Hi guys,
I got this error running eddy_cuda:
...................Allocated GPU # 0...................
thrust::system_error thrown in CudaVolume::common_assignment_from_newimage_vol after resize() with message: function_attributes(): after cudaFuncGetAttributes: invalid device function
terminate called after throwing an instance of 'thrust::system::system_error'
what(): function_attributes(): after cudaFuncGetAttributes: invalid device function
Aborted (core dumped)
I tried running the eddy_cuda that comes with version 5.0.11 with cuda 7.5 installed and updated nvidia drivers. I retried with replacing that eddy_cuda with eddy_cuda8.0 and installing cuda 8.0 instead. Same error.
I'm running this on Amazon Web Services
OS: Ubuntu 16.04
Video card: Tesla V100
FSL version: 5.0.11
NVIDIA drivers: 396
CUDA version: tried both 7.5 (with default eddy_cuda) and 8.0 (downloaded eddy_cuda8.0)
From an archived post (https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=fsl;b7d1c314.1711), this occurs due to wrong instructions for particular GPU architecture. "The solution is to compile a non-fat binary that includes only the instructions that are relevant for your GPU." Do you know if this is possible or if it already exists for the Tesla V100 GPU?
Finally - anyway to exploit multiple GPUs to accelerate pre-processing even further?
Many, many thanks!
-Sumit
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
|