search

soedinglab / CCMpred

Protein Residue-Residue Contacts from Correlated Mutations predicted quickly and accurately.
http://www.ncbi.nlm.nih.gov/pubmed/25064567
GNU Affero General Public License v3.0
107 stars 25 forks source link

Compiling with CUDA #16

Open huhlim opened 5 years ago

huhlim commented 5 years ago

I'm using a CUDA 9.2, and the nvcc did not support -arch=sm_20 option. I simply fixed the issue by replacing sm_20 to sm_70 in CMakeLists.txt file.

huhlim commented 5 years ago

With -arch=sm_70, I could compile it, but the compiled version failed to run on GTX980Ti. By using -arch=sm_50, I succeeded to run it.

croth1 commented 5 years ago

Thanks for reporting the problem @huhlim. Unfortunately, I do not have any cuda experience nor a graphics card so I cannot help maintaining the cuda support.

I will leave your issue open as your fix may help others struggling with cuda.

rmrao commented 5 years ago

Thanks for leaving this up! This fix also worked for me.

bognabognabogna commented 4 years ago

Hello, I've done the same but then I got the following error: /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/include/c++/v1/string.h:61:15: fatal error: 'string.h' file not found

does anyone know how to fix it?

huhlim commented 4 years ago

@bognabognabogna I think it is a completely different issue with your MacOS environment. Please check this out. https://stackoverflow.com/questions/26185978/macos-wchar-h-file-not-found

sseemayer commented 4 years ago

Hi all, please check out #20 to see if you can get this running on your machine with CUDA now.

rmrao commented 1 year ago

Since this issue is still open - just noting that this does work on latest CUDA versions now. Thanks for the fix!