0 The "As" value referred to when A field is 0 in RGBA16 format or
when the format is RGB24.
AEM 15 int0:1:0 Method of Expanding Texture Alpha
0 Processed normally even when R=G=B=0.
1 Treated as "Transparent" (A=0) when R=G=B=0.
TA1 39:32 int0:8:0 The "As" value referred to when A field is 1 in RGBA16 format.

TEXCLUT : CLUT Position Specification

0x1c
This register specifies the CLUT position in the buffer when the CLUT storage mode is CSM=1 (CSM2 mode).
It is disabled when CSM=0 (CSM1 mode).

BIT ASSIGN
2 1 1 0 0 0
1 2 1 6 5 0

COV COU CBW

FIELD
Name Pos. Format Contents
CBW 5:0 int 0:6:0 CLUT Buffer Width (CBW = Width in Units of Pixels /64)
COU 11:6 int 0:6:0 CLUT Offset U (COU = Offset in Units of Pixels/16)
COV 21:12 int 0:10:0 CLUT Offset V (COV = Offset in Units of Pixels)

TEXFLUSH : Texture Page Buffer Disabling

0x3f
This register waits for the completion of the current drawing process, and then disables the texture data read to
the texture page buffer. Any data can be written.
A drawing process succeeding the write to this register is started after the texture page buffer is disabled.

BIT ASSIGN

TRXDIR : Activation of Transmission between Buffers

0x53
This register specifies the transmission direction in the transmission between buffers, and activates transmission.
Appropriate settings must be made by the BITBLTBUF/TRXPOS/TRXREG before activating the
transmission.

BIT ASSIGN
0 0
1 0
X
D
I
R

FIELD
Name Pos. Format Contents
XDIR 1:0 int0:2:0 Transmission direction is specified and transmission is started.
00 Host -> Local Transmission
01 Local -> Host Transmission
10 Local -> Local Transmission
11 Transmission is deactivated.

TRXPOS : Specification of Transmission Areas in Buffers

0x51
This register specifies the position and scanning direction of the rectangular area in each buffer where buffer
transmission is performed. For details, see "4.1. Transmission Parameters".

BIT ASSIGN
6 5 5 4 4 3 2 1 1 0
0 9 8 8 2 2 6 6 0 0

DIR DSAY DSAX SSAY SSAX

FIELD
Name Pos. Format Contents
SSAX 10:0 int0:11:0 X Coordinate of Upper Left Point of Source Rectangular Area
SSAY 26:16 int0:11:0 Y Coordinate of Upper Left Point of Source Rectangular Area
DSAX 42:32 int0:11:0 X Coordinate of Upper Left Point in Destination Rectangular Area
DSAY 58:48 int0:11:0 Y Coordinate of Upper Left Point in Destination Rectangular Area
DIR 60:59 int0:2:0 Pixel Transmission Order (Enabled only in Local -> Local
Transmission.)
00 Upper Left -> Lower Right
01 Lower Left -> Upper Right
10 Upper Right -> Lower Left
11 Lower Right -> Upper Left
Note that the set rectangular area wraps around when exceeding the buffer width.

TRXREG : Specification of Transmission Areas in Buffers

0x52
This register specifies the size of the rectangular area, where the transmission between buffers is implemented, in
units of pixels. The pixel mode must be the one set by the BITBLTBUF register.

BIT ASSIGN
4 3 1 0
3 2 1 0

RRH RRW

FIELD
Name Pos. Format Contents
RRW 11:0 int0:12:0 Width of Transmission Area
RRH 43:32 int0:12:0 Height of Transmission Area

UV : Specification of Vertex Texture Coordinates

0x03
This register specifies the texel coordinate (UV) values of the vertex.
It is enabled when FST is set to 1 in the PRIM register. When FST=0, the value of the UV register is ignored.

BIT ASSIGN
2 1 1 0
9 6 3 0

V U

FIELD
Name Pos. Format Contents
U 13:0 int 0:10:4 Texel Coordinate (U)
V 29:16 int 0:10:4 Texel Coordinate (V)

XYOFFSET_1 / XYOFFSET_2 : Offset Value Setting

0x18 / 0x19
These registers set the offset value for converting from the primitive coordinate system to the window
coordinate system. XYOFFSET_1 sets Context 1 and XYOFFSET_2 sets Context 2.

BIT ASSIGN
4 3 1 0
7 2 5 0

OFY OFX

FIELD
Name Pos. Format Contents
OFX 15:0 int 0:12:4 Offset (X)
OFY 47:32 int 0:12:4 Offset (Y)

XYZ2 : Setting for Vertex Coordinate Values

0x05
This register sets the vertex coordinate values and moves the vertex queue a step forward. The coordinate
values X and Y are specified in the primitive coordinate system.
When the vertex information stored in the vertex queue reaches a fixed number, drawing is started. The method
of moving the vertex queue forward and the number of vertices at which drawing starts depend on the type of
the drawing primitive.

BIT ASSIGN
6 3 3 1 1 0
3 2 1 6 5 0

Z Y X

FIELD
Name Pos. Format Contents
X 15:0 int 0:12:4 Vertex Coordinate Value X (0 – 4095.9375)
Y 31:16 int 0:12:4 Vertex Coordinate Value Y (0 – 4095.9375)
Z 63:32 int 0:32:0 Vertex Coordinate Value Z

XYZ3 : Setting for Vertex Coordinate Values (without Drawing Kick)

0x0d
This register sets the coordinate values of the vertex and moves the vertex queue a step forward. The X and Y
coordinate values are specified in the primitive coordinate system.
Drawing is not started.

BIT ASSIGN
6 3 3 1 1 0
3 2 1 6 5 0

Z Y X

XYZF2 : Setting for Vertex Coordinate Values

0x04
This register sets the vertex coordinate values and the Fog coefficient, and then moves the vertex queue a step
forward. The X and Y coordinate values are specified in the primitive coordinate system.
When the vertex information stored in the vertex queue reaches a fixed number, drawing is started. The method
of moving the vertex queue forward and the number of vertices at which drawing starts depend on the type of
the drawing primitive.

BIT ASSIGN
6 5 5 3 3 1 1 0
3 6 5 2 1 6 5 0

F Z Y X

FIELD
Name Pos. Format Contents
X 15:0 int0:12:4 Vertex Coordinate Value X (0 – 4095.9375)
Y 31:16 int0:12:4 Vertex Coordinate Value Y (0 – 4095.9375)
Z 55:32 int0:24:0 Vertex Coordinate Value Z
F 63:56 int0:8:0 Fog Coefficient

XYZF3 : Setting for Vertex Coordinate Values (without Drawing Kick)

0x0c
This register sets the vertex coordinate values and the Fog coefficient, and then moves the vertex queue a step
forward.
Drawing is not started.

