Compile fails using CUDA 6.5RC

[Bryan Novack]

I'm attempting to compile cudpp 2.1 with cuda 6.5RC. It fails every time with the following error:

nvcc fatal : unsupported gpu architecture 'compute_10'

I did some digging into the CMakeLists files and found in src/cudpp/CMakeLists.txt (line 91) where it sets GENCODE to compute_10 without first checking the CUDPP_GENCODE_SM10 option.

I was able to get around the problem by commenting this line out, but I don't know if I am potentially introducing an instability to cudpp.

Thoughts?

Bryan

[Mark Harris]

Starting with CUDA 6.5 support is being dropped for sm_10 (the G80 architecture).  So we should also drop support from CUDPP (people who need it can use old versions of CUDA and CUDPP).  We should remove sm_10 in the master branch.

I created issue #141 to get this fixed.

https://github.com/cudpp/cudpp/issues/141

Mark

--
You received this message because you are subscribed to the Google Groups "CUDPP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cudpp+un...@googlegroups.com.
To post to this group, send email to cu...@googlegroups.com.
Visit this group at http://groups.google.com/group/cudpp.
For more options, visit https://groups.google.com/d/optout.