Task #3171
schedule CPU H2D force contribution in separate stream
Description
In order to be able to overlap the H2D copy of CPU force contribution with compute kernels running on the GPU, the copy needs to be scheduled in a different stream than the local/nonlocal NB streams where it it done currently. In the current setup, unless PME completes significantly later than the PP tasks, most/all H2D time ends up non-overlapped.
Related issues
History
#1 Updated by Szilárd Páll 29 days ago
- Related to Feature #2888: CUDA Update and Constraints module added
#2 Updated by Szilárd Páll 29 days ago
- Related to Feature #2817: GPU X/F buffer ops added
#3 Updated by Szilárd Páll 29 days ago
- Related to Feature #2890: GPU Halo Exchange added
#4 Updated by Szilárd Páll 5 days ago
- Related to Task #3170: investigate GPU f buffer ops use cases added