BIT ASSIGN
6 5 5 3 3 1 1 0
3 6 5 2 1 6 5 0

F Z Y X

ZBUF_1 / ZBUF_2 : Z Buffer Setting

0x4e / 0x4f
These registers make various settings regarding Z buffer. ZBUF_1 sets Context 1 and ZBUF_2 sets Context 2.

BIT ASSIGN
3 2 2 0 0
2 7 4 8 0
Z
M
PSM ZBP
S
K

FIELD
Name Pos. Format Contents
ZBP 8:0 int0:9:0 Z Buffer Base Pointer (ZBP = Word Address/2048)
PSM 27:24 int0:4:0 Z Value Storage Format
0000 PSMZ32
0001 PSMZ24
0010 PSMZ16
1010 PSMZ16S
ZMSK 32 int0:1:0 Z Value Drawing Mask
0 Z buffer is updated.
1 Z buffer is not updated regardless of the result of the depth
test.
The buffer width of the Z buffer is not set since it is the same size as that of the frame buffer.

7.2. Privileged Registers

The privileged registers are as shown below:

• System Control/Status Register

• PCRTC Setting Register
• Event ID Register

Since the privileged registers and general-purpose registers are mapped to different spaces, their register
addresses may be duplicated. Also, unlike the general-purpose registers, which have write access only, some of
the privileged registers can have read/write access. For clarification, the following information is added to the
register names in this section:

(r/w) Read/Write
(w) Write Only

BGCOLOR (w) : Background Color Setting

0x0e
This register sets the background color of the PCRTC with RGB value.

BIT ASSIGN
2 1 1 0 0 0
3 6 5 8 7 0

B G R

FIELD
Name Pos. Format Contents
R 7:0 int 0:8:0 Luminance of R element of background color
G 15:8 int 0:8:0 Luminance of G element of background color
B 23:16 int 0:8:0 Luminance of B element of background color

BUSDIR (w) : Host Interface Bus Switching

0x44
This register switches the direction of the FIFO used for data transmission with the host.
When switching, the FIFO must be empty. For details, see "4.2.2. Transmission from Local Buffer to Host".

BIT ASSIGN
0
0
D
I
R

FIELD
Name Pos. Format Contents
DIR 0 int 0:1:0 Transmission Direction of Interface
0 Host -> Local (Normal)
1 Local -> Host

CSR (r/w) : System Status

0x40
This register sets the mode and obtains the status of the GS system.

BIT ASSIGN
3 2 2 1 1 1 1 1 0 0 0 0 0 0 0 0 0
1 4 3 6 5 4 3 2 9 8 6 5 4 3 2 1 0
F F N R F E V H F S
I F E L D S S I I
I I
ID REV E E S U 0 0 W
I I I N
I
G
N
F L E S N N
L N S A
O D D T H T T T H L

FIELD
Name Pos. Format Contents
SIGNAL 0 int 0:1:0 SIGNAL Event Control
(Write)
0 Nothing is done.
1 Old event is cleared and event is enabled.
(Read)
0 SIGNAL event has not been generated.
1 SIGNAL event has been generated.
FINISH 1 int 0:1:0 FINISH Event Control
(Write)
0 Nothing is done.
1 Event is enabled.
(Read)
0 FINISH event has not been generated.
1 FINISH event has been generated.
HSINT 2 int 0:1:0 HSync Interrupt Control
(Write)
0 Nothing is done.
1 HSync interrupt is enabled.
(Read)
0 HSync interrupt has not been generated.
1 HSync interrupt has been generated.
VSINT 3 int 0:1:0 VSync Interrupt Control
(Write)
0 Nothing is done.
1 VSync interrupt is enabled.
(Read)
0 VSync interrupt has not been generated.
1 VSync interrupt has been generated.
EDWINT 4 int 0:1:0 Rectangular Area Write Termination Interrupt Control
(Write)
0 Nothing is done.
1 Rectangular area write interrupt is enabled.
(Read)
0 Rectangular area write interrupt has not been generated.
1 Rectangular area write interrupt has been generated.
FLUSH 8 int 0:1:0 Drawing Suspend and FIFO Clear (enabled during data write)
0 Not flushed.
1 Flushed.

Name Pos. Format Contents

RESET 9 int 0:1:0 GS System Reset (enabled during data write)
0 Not reset.
1 Reset.
NFIELD 12 int 0:1:0 Output Value of NFIELD Output
(Output value is updated by sampling at VSync.)
FIELD 13 int 0:1:0 Field Displayed Currently
(In Interlace Mode)
0 EVEN
1 ODD
(In Non-Interlace Mode)
0 Equivalent to EVEN
1 Equivalent to ODD
FIFO 15:14 int 0:2:0 Host Interface FIFO Status
00 Neither Empty nor Almost Full
01 Empty
10 Almost Full
11 Reserved
REV 23:16 int 0:8:0 Revision No. of the GS
ID 31:24 int 0:8:0 ID of the GS
Bit 5 and bit 6 should be set to 0 when data is written.

DISPFB1 (w) : Setting for Rectangular Area Read Output Circuit 1

0x07
This register makes settings for the frame buffer regarding information on Rectangular Area Read Output
Circuit 1 for the PCRTC.

BIT ASSIGN
5 4 4 3 1 1 1 0 0 0
3 3 2 2 9 5 4 9 8 0

DBY DBX PSM FBW FBP

FIELD
Name Pos. Format Contents
FBP 8:0 int0:9:0 Base Pointer (Address/2048)
FBW 14:9 int0:6:0 Buffer Width (Width/64)
PSM 19:15 int0:5:0 Pixel Storage Format
00000 PSMCT32
00001 PSMCT24
00010 PSMCT16
01010 PSMCT16S
10010 PS-GPU24
DBX 42:32 int0:11:0 X Position in Buffer of Upper Left Point of Rectangular Area
(in units of pixels)
DBY 53:43 int0:11:0 Y Position in Buffer of Upper Left Point of Rectangular Area
(in units of pixels)

DISPFB2 (w) : Setting for Rectangular Area Read Output Circuit 2

0x09
This register makes settings for the frame buffer regarding information on Rectangular Area Read Output
Circuit 2 for the PCRTC.

BIT ASSIGN
5 4 4 3 1 1 1 0 0 0
3 3 2 2 9 5 4 9 8 0

DBY DBX PSM FBW FBP

FIELD
Name Pos. Format Contents
FBP 8:0 int0:9:0 Base Pointer (Address/2048)
FBW 14:9 int0:6:0 Buffer Width (Width/64)
PSM 19:15 int0:5:0 Pixel Storage Format
00000 PSMCT32
00001 PSMCT24
00010 PSMCT16
01010 PSMCT16S
DBX 42:32 int0:11:0 X Position in Buffer of Upper Left Point of Rectangular Area
(in units of pixels)
DBY 53:43 int0:11:0 Y Position in Buffer of Upper Left Point of Rectangular Area
(in units of pixels)

DISPLAY1 (w) : Setting for Rectangular Area Read Output Circuit 1

0x08
This register makes settings for the display position on the screen regarding information on Rectangular Area
Read Output Circuit 1 for the PCRTC.

BIT ASSIGN
5 4 4 3 2 2 2 2 2 1 1 0
4 4 3 2 8 7 6 3 2 2 1 0
M M
A A
DH DW DY DX
G G
V H

FIELD
Name Pos. Format Contents
DX 11:0 int 0:12:0 X Position in the Display Area (in VCK units)
DY 22:12 int 0:11:0 Y Position in the Display Area (in Raster units)
MAGH 26:23 int 0:4:0 Magnification in H Direction
0000 x 1
0001 x 2
0010 x 3
0011 x 4
:
:
1111 x 16
MAGV 28:27 int 0:2:0 Magnification in V Direction
00 x1
01 x2
10 x3
11 x4
DW 43:32 int 0:12:0 Display Area Width - 1 (in VCK units)
DH 54:44 int 0:11:0 Display Area Height - 1 (in Pixel units)

DISPLAY2 (w) : Setting for Rectangular Area Read Output Circuit 2

0x0a
This register makes settings for the display position on the screen regarding information on Rectangular Area
Read Output Circuit 2 for the PCRTC.

BIT ASSIGN
5 4 4 3 2 2 2 2 2 1 1 0
4 4 3 2 8 7 6 3 2 2 1 0
M M
A A
DH DW DY DX
G G
V H

EXTBUF (w) : Setting for Feedback Write Buffer

0x0b
This register makes settings for the frame buffer used when feedback write is performed.

BIT ASSIGN
5 4 4 3 2 2 2 2 2 2 2 1 1 1 0
3 3 2 2 6 5 4 3 2 1 0 9 4 3 0
E E W
F
M M F
B
WDY WDX O O F EXBW EXBP
I
D D M
N
C A D

FIELD
Name Pos. Format Contents
EXBP 13:0 int 0:14:0 Base pointer of the buffer where data is written.
EXBP=Word Address/64
EXBW 19:14 int 0:6:0 Width of the buffer where data is written.
EXBW=Width in Pixel units/64
FBIN 21:20 int 0:2:0 Selection of Input Source
00 OUT1
01 OUT2
WFFMD 22 int 0:1:0 Interlace Mode
0 FIELD (Written to every other raster.)
1 FRAME (Written to every raster.)
EMODA 24:23 int 0:2:0 Method of Processing Input Alpha Value
00 Input Alpha value is written as it is.
01 Value converted from Input RGB to Luminance value Y is written.
10 Value converted from Input RGB to Luminance value Y and
reduced by half is written.
11 Always 0
EMODC 26:25 int 0:2:0 Method of Processing Input Color Value
00 Input RGB is written as it is.
01 Value converted from Input RGB to Luminance value Y is written
to RGB respectively.
10 Value converted from Input RGB to YCbCr is written.
11 Input Alpha value is written to RGB respectively.
WDX 42:32 int 0:11:0 X Coordinate in the buffer of upper left point of the rectangular area where
input image data is written.
WDY 53:43 int 0:11:0 Y Coordinate in the buffer of upper left point of the rectangular area where
input image data is written.

EXTDATA (w) : Feedback Write Setting

0x0c
This register sets the area to be read when feedback write is performed.

BIT ASSIGN
5 4 4 3 2 2 2 2 2 1 1 0
4 4 3 2 8 7 6 3 2 2 1 0
S
M
WH WW SMPH SY SX
P
V

FIELD
Name Pos. Format Contents
SX 11:0 int0:12:0 X Coordinate of upper left point of the rectangular area where input
image is written (in VCK units).
SY 22:12 int0:11:0 Y Coordinate of upper left point of the rectangular area where input
image is written (in Pixel units).
SMPH 26:23 int0:4:0 Sampling Rate in H Direction (in VCK units)
0000 Every VCK
0001 At intervals of 1 VCK
0010 At intervals of 2 VCKs
0011 At intervals of 3 VCKs
:
:
1111 At intervals of 15 VCKs
SMPV 28:27 int0:2:0 Sampling Rate in V Direction
00 Every H-Sync
01 At intervals of 1 H-Sync
10 At intervals of 2 H-Syncs
11 At intervals of 3 H-Syncs
WW 43:32 int0:12:0 Rectangular Area Width - 1
WH 54:44 int0:11:0 Rectangular Area Height - 1

EXTWRITE (w) : Feedback Write Function Control

0x0d
This register controls the feedback write operation (start-stop).

BIT ASSIGN
0
0
W
R
I
T
E

FIELD
Name Pos. Format Contents
WRITE 0 int 0:1:0 Activation/Deactivation of Write
0 Write to memory is completed in the current frame.
1 Write to memory is started from the next frame.

IMR (w) : Interrupt Mask Control

0x41
This register specifies whether various kinds of interrupts are masked.
All initial values after reset are 1 (masked).

BIT ASSIGN
1 1 1 1 1 0 0
4 3 2 1 0 9 8
F
E V H I S
D S S N I
W I G
* * M M S
M H M
S S S M S
K K K S K
K

FIELD
Name Pos. Format Contents
SIGMSK 8 int 0:1:0 SIGNAL Event Interrupt Mask
0 Not masked.
1 Masked.
FINISHMSK 9 int 0:1:0 FINISH Event Interrupt Mask
0 Not masked.
1 Masked.
HSMSK 10 int 0:1:0 HSync Interrupt Mask
0 Not masked.
1 Masked.
VSMSK 11 int 0:1:0 VSync Interrupt Mask
0 Not masked.
1 Masked.
EDWMSK 12 int 0:1:0 Rectangular Area Write Termination Interrupt Mask
0 Not masked.
1 Masked.
* Undefined bits should be set to 1.

