AMD Granville

Granville

Block Diagram

AMD's Granville GPU uses the TeraScale 2 architecture and is made using a 40 nm production process at TSMC. With a die size of 166 mm² and a transistor count of 1,040 million it is a small chip. Granville supports DirectX 11.2 (Feature Level 11_0). For GPU compute applications, OpenCL version 1.2 can be used. It features 800 shading units, 40 texture mapping units and 16 ROPs.

Further reading: Evergreen Series Instruction Set Architecture

Graphics Processor

Released: Jan 4th, 2011

GPU Name: Granville

Desktop Variant: Juniper

Generation: Evergreen

Architecture: TeraScale 2

Foundry: TSMC

Process Size: 40 nm

Transistors: 1,040 million

Density: 6.3M / mm²

Die Size: 166 mm²

Package: FCBGA-962

Graphics Features

DirectX: 11.2 (11_0)

OpenGL: 4.4

OpenCL: 1.2

Vulkan: N/A

Shader Model: 5.0

WDDM: 1.3

Compute: GFX4

DCE: 4.0

UVD: 2.3

Render Config

Shading Units: 800

TMUs: 40

ROPs: 16

Compute Units: 10

Z-Stencil: 40

L1 Cache: 8 KB per CU

L2 Cache: 256 KB

Max. TDP: 50 W

All TeraScale 2 GPUs

AMD GPU Architecture History

2023 CDNA 3.0
2022-2024 RDNA 3.0
2021 CDNA 2.0
2020 CDNA 1.0
2020-2023 RDNA 2.0
2019-2020 RDNA 1.0
2018-2022 GCN 5.1
2017-2020 GCN 5.0
2016-2020 GCN 4.0
2014-2019 GCN 3.0
2013-2017 GCN 2.0
2011-2020 GCN 1.0
2010-2013 TeraScale 3
2009-2015 TeraScale 2
2005-2013 TeraScale
2005-2007 Ultra-Threaded SE

Graphics cards using the AMD Granville GPU

Name	Chip	Memory	Shaders	TMUs	ROPs	GPU Clock	Memory Clock
AMD Radeon HD 6870M	Granville XT	1024 MB	800	40	16	675 MHz	1000 MHz
AMD Radeon HD 6830M	Granville LP	1024 MB	800	40	16	575 MHz	800 MHz
AMD Radeon HD 6850M	Granville PRO	1024 MB	800	40	16	675 MHz	800 MHz