CUDA_ERROR_NO_BINARY_FOR_GPU

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Clean install of pyrit with cuda
2. Run pyrit list_cores
3.

What is the expected output? What do you see instead?
I expect the cores (mine is GeForce GT 330M)

what I see instead is:
Pyrit 0.4.1-dev (svn r308) (C) 2008-2011 Lukas Lueg http://pyrit.googlecode.com
This code is distributed under the GNU General Public License v3+

Traceback (most recent call last):
  File "/usr/local/bin/pyrit", line 6, in <module>
    pyrit_cli.Pyrit_CLI().initFromArgv()
  File "/Library/Python/2.7/site-packages/pyrit_cli.py", line 117, in initFromArgv
    func(self, **options)
  File "/Library/Python/2.7/site-packages/pyrit_cli.py", line 293, in list_cores
    with cpyrit.cpyrit.CPyrit() as cp:
  File "/Library/Python/2.7/site-packages/cpyrit/cpyrit.py", line 440, in __init__
    self.cores.append(CUDACore(queue=self, dev_idx=dev_idx))
  File "/Library/Python/2.7/site-packages/cpyrit/cpyrit.py", line 245, in __init__
    _cpyrit_cuda.CUDADevice.__init__(self, dev_idx)
SystemError: CUDA_ERROR_NO_BINARY_FOR_GPU

What version of the product are you using? On what operating system?

I'm using the latest svn, on OSX 10.9.4. Pyrit is working fine, cpyrit installs 
without issues after modify setup.py to identify the path for nvcc. But it 
doesn't run properly as I reported above. CUDA drivers and toolkit are 6.5.14. 

Please provide any additional information below.

I've followed the instructions to test CUDA: it is working properly on my 
system (I've tested all of the samples that come with the toolkit). I checked 
the previous posts for similar issues. Some of them seem similar but the 
solutions do not fit my specific case. I've checked google and the various 
tutorials and guides. Probably I'm missing something stupid. But it is missing. 
Annoying.

Original issue reported on code.google.com by francesc...@gmail.com on 13 Sep 2014 at 9:06

[GoogleCodeExporter]

Same issue here. 

Original comment by bag...@gmail.com on 5 Oct 2014 at 12:29

[GoogleCodeExporter]

I also have this problem

Original comment by sc2big...@gmail.com on 7 Oct 2014 at 1:33

[GoogleCodeExporter]

Ran into the same problem with no avail.  All CUDA samples run fine, but 
receive this error from pyrit's list_cores and selftest. 

I am using Linux Mint 17 (Ubuntu 14.04 base) and complied both pyrit and cpyrit 
from source, running 0.4.0

Original comment by erich...@gmail.com on 24 Nov 2014 at 7:18

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

I had the same problem and fixed it by modifying setup.py and specifying the 
correct gpu architecture for my graphics card. I had a GeForce GT 330 which is 
compatible with compute 1.2, so in the call to nvcc you have to add an option 
to specify gpu architecture:

nvcc_cmd = NVCC + bit_flag + ' --host-compilation C'\
                                         ' --gpu-architecture compute_12'\
                                         ' --verbose'\
                                         ' -Xcompiler "-fPIC" --ptx' \
                                         ' ./_cpyrit_cudakernel.cu'

otherwise it specifies compute_20 which is not supported by the card.

unfortunately this old card doesn't provide too much better speed, but at least 
it works

Original comment by nice...@gmail.com on 13 Dec 2014 at 6:36

[GoogleCodeExporter]

i had the same problem. Geforce GTX 660 OC, AMD FX 6100 Sixcore and 16GB RAM 
DDR3.

The Solution by "#5" is not working. My GPU is 3.0

Not working whith:
' --gpu-architecture compute_12'\
' --gpu-architecture compute_20'\
' --gpu-architecture compute_30'\

nvida driver and cuda toolkit working.
no one has a solution? There are other programs that crack with GPU?

Original comment by marco.we...@googlemail.com on 5 Jun 2015 at 10:10