PMODE (w) : PCRTC Mode Setting

0x00
This register makes various settings for the PCRTC.

BIT ASSIGN
1 1 0 0 0 0 0 0 0 0
6 5 8 7 6 5 4 2 1 0
S A M
E E
L M M
* ALP CRTMD N N
B O O
2 1
G D D

FIELD
Name Pos. Format Contents
EN1 0 int 0:1:0 Read Circuit 1 ON/OFF
0 OFF
1 ON
EN2 1 int 0:1:0 Read Circuit 2 ON/OFF
0 OFF
1 ON
CRTMD 4:2 int 0:3:0 CRT Output Switching
Always 001
MMOD 5 int 0:1:0 Alpha Value Selection for Alpha Blending
0 Alpha Value of Read Circuit 1
1 ALP Register Value
AMOD 6 int 0:1:0 OUT1 Alpha Output Selection
0 Alpha Value of Read Circuit 1
1 Alpha Value of Read Circuit 2
SLBG 7 int 0:1:0 Alpha Blending Method Selection
0 Blended with the output of Read Circuit 2.
1 Blended with the background color.
ALP 15:8 int 0:8:0 Fixed Alpha Value (0xff = 1.0)
* Undefined bits should be set to 0.

SIGLBLID (r/w) : Signal ID Value Read

0x48
This register obtains ID values of the SIGNAL event and LABEL event.
BIT ASSIGN
6 3 3 0
3 2 1 0

LBLID SIGID

FIELD
Name Pos. Format Contents
SIGID 31:0 int 0:32:0 ID Value Set by SIGNAL Register
LBLID 63:32 int 0:32:0 ID Value Set by LABEL Register

SMODE2 (w) : Setting for Modes Related to Video Synchronization

0x02
This register makes settings related to PCRTC video synchronization.

BIT ASSIGN
0 0 0 0
3 2 1 0
D F
I
P F
N
M M
T
S D

FIELD
Name Pos. Format Contents
INT 0 int 0:1:0 Interlace Mode Setting
0 Non-Interlace Mode
1 Interlace Mode
FFMD 1 int 0:1:0 Setting in Interlace Mode
0 FIELD Mode (Read every other line.)
1 FRAME Mode (Read every line.)
DPMS 3:2 int 0:2:0 VESA DPMS Mode Setting
00 On
01 Stand-by
10 Suspend
11 Off

7.3. Register List in Address Order

General Purpose Registers
Address Register Name Description
0x00 PRIM Drawing primitive setting
0x01 RGBAQ Vertex color setting
0x02 ST Vertex texture coordinate setting (texture coordinates)
0x03 UV Vertex texture coordinate setting (texel coordinates)
0x04 XYZF2 Vertex coordinate value setting
0x05 XYZ2 Vertex coordinate value setting
0x06 TEX0_1 Texture information setting
0x07 TEX0_2 Texture information setting
0x08 CLAMP_1 Texture wrap mode
0x09 CLAMP_2 Texture wrap mode
0x0a FOG Vertex fog value setting
0x0c XYZF3 Vertex coordinate value setting (without drawing kick)
0x0d XYZ3 Vertex coordinate value setting (without drawing kick)
0x14 TEX1_1 Texture information setting
0x15 TEX1_2 Texture information setting
0x16 TEX2_1 Texture information setting
0x17 TEX2_2 Texture information setting
0x18 XYOFFSET_1 Offset value setting
0x19 XYOFFSET_2 Offset value setting
0x1a PRMODECONT Specification of primitive attribute setting method
0x1b PRMODE Drawing primitive attribute setting
0x1c TEXCLUT CLUT position setting
0x22 SCANMSK Raster address mask setting
0x34 MIPTBP1_1 MIPMAP information setting (Level 1 – 3)
0x35 MIPTBP1_2 MIPMAP information setting (Level 1 – 3)
0x36 MIPTBP2_1 MIPMAP information setting (Level 4 – 6)
0x37 MIPTBP2_2 MIPMAP information setting (Level 4 – 6)
0x3b TEXA Texture alpha value setting
0x3d FOGCOL Distant fog color setting
0x3f TEXFLUSH Texture page buffer disabling
0x40 SCISSOR_1 Scissoring area setting
0x41 SCISSOR_2 Scissoring area setting
0x42 ALPHA_1 Alpha blending setting
0x43 ALPHA_2 Alpha blending setting
0x44 DIMX Dither matrix setting
0x45 DTHE Dither control
0x46 COLCLAMP Color clamp control
0x47 TEST_1 Pixel test control
0x48 TEST_2 Pixel test control
0x49 PABE Alpha blending control in pixel units
0x4a FBA_1 Alpha correction value
0x4b FBA_2 Alpha correction value
0x4c FRAME_1 Frame buffer setting
0x4d FRAME_2 Frame buffer setting
0x4e ZBUF_1 Z buffer setting
0x4f ZBUF_2 Z buffer setting
0x50 BITBLTBUF Setting for transmission between buffers
0x51 TRXPOS Specification for transmission area in buffers

Address Register Name Description

0x52 TRXREG Specification for transmission area in buffers
0x53 TRXDIR Activation of transmission between buffers
0x54 HWREG Data port for transmission between buffers
0x60 SIGNAL SIGNAL event occurrence request
0x61 FINISH FINISH event occurrence request
0x62 LABEL LABEL event occurrence request

Privileged Registers
Address Register Name Description
0x00 PMODE PCRTC mode setting
0x02 SMODE2 Mode setting related to video synchronization
0x07 DISPFB1 Setting for rectangular area read output circuit 1
0x08 DISPLAY1 Setting for rectangular area read output circuit 1
0x09 DISPFB2 Setting for rectangular area read output circuit 2
0x0a DISPLAY2 Setting for rectangular area read output circuit 2
0x0b EXTBUF Feedback write buffer setting
0x0c EXTDATA Feedback write setting
0x0d EXTWRITE Feedback write control
0x0e BGCOLOR Background color setting
0x40 CSR System status
0x41 IMR Interrupt mask control
0x44 BUSDIR Host interface bus switching
0x48 SIGLBLID Signal ID value read

(This page is left blank intentionally)

8. Details of GS Local Memory

The GS uses non-linear address conversion rules when converting from a two-dimensional address to a one-
dimensional physical memory address to improve the efficiency of memory access. This chapter describes the
address conversion rules.

