Skip to content

128/256 threads/block #92

Description

@ancahamuraru

Update the OpenCL kernel for 128/256 threads/block based on the equivalent CUDA kernel - see commit f2b9db2 from the main Gromacs master branch: gromacs@f2b9db2

Evaluate the performance of the new kernel for AMD and NVIDIA GPUs and decide on the final version or versions of the OpenCL kernel that will be used.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions