分类
学界业界动态

面向深度学习用户的Nvidia各版本GPU参数汇总整理

浏览量： 175

做深度学习离不开算力，就像到达诗和远方的旅途离不开交通工具。本文面向深度学习用户整理市面上常用的各个版本GPU的关键参数，进行汇总整理，方便大家随时查阅。如果数据有误，或有新GPU问世，或者需要补充的参数维度，各位看官可以及时通过联系方式来找我更新数据。

注：仅统计Memory在 11GB以上、发售时间在2016年及以后、算力性能满足深度学习计算要求(Cuda算力性能>=3.5)的GPU。若表中单元格为空，则为暂无数据或数据待补充。博主将不定期移除上市时间最久、已停产过时、市面上已大量退役的GPU数据信息项。

(当前数据整理截止为2024年11月)

GeForce系列

版本型号MemoryCUDA CoreCUDA性能半精度TFLOPs单精度TFLOPs双精度TFLOPs显存位宽总线带宽最大功率发布时间原价￥
50系RTX 509032GB
GDDR7		21760512位600w未上市15999
19999
50系RTX 508016GB
GDDR7		10752256位400w未上市
40系RTX 4090 Ti24GB未上市15999
40系RTX 409024GB
GDDR6X		163848.982.5882.581.290384位1008GB/s450w2022
10.12		12999
40系RTX 4090D24GB145928.973.5473.541.149384位1008GB/s425w2023
12.28		11999
40系RTX 4080 Super16GB102408.951.351.30.802256位736GB/s320w2024
01.31		8099
40系RTX 408016GB97288.948.7448.740.762256位716.8GB/s320w2022
11.16		9499
40系RTX 4070 Ti Super16GB84488.944.144.10.689256位672GB/s285w2024
01.24		6499
40系RTX 4070 Ti12GB76808.940.0940.090.626192位504.2GB/s285w2023
01.05		6499
40系RTX 4070 Super12GB71688.935.4835.480.554192位504.2GB/s220w2024
01.17		4899
40系RTX 407012GB58888.929.1529.150.455192位504.2GB/s200w2023
04.13		4799
30系RTX 3090 Ti24GB107528.633.54
39.99		33.54
39.99		0.524
0.625		384位1008GB/s450w2022
03.29		14999
30系RTX 309024GB104968.629.38
35.68		29.28
35.58		0.459
0.558		384位935.8GB/s350w2020
09.02		11999
30系RTX 3080 Ti12GB102408.628.06
34.10		28.57
34.71		0.438
0.533		384位912.4GB/s350w2021
06.03		8999
30系RTX 308012GB
10GB		8960
8704		8.625.06
29.76		25.07
29.77		0.392
0.465		384位
320位		760GB/s350w
320w		2021
06.03		6357
20系RTX 2080 Ti11GB
GDDR6		43527.523.5
(26.9)		11.8
(13.4)		0.37
(0.42)		352位616GB/s260w2018
09.20		9999
10系GTX 1080 Ti11GB
GDDR5X		35846.10.166
(0.177)		10.609
(11.34)		0.332
(0.354)		352位484GB/s250w2017
03.10		5699

Tesla系列

版本型号MemoryCUDA CoreCUDA性能半精度TFLOPs单精度TFLOPs双精度TFLOPs总线位宽总线带宽最大功率发布时间原价￥
H系H10080GB
HBM2e/HBM3		14592
16896		9.0756.449
989.43		51.2
66.9		25.6
33.5		5120位2039GB/s350w
700w		2022.03.22
L系L4048GB
GDDR6		181768.9362.06690.5161.414384位864GB/s300w2022.10.13
L系L424GB
GDDR6		74248.912130.30.49192位300GB/s72w2023.03.21
A系A10040GB/80GB
HBM2		69128.031219.59.75120位1555GB/s250w2020.05.14
A系A4048GB
GDDR6		107528.6149.6837.421.168384位695.8GB/s300w2020.10.05
A系A3024GB
HBM2		35848.0165.1210.325.1613072位933.1GB/s165w2021.04.12
A系A1024GB
GDDR6		92168.6124.9631.240.976384位600GB/s150w2021.04.12
A系A164*16GB
GDDR6		4*12808.64*18.4324*4.6081.08484*128位4*200GB/s250w2021.04.12
A系A216GB
GDDR6		12808.618.1244.5310.14128位200GB/s40-60w2021.11.10
T系T416GB
GDDR6		25607.564.88.1256位320GB/s70w2018.09.12
V系V10016GB/32GB
HBM2		51207.0119.192
112.224
105.680		14.899
14.028
13.210		7.450
7.014
6.605		4096位900GB/s
829.44GB/s		250w2017.05.10
P系P10016GB/12GB
HBM2		35846.0不支持8.071‒9.344.036‒4.674096位
3072位		732GB/s
549GB/s		250w2016.04.05
P系P4024GB
GDDR5		38406.1不支持10.007–11.7580.3127–0.3674384位345.6GB/s250w2016.09.13

Quadro系列

型号
MemoryCUDA CoreCUDA性能半精度TFLOPs单精度TFLOPs双精度TFLOPs显存位宽总线带宽最大功率发布时间原价￥
RTX 600048GB
GDDR6		181768.991.1384位960GB/s300w
RTX 500032GB
GDDR6		250w
RTX 450024GB
GDDR6		210w
RTX 400020GB
GDDR6		130w
RTX 4000 SFF20GB
GDDR6		70w
RTX 200016GB
GDDR6		70w
RTX A600048GB
GDDR6		8.6300w
RTX A500024GB
GDDR6		8.6230w
RTX A450020GB
GDDR6		200w
RTX A400016GB
GDDR6		8.6140w
Quadro RTX 800048GB
GDDR6		7.5
Quadro RTX 600024GB
GDDR6		7.5
Quadro RTX 500016GB
GDDR6		7.5
Quadro GV10032GB
HBM2		7.0
Quadro GP10016GB
HBM2		6.02017
Quadro P600024GB
GDDR5X		6.1
Quadro P500016GB
GDDR5X		6.1
Quadro M6000 24GB24GB
GDDR5		5.22016.03.05

参考来源

  1. https://www.nvidia.cn/geforce/graphics-cards/40-series/
  2. https://www.nvidia.com/en-us/geforce/graphics-cards/40-series/rtx-4090/
  3. https://www.nvidia.cn/geforce/graphics-cards/30-series/
  4. https://www.nvidia.cn/geforce/graphics-cards/compare/
  5. https://detail.zol.com.cn/1208/1207097/param.shtml
  6. https://developer.nvidia.com/cuda-gpus
  7. https://zh.wikipedia.org/wiki/NVIDIA_Tesla
  8. https://www.bilibili.com/read/cv33922816/
  9. https://zh.wikipedia.org/wiki/NVIDIA_GeForce_40%E7%B3%BB%E5%88%97
  10. https://zh.wikipedia.org/wiki/NVIDIA_GeForce_30%E7%B3%BB%E5%88%97
  11. https://zh.wikipedia.org/wiki/NVIDIA_GeForce_20%E7%B3%BB%E5%88%97
  12. https://zh.wikipedia.org/wiki/NVIDIA_GeForce_10%E7%B3%BB%E5%88%97
  13. https://ai.oldpan.me/t/topic/287
  14. https://en.wikipedia.org/wiki/GeForce_40_series
  15. https://www.nvidia.com/en-us/design-visualization/rtx-6000/
  16. https://www.nvidia.com/en-us/design-visualization/desktop-graphics/
  17. https://resources.nvidia.com/en-us-design-viz-stories-ep/l40-linecard?lx=CCKW39&&search=professional%20graphics
  18. https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/quadro-product-literature/NV-DS-Quadro-M6000-24GB-US-NV-fnl-HR.pdf
  19. https://images.nvidia.com/content/pdf/quadro/data-sheets/192195-DS-NV-Quadro-P5000-US-12Sept-NV-FNL-WEB.pdf
  20. https://images.nvidia.com/content/pdf/quadro/data-sheets/192152-NV-DS-Quadro-P6000-US-12Sept-NV-FNL-WEB.pdf
  21. https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/productspage/quadro/quadro-desktop/quadro-pascal-gp100-data-sheet-us-nv-704562-r1.pdf
  22. https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/productspage/quadro/quadro-desktop/quadro-volta-gv100-data-sheet-us-nvidia-704619-r3-web.pdf
  23. https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/quadro-product-literature/quadro-rtx-5000-data-sheet-us-nvidia-704120-r4-web.pdf
  24. https://www.nvidia.cn/design-visualization/rtx-5000/
  25. https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/quadro-product-literature/quadro-rtx-6000-us-nvidia-704093-r4-web.pdf
  26. https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/quadro-product-literature/quadro-rtx-8000-us-nvidia-946977-r1-web.pdf
版权声明
本博客的文章除特别说明外均为原创，本人版权所有。欢迎转载，转载请注明作者及来源链接，谢谢。
本文地址: https://blog.ailemon.net/2024/09/25/nvidia-gpu-params-for-deep-learning/
All articles are under Attribution-NonCommercial-ShareAlike 4.0
发表你的看法 “点击发表你的看法”

关注“AI柠檬博客”微信公众号，及时获取你最需要的干货。


打赏微海报分享

发表回复

您的电子邮箱地址不会被公开。 必填项已用 * 标注

17 − 17 =

如果您是第一次在本站发布评论，内容将在博主审核后显示，请耐心等待