8.1. Memory Access Units

The GS local memory can be considered to be a linear address space of 32-bit word units, but it cannot be
accessed uniformly due to its complex configuration affected by the DRAM's physical structure.
Since accessible units vary according to the type of processing, 3 processing units are defined as follows:

Page
Size: 8 Kbytes
FBP of FRAME (frame buffer) and ZBP of ZBUF (Z buffer) can point to page boundaries.
This is the same as a DRAM page. An access to a page does not result in a page break.

Block
Size: 256 bytes
TBP0 of TEX0 and TBPn of MIPTBP1/2 (texture buffer), and DBP and SBP of BITBLTBUF (destination
and source buffers during transmission between buffers) point to block boundaries.

Column
Size: 64 bytes (512 bits)
A column in a buffer is accessible in a single cycle.
Data can be read from and written to a column in the frame buffer and Z buffer in a single cycle (2048 bits
in total).

The relationships between page, block, and column are as follows from the point of view of memory
capacity:
Entire local memory = 512 pages = 16,384 blocks = 65,536 columns
1 page = 32 blocks
1 block = 4 columns
Each of page, block, and column corresponds to the pixels arranged in the form of a rectangle, in the two-
dimensional space in the buffer (i.e. when viewed as a drawn image or a texture image). The width and
height of the rectangle vary according to the pixel storage format.

Pixel Storage Format Page (W x H) Block (W x H) Column (W x H)

PSMCT32, PSMCT24, 64 x 32 pixels 8 x 8 pixels 8 x 2 pixels
PSMZ32, PSMZ24
PSMCT16, PSMCT16S, 64 x 64 pixels 16 x 8 pixels 16 x 2 pixels
PSMZ16, PSMZ16S
PSMT8 128 x 64 pixels 16 x 16 pixels 16 x 4 pixels
PSMT4 128 x 128 pixels 32 x 16 pixels 32 x 4 pixels

8.2. Page Arrangement in Buffer

The following is the page arrangement order in a buffer in a two-dimensional space:

(e.g.) Frame Buffer (FBP=0 and FBW=10)

y
0 1 2 3 4 cc 9

10 11 12 13 14 cc 19

20 21 cc

cc
Figure 8-1 Page Arrangement Order in Buffer

8.3. Data Structure in a Page

The relationship between page, block, and column, such as block allocation in a page or pixel arrangement order
in a column, varies from one pixel storage format to another.
It is illustrated by pixel storage format below. Data with the same block number and column number in the
same page shows the same memory regardless of the pixel storage format.

8.3.1. PSMCT32/PSMCT24/PSMZ32/PSMZ24

Block Configuration in a Page (Numbers in the figures show block addresses.)

PSMCT32, PSMCT24 PSMZ32, PSMZ24
64 pixels 64 pixels
0 1 4 5 16 17 20 21 24 25 28 29 8 9 12 13
2 3 6 7 18 19 22 23 26 27 30 31 10 11 14 15
8 9 12 13 24 25 28 29 16 17 20 21 0 1 4 5
10 11 14 15 26 27 30 31 18 19 22 23 2 3 6 7

Column Configuration in
a Block 8 pixels

COLUMN 0 2 pixels

COLUMN 1

8 pixels
COLUMN 2

COLUMN 3 Block size = 8x8 (pixels)

Column size = 8x2 (pixels)

Pixel Arrangement Order in a

Column
(Numbers in the figure show 32-bit word
addresses.) 0 1 4 5 8 9 12 13

2 3 6 7 10 11 14 15

PSMT8H, PSMT4HH, and PSMT4HL have the same data structure as PSMCT32. Effective pixel data is put
in the position of bits 31 to 24, 31 to 27, and 27 to 24 respectively.

8.3.2. PSMCT16/PSMCT16S

Block Configuration in a Page (Numbers in the figures show block addresses.)

PSMCT16 PSMZ16

64 pixels 64 pixels
0 2 8 10 24 26 16 18
1 3 9 11 25 27 17 19
4 6 12 14 28 30 20 22
5 7 13 15 29 31 21 23
16 18 24 26 8 10 0 2
17 19 25 27 9 11 1 3
20 22 28 30 12 14 4 6
21 23 29 31 13 15 5 7

PSMCT16S PSMZ16S

64 pixels 64 pixels
0 2 16 18 24 26 8 10
1 3 17 19 25 27 9 11
8 10 24 26 16 18 0 2
9 11 25 27 17 19 1 3
4 6 20 22 28 30 12 14
5 7 21 23 29 31 13 15
12 14 28 30 20 22 4 6
13 15 29 31 21 23 5 7

Column Configuration in a Block

16 pixels

COLUMN 0 2 pixels

COLUMN 1
8 pixels

COLUMN 2 Block size = 16x8 (pixels)

Column size = 16x2 (pixels)
COLUMN 3

Pixel Arrangement Order in a Column (Numbers in the figure show 32-bit word addresses.)

0 1 4 5 8 9 12 13 0 1 4 5 8 9 12 13

2 3 6 7 10 11 14 15 2 3 6 7 10 11 14 15

Bits 0-15 Bits 16-31

8.3.3. PSMT8

Block Configuration in a Page (Numbers in the figure show block addresses.)

128 pixels

0 1 4 5 16 17 20 21

2 3 6 7 18 19 22 23

8 9 12 13 24 25 28 29

10 11 14 15 26 27 30 31

Column Configuration in a Block

16 pixels

COLUMN 0 4 pixels

COLUMN 1

16 pixels

COLUMN 2

COLUMN 3

Pixel Arrangement Order in a Column (Numbers in the figures show 32-bit word addresses.)
Pixel arrangement order varies according to the column number in PSMT8.
COLUMNS 0 and 2
Bits 0-7 Bits 16-23

0 1 4 5 8 9 12 13 0 1 4 5 8 9 12 13

2 3 6 7 10 11 14 15 2 3 6 7 10 11 14 15

8 9 12 13 0 1 4 5 8 9 12 13 0 1 4 5

10 11 14 15 2 3 6 7 10 11 14 15 2 3 6 7

Bits 8-15 Bits 24-31

COLUMNS 1 and 3
Bits 0-7 Bits 16-23

8 9 12 13 0 1 4 5 8 9 12 13 0 1 4 5

10 11 14 15 2 3 6 7 10 11 14 15 2 3 6 7

0 1 4 5 8 9 12 13 0 1 4 5 8 9 12 13

2 3 6 7 10 11 14 15 2 3 6 7 10 11 14 15

Bits 8-15 Bits 24-31

8.3.4. PSMT4

Block Configuration in a Page (Numbers in the figure show block addresses.)

128 pixels

0 2 8 10

1 3 9 11

4 6 12 14

5 7 13 15

16 18 24 26

17 19 25 27

20 22 28 30

21 23 29 31

Column Configuration in a Block

32 pixels

4 pixels
COLUMN 0

COLUMN 1

16 pixels

COLUMN 2

COLUMN 3

Pixel Arrangement Order in a Column (Numbers in the figures show 32-bit word addresses.)
Pixel arrangement order varies according to the column number in PSMT4.
COLUMNS 0 and 2
Bits 0-3 Bits 8-11 Bits 16-19 Bits 24-27

0 1 4 5 8 9 12 13 0 1 4 5 8 9 12 13 0 1 4 5 8 9 12 13 0 1 4 5 8 9 12 13

2 3 6 7 10 11 14 15 2 3 6 7 10 11 14 15 2 3 6 7 10 11 14 15 2 3 6 7 10 11 14 15
8 9 12 13 0 1 4 5 8 9 12 13 0 1 4 5 8 9 12 13 0 1 4 5 8 9 12 13 0 1 4 5

10 11 14 15 2 3 6 7 10 11 14 15 2 3 6 7 10 11 14 15 2 3 6 7 10 11 14 15 2 3 6 7

Bits 4-7 Bits 12-15 Bits 20-23 Bits 28-31

COLUMNS 1 and 3
Bits 0-3 Bits 8-11 Bits 16-19 Bits 24-27

8 9 12 13 0 1 4 5 8 9 12 13 0 1 4 5 8 9 12 13 0 1 4 5 8 9 12 13 0 1 4 5

10 11 14 15 2 3 6 7 10 11 14 15 2 3 6 7 10 11 14 15 2 3 6 7 10 11 14 15 2 3 6 7

0 1 4 5 8 9 12 13 0 1 4 5 8 9 12 13 0 1 4 5 8 9 12 13 0 1 4 5 8 9 12 13

2 3 6 7 10 11 14 15 2 3 6 7 10 11 14 15 2 3 6 7 10 11 14 15 2 3 6 7 10 11 14 15

Bits 4-7 Bits 12-15 Bits 20-23 Bits 28-31

8.4. Occupied Memory Area

8 (W) x 8 (H) in PSMCT32 32 (W) x 32 (H) in PSMCT32
Occupied block: 0 only Occupied blocks: 0 to 15
0 1 4 5 16 17 20 21 0 1 4 5 16 17 20 21

2 3 6 7 18 19 22 23 2 3 6 7 18 19 22 23

8 9 12 13 24 25 28 29 8 9 12 13 24 25 28 29

10 11 14 15 26 27 30 31 10 11 14 15 26 27 30 31

8 (W) x 16 (H) in PSMCT32 8 (W) x 32 (H) in PSMCT32

Occupied blocks: 0 and 2 Occupied blocks: 0, 2, 8, and 10
0 1 4 5 16 17 20 21 0 1 4 5 16 17 20 21

2 3 6 7 18 19 22 23 2 3 6 7 18 19 22 23

8 9 12 13 24 25 28 29 8 9 12 13 24 25 28 29

10 11 14 15 26 27 30 31 10 11 14 15 26 27 30 31

8.5. Pointing within a Page

A buffer starting address points to a block boundary (not a page boundary), when using a texture (TBP0 of
TEX0 or TBPn of MIPTBP1/2) or transferring data between buffers (SBP or DBP of BITBLTBUF). When it
points to a block within a page, blocks are arranged as follows from the point of view of the texture buffer:
• The block pointed to is located at the upper left point in a two-dimensional space
• The block arrangement order conforms to the one obtained when the start of a page is pointed to
• The block pointed to and the ones following configure a one-page rectangle (64 x 32 in PSMCT32)
Examples of block arrangement order are shown below. Light gray blocks show the next page, followed by the
dark gray page.

Pointing to Block 1 in PSMCT32

64 pixels
1 2 5 6 17 18 21 22 1 2 5 6 17 18 21 22

3 4 7 8 19 20 23 24 3 4 7 8 19 20 23 24

9 10 13 14 25 26 29 30 9 10 13 14 25 26 29 30

11 12 15 16 27 28 31 0 11 12 15 16 27 28 31 0

Pointing to Block 9 in PSMCT32

64 pixels
9 10 13 14 25 26 29 30 9 10 13 14 25 26 29 30

11 12 15 16 27 28 31 0 11 12 15 16 27 28 31 0

17 18 21 22 1 2 5 6 17 18 21 22 1 2 5 6

19 20 23 24 3 4 7 8 19 20 23 24 3 4 7 8

8.6. Pixel Storage Format Conversion

When transferring PSMT4/PSMT8/PSMT16 texture data as PSMCT32 data, use the following rules to convert
pixel storage format in the preprocessing stage. This allows textures of different modes to be packed together
and transferred at one time as PSMCT32 data.
8.6.1. PSMT4
The following is an example of converting 1-column PSMT4 data (32 x 4) to PSMCT32 (8 x 2).

1-column PSMT4 Data (Pixel numbers in the figure are arranged in scan line order.)
32 pixels
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63
64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95
96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127

1-column PSMCT32 Data

8 pixels
A B C D E F G H

I J K L M N O P

PSMCT32 pixels (A to P) correspond to PSMT4 pixels (0 to 127) as follows:

COLUMNS 0 and 2 COLUMNS 1 and 3

Bit 31 Bit 0 Bit 31 Bit 0

A 92 24 84 16 76 88 68 00 A
88 28 80 20 72 12 64 4

B 93 25 85 17 77 99 69 11 B
89 29 81 21 73 13 65 5

C 94 26 86 18 78 10 70 22 C
90 30 82 22 74 14 66 6

D 95 27 87 19 79 11 71 33 D
91 31 83 23 75 15 67 7

E 88 28 80 20 72 12 64 44 E
92 24 84 16 76 8 68 0

F 89 29 81 21 73 13 65 55 F
93 25 85 17 77 9 69 1

G 90 30 82 22 74 14 66 66 G
94 26 86 18 78 10 70 2

H 91 31 83 23 75 15 67 77 H
95 27 87 19 79 11 71 3

I I
124
124 56 116
116 48 108
108 40 100
100 32 120 60 112 52 104 44 96 36

J J
125
125 57 117
117 49 109
109 41 101
101 33 121 61 113 53 105 45 97 37

K 102
102 K
126
126 58 118
118 50 110
110 42 34 122 62 114 54 106 46 98 38

L L
127
127 59 119
119 51 111
111 43 103
103 35 123 63 115 55 107 47 99 39

M M
120
120 60 112
112 52 104
104 44 96 36 124 56 116 48 108 40 100 32

N N
121
121 61 113
113 53 105
105 45 97 37 125 57 117 49 109 41 101 33

O O
122
122 62 114
114 54 106
106 46 98 38 126 58 118 50 110 42 102 34

P P
123
123 63 115
115 55 107
107 47 99 39 127 59 119 51 111 43 103 35

8.6.2. PSMT8
The following is an example of converting 1-column PSMT8 data (16 x 4) to PSMCT32 (8 x 2).

1-column PSMT8 Data (Pixel numbers in the figure are arranged in scan line order.)
16 pixels
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47

