Skip to content

Instantly share code, notes, and snippets.

Embed
What would you like to do?
Nvidia GPUs sorted by CUDA cores

List of desktop Nvidia GPUS ordered by CUDA core count

I created it for those who use Neural Style

Guys, please add your hardware setups, neural-style configs and results in comments!

GPU CUDA cores Memory Processor frequency
GeForce GTX TITAN Z 5760 12 GB 705 / 876
NVIDIA TITAN Xp 3840 12 GB 1582
GeForce GTX 1080 Ti 3584 11 GB 1582
GeForce GTX TITAN X 3072 12 GB 1000 / 1075
GeForce GTX 690 3072 4 GB 915 / 1019
GeForce GTX TITAN Black 2880 6 GB 889 / 980
GeForce GTX 780 Ti 2880 3 GB 875 / 928
GeForce GTX 980 Ti 2816 6 GB 1000 / 1075
GeForce GTX TITAN 2688 6 GB 837 / 876
GeForce GTX 1080 2560 8 GB 1607 / 1733
GeForce GTX 780 2304 3 GB 863 / 900
GeForce GTX 980 2048 4 GB 1126 / 1216
GeForce GTX 1070 1920 8 GB 1506 / 1683
GeForce GTX 970 1664 4 GB 1050 / 1178
GeForce GTX 770 1536 2 GB 1046 / 1085
GeForce GTX 680 1536 2 GB 1006 / 1058
GeForce GTX 760 Ti (OEM) 1344 2 GB 960
GeForce GTX 670 1344 2 GB 915 / 980
GeForce GTX 660 Ti 1344 2 GB 915 / 980
GeForce GTX 1060 (6GB) 1280 6 GB 1506 / 1708
GeForce GTX 960 (OEM) 1280 3 GB 924 / 980
GeForce GTX 760 192-bit(OEM) 1152 1.5 GB / 3 GB 980 / 1033
GeForce GTX 760 1152 2 GB 980 / 1033
GeForce GTX 1060 (3GB) 1152 3 GB 1506 / 1708
GeForce GTX 660 (OEM) 1152 1.5 GB / 3 GB 823 / 888
GeForce GTX 960 1024 2 GB 1127 / 1178
GeForce GTX 950 (OEM) 1024 2 GB 935 / 980
GeForce GTX 590 1024 3 GB 630
GeForce GTX 660 960 2 GB 980 / 1033
GeForce GTX 1050 Ti 768 4 GB 1290 / 1392
GeForce GTX 950 768 2 GB 1024 / 1188
GeForce GTX 650 Ti BOOST 768 2 GB 980 / 1033
GeForce GTX 650 Ti 768 1 GB 928
GeForce GTX 1050 640 2 GB 1354 / 1455
GeForce GTX 750 Ti 640 2 GB 1020 / 1075
GeForce GTX 645 (OEM) 576 1 GB 823
GeForce GTX 750 512 1 GB 1020 / 1085
GeForce GTX 580 512 1536 MB
GeForce GTX 480 480 1536 MB
GeForce GTX 570 480 1280 MB
GeForce GTX 295 480 1792 MB
GeForce GTX 470 448 1280 MB
GeForce GTX 745 (OEM) 384 4 GB
GeForce GT 740 384 1 GB / 2 GB
GeForce GT 730 96-384 1 GB / 2 GB 700 / 902
GeForce GT 635 (OEM) 384 2 GB
GeForce GTX 650 384 1 GB
GeForce GTX 560 Ti 384 1 GB
GeForce GTX 560 (OEM) 384 1280 MB / 2560 MB
GeForce GT 640 384 2 GB
GeForce GTX 465 352 1 GB
GeForce GTX 560 Ti (OEM) 352 1280 GB / 2560 GB
GeForce GTX 460 336 1 GB
GeForce GTX 560 336 1 GB
GeForce GTX 460 SE 288 1 GB
GeForce GTX 555 (OEM) 288 1 GB
GeForce GTX 285 for Mac 240 1 GB
GeForce GTX 285 240 1 GB
GeForce GTX 280 240 1 GB
GeForce GT 720 192 1 GB / 2 GB
GeForce GT 710 192 2 GB 954
GeForce GTS 450 192 1 GB
GeForce GTX 550 Ti 192 1 GB
GeForce GT 630 (OEM) 192 1 GB / 2 GB
GeForce GT 640 (OEM) 144 / 384 1 GB / 2 GB
GeForce GT 545 GDDR5 (OEM) 144 1 GB
GeForce GT 545 DDR3 144 1.5 GB / 3 GB
GeForce GTS 250 128 1 GB
GeForce GTS 150 128 1 GB
GeForce GTS 240 (OEM Product) 112 1 GB
GeForce GT 630 96 1 GB 700~902
GeForce GT 620 96 1 GB 700
GeForce GT 440 96 1 GB 810
GeForce GT 430 96 1 GB 700
GeForce GT 530 (OEM) 96 1 GB / 2GB
GeForce GT 340 (OEM) 96 1 GB
GeForce GT 330 (OEM) 96-112 1 GB / 2GB
GeForce GT 240 96 1GB
GeForce GT 320 (OEM Product) 72 1 GB
GeForce GT 705 (OEM) 48 1 GB
GeForce GT 620 (OEM) 48 1 GB
GeForce GT 610 48 1 GB
GeForce GT 520 (OEM) 48 1 GB / 2 GB
GeForce GT 520 48 1 GB
GeForce GT 220 48 1 GB
GeForce 605 (OEM) 48 1 GB
GeForce 510 (OEM) 48 1 GB / 2 GB
GeForce 405 (OEM) 16 1 GB
GeForce 310 (OEM) 16 1 GB
@cavinsmith

This comment has been minimized.

Copy link
Owner Author

commented May 11, 2016

Neural style configuration working on macbook (gt640m 384 cores, 625mhz):

th neural_style.lua -model_file models/nin_imagenet_conv.caffemodel -proto_file models/train_val.prototxt -gpu 0 -num_iterations 800 -seed 123 -content_layers relu0,relu3,relu7,relu12 -style_layers relu0,relu3,relu7,relu12 -content_weight 10 -style_weight 1000 -init image -save_iter 100 -print_iter 10 -optimizer adam -style_image art/art.jpg -content_image images/image.jpg -image_size 750

To avoid memory issues close browsers, adobe products and itunes prior to run. If you still have issues after it - try smaller image_size should work.

@MShahzaib

This comment has been minimized.

Copy link

commented Nov 10, 2017

Is there any way you can also report memory bandwidth for each card

@wureka

This comment has been minimized.

Copy link

commented Mar 18, 2018

Could you provide Jetson TX2's cuda score ?
Thanks

@LuanWH

This comment has been minimized.

Copy link

commented May 22, 2018

When I worked on CUDA projects back in 2014/5, my GT635M only has 144 CUDA cores and I felt like it was sooooo powerful. Now take a look at the bloody list.....

@Lordbl4

This comment has been minimized.

Copy link

commented Sep 21, 2018

I have GT330 2gb ddr2 version. It is not work with torch. Latest nvidia driver for this card is 340.104, cuda version 6.5.14, OS: Ubuntu 16.04 Server.
Driver and cuda is ok, but torch cant compile on stage cd ~/torch; ./install.sh

CMake Error at THC_generated_THCReduceApplyUtils.cu.o.cmake:207
nvcc fatal : Unsupported gpu architecture 'compute_52' - OpenCV 2.5 build

@Starscream9559

This comment has been minimized.

Copy link

commented Jun 23, 2019

This table is updated as of... 2018? 2019?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
You can’t perform that action at this time.