LAMMPS GPU Cuda driver error 101

nvc_get_devices returns:

Found 1 platform(s).
Using platform: NVIDIA Corporation NVIDIA CUDA Driver
CUDA Driver Version: 5.0

Device 0: “GeForce GTX 580”
Type of device: GPU
Compute capability: 2
Double precision support: Yes
Total amount of global memory: 2.99957 GB
Number of compute units/multiprocessors: ! ; 16
Number of cores: 512
Total amount of constant memory: 65536 bytes
Total amount of local/shared memory per block: 49152 bytes
Total number of registers available per block: 32768
Warp size: 32
Maximum number of threads per block: 1024
Maximum group size (# of threads per block) 1024 x 1024 x 64
Maximum item sizes (# thre! ads for each dim) 65535 x 65535 x 65535
&! nbsp; Ma ximum memory pitch: 2147483647 bytes
Texture alignment: 512 bytes
Clock rate: 1.71 GHz
Run time limit on kernels: Yes
Integrated: &nbs! p; No
Support host page-locked memory mapping: Yes
Compute mode: Default
Concurrent kernel execution: Yes
Device has ECC support enabled: No

Device 1: “GeForce GTX 580”
Type of device: GPU
Compute capability: ! ; &! nbsp;&nb sp; 2
Double precision support: Yes
Total amount of global memory: 2.99969 GB
Number of compute units/multiprocessors: 16
Number of cores: 512
Total amount of constant memory: 65536 bytes
Total amount of local/shared memory per block: 49152 bytes
Total number of registers available per block: 32768
Warp size: ! ; 32
Maximum number of threads per block: 1024
Maximum group size (# of threads per block) 1024 x 1024 x 64
Maximum item sizes (# threads for each dim) 65535 x 65535 x 65535
Maximum memory pitch: 2147483647 bytes
Texture alignment: 512 bytes
Clock rate: &nb! sp; ! ; & nbsp; 1.71 GHz
Run time limit on kernels: No
Integrated: No
Support host page-locked memory mapping: Yes
Compute mode: Default
Concurrent kernel execution: Yes
Device has ECC support enabled: ! No

launch command, while in the bench folder is:

mpirun -np 12 …/…/src/lmp_openmpi -sf gpu -c off -v g 2 -v x 1 -v y 1 -v z 1 -v t 1000 < in.rhodo.gpu

Without changing anything but -np 12 to -np 1 it works. I then went back to -np 12 and it is working now. I don’t know why this is the case and it seems almost haphazard that it is working now when before it wasn’t. Do I need to run nvc_get_devices each time on startup to prevent the error? Or maybe the problem is between the chair and the keyboard. If the problem crops up again should I continue this thread or make a new one?

Thank you for your time,

David W. Caldwell II

I don't have any good explanation for the error you were seeing. I will be testing with CUDA 5 extensively as we prepare for the titan upgrade. If I notice any issues that need to be fixed, I will let you know. Thanks. - Mike

FYI: I am able to reproduce this issue with CUDA 5.0 release candidate, so it is not a problem with your local setup.

Will update when i know whether the change will be in cuda or the lammps code...

- Mike