48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63

1-column PSMCT32 Data

8 pixels

A B C D E F G H

I J K L M N O P

PSMCT32 pixels (A to P) correspond to PSMT8 pixels (0 to 63) as follows:

COLUMNS 0 and 2 COLUMNS 1 and 3

Bit 31 Bit 0 Bit 31 Bit 0

A 44 8 36 0 A 40 12 32 4

B 45 9 37 1 B 41 13 33 5

C 46 10 38 2 C 42 14 34 6

D 47 11 39 3 D 43 15 35 7

E 40 12 32 4 E 44 8 36 0

F 41 13 33 5 F 45 9 37 1

G 42 14 34 6 G 46 10 38 2

H H
43 15 35 7 47 11 39 3

I I
60 24 52 16 56 28 48 20

J J
61 25 53 17 57 29 49 21

K K
62 26 54 18 58 30 50 22

L L
63 27 55 19 59 31 51 23

M M
56 28 48 20 60 24 52 16

N N
57 29 49 21 61 25 53 17

O O
58 30 50 22 62 26 54 18

P P
59 31 51 23 63 27 55 19

8.6.3. PSMT16
The following is an example of converting 1-column PSMT16 data (16 x 2) to PSMCT32 (8 x 2).

1-column PSMT16 Data (Pixel numbers in the figure are arranged in scan line order.)
16 pixels
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

1-column PSMCT32 Data

8 pixels
A B C D E F G H

I J K L M N O P

PSMCT32 pixels (A to P) correspond to PSMT16 pixels (0 to 31) as follows:

Bit 31 Bit 0

A 8 0

B 9 1

C 10 2

D 11 3

E 12 4

F 13 5

G 14 6

H 15 7

I 24 16

J 25 17

K 26 18

L 27 19

M 28 20

N 29 21

O 30 22

P 31 23

(This page is left blank intentionally)

