ATI R600

R600

R600 GT

R600 XTX A11

R600 XTX A13

R600 GL

SIC66SIC66

Die Shot

Block Diagram

Block Diagram 2

Dispatch

ATI's R600 GPU uses the TeraScale architecture and is made using a 80 nm production process at TSMC. With a die size of 420 mm² and a transistor count of 720 million it is a very big chip. R600 supports DirectX 10.0 (Feature Level 10_0). For GPU compute applications, OpenCL version N/A can be used. It features 320 shading units, 16 texture mapping units and 16 ROPs.

Further reading: R600 Series Instruction Set Architecture

Graphics Processor

Released: May 14th, 2007

GPU Name: R600

Codename: Pele

Architecture: TeraScale

Foundry: TSMC

Process Size: 80 nm

Transistors: 720 million

Density: 1.7M / mm²

Die Size: 420 mm²

Graphics Features

DirectX: 10.0 (10_0)

OpenGL: 3.3 (full)
4.0 (partial)

OpenCL: N/A

Vulkan: N/A

Shader Model: 4.0

WDDM: 1.1

Compute: GFX3

DCE: 2.0

Render Config

Shading Units: 320

TMUs: 16

ROPs: 16

Compute Units: 4

Vertex Threads: 3

Pixel Threads: 4

Vertex Cache: 32 KB

Texture Cache: 32 KB

Tex L1 Cache: 32 KB per 4 SPs

L2 Cache: 256 KB

Max. TDP: 240 W

All TeraScale GPUs

ATI GPU Architecture History

2009-2015 TeraScale 2
2005-2013 TeraScale
2005-2007 Ultra-Threaded SE
2004-2008 R400
2002-2008 R300
2001-2006 Rage 7
2000 Rage 5
2000-2007 Rage 6
1999-2001 IBM
1998-1999 Rage 4
1997 Rage 2
1997-1999 Rage 3
1992-1997 Mach
1986-1992 Wonder

Graphics cards using the ATI R600 GPU

Name	Chip	Memory	Shaders	TMUs	ROPs	GPU Clock	Memory Clock
ATI Radeon HD 2900 XT	R600 XT	512 MB	320	16	16	743 MHz	828 MHz
ATI Radeon HD 2900 GT	R600 GT	256 MB	240	12	12	601 MHz	800 MHz
ATI Radeon HD 2900 PRO	R600 PRO	512 MB	320	16	16	600 MHz	800 MHz
ATI FireGL V8600	R600 V8600	1024 MB	320	16	16	688 MHz	868 MHz
ATI FireGL V8650	R600 V8650	2 GB	320	16	16	688 MHz	868 MHz
ATI FireGL V7600	R600 GL	512 MB	320	16	16	600 MHz	800 MHz
ATI Radeon HD 2900 XTX	R600 XTX A11	512 MB	320	16	16	743 MHz	828 MHz