Skip to content

Commit

Permalink
add graphics architecture code and computing power
Browse files Browse the repository at this point in the history
  • Loading branch information
qinzhaoyu committed Dec 22, 2021
1 parent dc98a25 commit c5337cd
Showing 1 changed file with 66 additions and 0 deletions.
66 changes: 66 additions & 0 deletions capter2/ReadMe.md
Original file line number Diff line number Diff line change
Expand Up @@ -136,6 +136,7 @@ nvcc 会先将设备代码编译为 PTX(parrallel thread execution)伪汇编
在编译为 PTX 代码时,需要选项 `-arch=compute_XY` 指定一个虚拟架构的计算能力;在编译为 cubin 代码时,
需要选项 `-code=sm_ZW` 指定一个真实架构的计算能力,以确定可执行文件能够使用的 GPU。


真实架构的计算能力必须大于等于虚拟架构的计算能力,例如:

-arch=compute_35 -code=sm_60 (right)
Expand Down Expand Up @@ -167,3 +168,68 @@ nvcc 会先将设备代码编译为 PTX(parrallel thread execution)伪汇编

------

## 显卡架构和计算能力

1. 费米 Fermi(cuda 3.2~cuda 8)
SM20 or SM_20, compute_30 –
GeForce 400, 500, 600, GT-630.
CUDA 10 以后就完全不支持了。

2. 开普勒 Kepler(cuda 5~cuda 10)
SM30 or SM_30, compute_30 –
GeForce 700, GT-730
支持了统一内存模型编程

SM35 or SM_35, compute_35 –
Tesla K40.
支持动态并行化。

SM37 or SM_37, compute_37 –
Tesla K80.
增加了一些寄存器。

CUDA 11 以后就完全不支持了。

2. 麦克斯韦 Maxwell(CUDA 6~CUDA 11)
SM50 or SM_50, compute_50 –
Tesla/Quadro M 系列

SM52 or SM_52, compute_52 –
Quadro M6000 , GeForce 900, GTX-970, GTX-980, GTX Titan X

SM53 or SM_53, compute_53 –
Tegra (Jetson) TX1 / Tegra X1, Drive CX, Drive PX, Jetson Nano

cuda 11 以后彻底不支持

4. 帕斯卡 Pascal (CUDA 8 ~今)
SM60 or SM_60, compute_60 –
Quadro GP100, Tesla P100, DGX-1 (Generic Pascal)

SM61 or SM_61, compute_61–
GTX 1080, GTX 1070, GTX 1060, GTX 1050, GTX 1030, Titan Xp, Tesla P40, Tesla P4, Discrete GPU on the NVIDIA Drive PX2

SM62 or SM_62, compute_62 –
Integrated GPU on the NVIDIA Drive PX2, Tegra (Jetson) TX2

5. 伏特 Volta (CUDA 9 ~今)
SM70 or SM_70, compute_70 –
DGX-1 with Volta, Tesla V100, GTX 1180 (GV104), Titan V, Quadro GV100

SM72 or SM_72, compute_72 –
Jetson AGX Xavier, Drive AGX Pegasus, Xavier NX

6. 图灵Turing (CUDA 10 ~今)
SM75 or SM_75, compute_75 –
GTX/RTX Turing – GTX 1660 Ti, RTX 2060, RTX 2070, RTX 2080, Titan RTX, Quadro RTX 4000, Quadro RTX 5000, Quadro RTX 6000, Quadro RTX 8000, Quadro T1000/T2000, Tesla T4

7. 安培Ampere (CUDA 11 ~今)
SM80 or SM_80, compute_80 –
NVIDIA A100 (不再用特斯拉(Tesla)做名字了 – GA100), NVIDIA DGX-A100

SM86 or SM_86, compute_86 – (from CUDA 11.1 onwards)
Tesla GA10x, RTX Ampere – RTX 3080, GA102 – RTX 3090, RTX A6000, RTX A40

8. 哈珀Hopper (CUDA 12 计划中)
SM90 or SM_90, compute_90 –
NVIDIA H100 (GH100)

0 comments on commit c5337cd

Please sign in to comment.