EE Core Users Manual
No ratings yet
EE Core Users Manual
180 pages
EE Core Instruction Set Manual
No ratings yet
EE Core Instruction Set Manual
408 pages
Yoshi's Nightmare
No ratings yet
Yoshi's Nightmare
15 pages
VRay Rendering Tips for Beginners
No ratings yet
VRay Rendering Tips for Beginners
8 pages
CELL Handbook
No ratings yet
CELL Handbook
884 pages
Transmeta Crusoe: A Revolutionary CPU For Mobile Computing Ashraful Alam
No ratings yet
Transmeta Crusoe: A Revolutionary CPU For Mobile Computing Ashraful Alam
23 pages
The Pixelflow Texture and Image Subsystem: Steven Molnar
100% (2)
The Pixelflow Texture and Image Subsystem: Steven Molnar
11 pages
Hydra Pixar Presentation S5327-Jeremy-Cowles
No ratings yet
Hydra Pixar Presentation S5327-Jeremy-Cowles
23 pages
MDN 04 0213DG
No ratings yet
MDN 04 0213DG
95 pages
Formats: PSD AI Indd Jpeg PNG Gif PDF Tiff Eps SVG
No ratings yet
Formats: PSD AI Indd Jpeg PNG Gif PDF Tiff Eps SVG
2 pages
Cgworld 2017 08
No ratings yet
Cgworld 2017 08
119 pages
FormZ Users Manual
No ratings yet
FormZ Users Manual
2,008 pages
Advanced RenderMan 3-Render Harder
100% (1)
Advanced RenderMan 3-Render Harder
184 pages
KeyShot 4 - Manual
No ratings yet
KeyShot 4 - Manual
241 pages
Introduction To The Graphics Pipeline of The PS3
No ratings yet
Introduction To The Graphics Pipeline of The PS3
29 pages
Crusoe Processor Model TM5800 Features
No ratings yet
Crusoe Processor Model TM5800 Features
8 pages
3DO M2: Next-Gen 64-Bit Gaming Tech
No ratings yet
3DO M2: Next-Gen 64-Bit Gaming Tech
4 pages
DSL: An Overview: by M. V. Ramana Murthy
No ratings yet
DSL: An Overview: by M. V. Ramana Murthy
9 pages
Crusoe Processor: Seminar Guide: - By: - Prof. H. S. Kulkarni Ashish
No ratings yet
Crusoe Processor: Seminar Guide: - By: - Prof. H. S. Kulkarni Ashish
26 pages
Workstation Resource Guide
No ratings yet
Workstation Resource Guide
20 pages
PC Mag 1985 04 30 PDF
No ratings yet
PC Mag 1985 04 30 PDF
264 pages
The End of The Gpu Roadmap: Tim Sweeney CEO, Founder Epic Games
No ratings yet
The End of The Gpu Roadmap: Tim Sweeney CEO, Founder Epic Games
74 pages
Graphic Engines 3
No ratings yet
Graphic Engines 3
8 pages
ESXi Architecture
No ratings yet
ESXi Architecture
10 pages
Get Ready For Photogrammetry: Unity For Games E-Book
No ratings yet
Get Ready For Photogrammetry: Unity For Games E-Book
21 pages
Amd Cdna Whitepaper
No ratings yet
Amd Cdna Whitepaper
11 pages
3D Artist - Issue 131 2019
No ratings yet
3D Artist - Issue 131 2019
100 pages
Realism in Vue by Dax Pandhi
No ratings yet
Realism in Vue by Dax Pandhi
322 pages
Technical Artist Roadmap
No ratings yet
Technical Artist Roadmap
9 pages
Software Engineering: A Challenge of Designing A System: Nguyen Trung Dang Thuy s3509547
No ratings yet
Software Engineering: A Challenge of Designing A System: Nguyen Trung Dang Thuy s3509547
2 pages
Smart City: IEEE Standards Help Enable Smart City Technologies For Humanity
No ratings yet
Smart City: IEEE Standards Help Enable Smart City Technologies For Humanity
2 pages
2620 Final PDF
No ratings yet
2620 Final PDF
45 pages
IEEE Standard For POSIX Realtime and Embedded Application
No ratings yet
IEEE Standard For POSIX Realtime and Embedded Application
189 pages
Course28-Advanced Real-Time Rendering in 3D Graphics and Games SIGGRAPH07
No ratings yet
Course28-Advanced Real-Time Rendering in 3D Graphics and Games SIGGRAPH07
144 pages
Circular Polarization
No ratings yet
Circular Polarization
13 pages
Dragons Crown Art Works
No ratings yet
Dragons Crown Art Works
65 pages
Olivetti - MS-DOS 3.30 - User Guide
No ratings yet
Olivetti - MS-DOS 3.30 - User Guide
168 pages
Cell Processor Architecture Overview
100% (2)
Cell Processor Architecture Overview
27 pages
Introduction To Lighting: Light and Material
No ratings yet
Introduction To Lighting: Light and Material
2 pages
Paint Tool Sai Tutorial
100% (1)
Paint Tool Sai Tutorial
8 pages
Saunders y Novak - 2007 - Game Development Essentials Game Interface Design
No ratings yet
Saunders y Novak - 2007 - Game Development Essentials Game Interface Design
300 pages
Sc88 Pro Unlocked
No ratings yet
Sc88 Pro Unlocked
216 pages
A Look To Holocure
No ratings yet
A Look To Holocure
8 pages
User Guide
No ratings yet
User Guide
592 pages
Dmtllvip Regular
No ratings yet
Dmtllvip Regular
18 pages
SPR Flow
No ratings yet
SPR Flow
136 pages
Silicon One Buffer
No ratings yet
Silicon One Buffer
9 pages
(Ebook PDF) Design Basics 9th Edition by Stephen Pentak Download
100% (6)
(Ebook PDF) Design Basics 9th Edition by Stephen Pentak Download
44 pages
LightWave V9-Modeler Print
No ratings yet
LightWave V9-Modeler Print
428 pages
Vcarve Level 1 Step by Step
No ratings yet
Vcarve Level 1 Step by Step
17 pages
GS Users Manual Supplement
No ratings yet
GS Users Manual Supplement
22 pages
EE Overview Manual
No ratings yet
EE Overview Manual
64 pages
VU Users Manual
No ratings yet
VU Users Manual
368 pages
Gsprogramming 1
No ratings yet
Gsprogramming 1
48 pages
SPU2 Overview Manual
No ratings yet
SPU2 Overview Manual
80 pages
PS2 Optimisations
No ratings yet
PS2 Optimisations
39 pages
Literature Review On Ai Based 3d Gameli
No ratings yet
Literature Review On Ai Based 3d Gameli
7 pages
Architecture of The Super NES
No ratings yet
Architecture of The Super NES
18 pages
Review2 PPT - Meghana
No ratings yet
Review2 PPT - Meghana
32 pages
Netcool
No ratings yet
Netcool
39 pages
Python Variables and Data Types
No ratings yet
Python Variables and Data Types
49 pages
Performance Task 1 - Attempt Review Prog 114
100% (1)
Performance Task 1 - Attempt Review Prog 114
4 pages
ML All Chapter
No ratings yet
ML All Chapter
118 pages
Overview of SIRA Portal
No ratings yet
Overview of SIRA Portal
6 pages
Instructions: 1.apply The Shared Preferences Concepts. 2.create The Field Like in Given Task
No ratings yet
Instructions: 1.apply The Shared Preferences Concepts. 2.create The Field Like in Given Task
3 pages
Lesson 5
No ratings yet
Lesson 5
10 pages
Group Policy For Microsoft Office 2010
No ratings yet
Group Policy For Microsoft Office 2010
259 pages
iDS-7208HUHI-M1 X Datasheet 20241018
No ratings yet
iDS-7208HUHI-M1 X Datasheet 20241018
6 pages
OpenStack Install Guide 2024
No ratings yet
OpenStack Install Guide 2024
149 pages
DE10-Lite User Manual: June 5, 2020
No ratings yet
DE10-Lite User Manual: June 5, 2020
74 pages
EL1TNTT Training v0.1 31 02
No ratings yet
EL1TNTT Training v0.1 31 02
154 pages
Audio Headroom Essentials
No ratings yet
Audio Headroom Essentials
9 pages
Macronix Nbit Memory Family: 64M-Bit (X 1) Cmos Serial Eliteflash Memory
No ratings yet
Macronix Nbit Memory Family: 64M-Bit (X 1) Cmos Serial Eliteflash Memory
46 pages
Consumer Electronics Lab Manual
No ratings yet
Consumer Electronics Lab Manual
53 pages
Linux 2025
No ratings yet
Linux 2025
3 pages
Questions Bank With Answer Key
No ratings yet
Questions Bank With Answer Key
7 pages
Case Study of Linux
100% (1)
Case Study of Linux
8 pages
Noise Removal
No ratings yet
Noise Removal
16 pages
Towards Assessing The Maturity of OT Security Control Standards and Guidelines
No ratings yet
Towards Assessing The Maturity of OT Security Control Standards and Guidelines
6 pages
4-Bit Synchronous Counter: Final Project
No ratings yet
4-Bit Synchronous Counter: Final Project
7 pages
Automated Irrigation System Based On Irrigation Gates Using Fuzzy Logic
No ratings yet
Automated Irrigation System Based On Irrigation Gates Using Fuzzy Logic
5 pages
Empowerment Technology: Quarter 1 - Module 8
No ratings yet
Empowerment Technology: Quarter 1 - Module 8
24 pages
Sptve Icf 8 Q1 M6
No ratings yet
Sptve Icf 8 Q1 M6
13 pages
Online Job Application User Manual
No ratings yet
Online Job Application User Manual
4 pages
DevOps Q.
No ratings yet
DevOps Q.
13 pages
CV Trajkovic Jelena
No ratings yet
CV Trajkovic Jelena
2 pages
Alfa Account Server V2.2.36
No ratings yet
Alfa Account Server V2.2.36
26 pages
Data Analyst Cheat Sheet
No ratings yet
Data Analyst Cheat Sheet
28 pages