You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I was wondering about the behavior of gpuarray when the PCI busid for the CUDA device is over 13 characters in length ?
I had an issue where the PCI bus id on my machine is over 13 characters and I received the following error when running a simple gpu test program with Theano.
ERROR (theano.gpuarray): Could not initialize pygpu, support disabled Traceback (most recent call last): File "/usr/local/lib/python2.7/dist-packages/theano/gpuarray/__init__.py", line 227, in <module> use(config.device) File "/usr/local/lib/python2.7/dist-packages/theano/gpuarray/__init__.py", line 214, in use init_dev(device, preallocate=preallocate) File "/usr/local/lib/python2.7/dist-packages/theano/gpuarray/__init__.py", line 166, in init_dev unique_id = '(' + context.unique_id + ')' File "pygpu/gpuarray.pyx", line 1102, in pygpu.gpuarray.GpuContext.unique_id.__get__ File "pygpu/gpuarray.pyx", line 487, in pygpu.gpuarray.ctx_property GpuArrayException: cuDeviceGetPCIBusId((char *)res, 13, id): CUDA_ERROR_INVALID_VALUE: invalid argument [Elemwise{exp,no_inplace}(<TensorType(float32, vector)>), GpuFromHost<None>(Elemwise{exp,no_inplace}.0)]
Looking at the gpuarray code, I found that the line where this error occurred was trying to print the active device configuration. I tried setting print_active_device=false
as part of my THEANO_FLAGS and interestingly the code worked and I was able to monitor with nvidia-smi that the code used the gpu.
I was wondering if there is some issue when the PCI BUSID is over a certain length, the busID on my machine is 17 characters in total.
Any ideas ?
Thanks.
The text was updated successfully, but these errors were encountered:
Hi,
I was wondering about the behavior of gpuarray when the PCI busid for the CUDA device is over 13 characters in length ?
I had an issue where the PCI bus id on my machine is over 13 characters and I received the following error when running a simple gpu test program with Theano.
ERROR (theano.gpuarray): Could not initialize pygpu, support disabled Traceback (most recent call last): File "/usr/local/lib/python2.7/dist-packages/theano/gpuarray/__init__.py", line 227, in <module> use(config.device) File "/usr/local/lib/python2.7/dist-packages/theano/gpuarray/__init__.py", line 214, in use init_dev(device, preallocate=preallocate) File "/usr/local/lib/python2.7/dist-packages/theano/gpuarray/__init__.py", line 166, in init_dev unique_id = '(' + context.unique_id + ')' File "pygpu/gpuarray.pyx", line 1102, in pygpu.gpuarray.GpuContext.unique_id.__get__ File "pygpu/gpuarray.pyx", line 487, in pygpu.gpuarray.ctx_property GpuArrayException: cuDeviceGetPCIBusId((char *)res, 13, id): CUDA_ERROR_INVALID_VALUE: invalid argument [Elemwise{exp,no_inplace}(<TensorType(float32, vector)>), GpuFromHost<None>(Elemwise{exp,no_inplace}.0)]
Looking at the gpuarray code, I found that the line where this error occurred was trying to print the active device configuration. I tried setting
print_active_device=false
as part of my THEANO_FLAGS and interestingly the code worked and I was able to monitor with nvidia-smi that the code used the gpu.
I was wondering if there is some issue when the PCI BUSID is over a certain length, the busID on my machine is 17 characters in total.
Any ideas ?
Thanks.
The text was updated successfully, but these errors were encountered: