-
Notifications
You must be signed in to change notification settings - Fork 3
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Log here errors related to running Gromacs with OpenCL enabled #44
Comments
Underflow error in put_atoms_in_box(...). It blocks either in the first or in the second while loop:
It happens after > 50 kernel executions. NVIDIA 660M GTX, Windows 7 |
On NVIDIA 660M GTX, it runs through the end, no errors, no hanging, no crashes if configuring gmx to be single-threaded:
|
On I7-3610 QM Intel CPU it runs through the end, no errors, no hanging, no crashes if configuring gmx to be single-threaded:
and if explicitly adding barrier(CLK_LOCAL_MEM_FENCE); everywhere where #if WARP_SIZE_TEST==1 is present. (The #if WARP_SIZE_TEST == 1 path is not active as the WARP_SIZE_TEST is 128). Moreover, the output is the same as the output of running the OpenCL implementation on NVIDIA 660M GTX. Unfortunately, these results are different from those generated by the CUDA implementation. Running gmx check with -abstol 1 for the energies shows 13 differences: Reading energy frame 0 time 5.000 Bond step 5000: |
Specify the error (log error/execution error/crash, etc) and also details about the OpenCL device in use.
The text was updated successfully, but these errors were encountered: