Cannot compile on multi gpu machine

GoogleCodeExporter commented 8 years ago

What steps will reproduce the problem?
1. Running cmake
2. Run make
3.

What is the expected output? What do you see instead?

 Building NVCC (Device) object src/cudpp/CMakeFiles/cudpp.dir/app/./cudpp_generated_tridiagonal_app.cu.o
ptxas /tmp/tmpxft_00005ec8_00000000-5_tridiagonal_app.compute_10.ptx, line 615; 
warning : Double is not supported. Demoting to float
nvcc �%6X: Option '-cubin' is not allowed when compiling for multiple GPU 
code instances
CMake Error at /project/GPU0/cudpp/cmake/FindCUDA/parse_cubin.cmake:38 (file):
  file Internal CMake error when trying to open file:
   /project/GPU0/cudpp/src/cudpp/CMakeFiles/cudpp.dir/app/./cudpp_generated_tridiagonal_app.cu.o.cubin.txt
  for reading.

What version of the product are you using? On what operating system?
cudpp 2.0 , CUDA Version:3.10 , Fedora release 10 (Cambridge)  Linux x86_64 
Platform 

Please provide any additional information below.

Using the following card : Nvidia Tesla S1070 1U

Original issue reported on code.google.com by judeb...@gmail.com on 13 Aug 2012 at 4:21

GoogleCodeExporter commented 8 years ago

Please upgrade your CUDA version to at least 4.0.

What version of CMake are you using?  Our CMake files should not be building 
with -cubin option.

Original comment by harr...@gmail.com on 10 Sep 2012 at 10:07

GoogleCodeExporter commented 8 years ago

No response.  Closing as invalid.  Please reopen if you provide more 
information.

Original comment by harr...@gmail.com on 24 Sep 2012 at 10:16

Changed state: Invalid

tozhovez / cudpp

Cannot compile on multi gpu machine #119