Project

General

Profile

Task #3128

Feature #2816: GPU offload / optimization for update&constraits, buffer ops and multi-gpu communication

Feature #2817: GPU X/F buffer ops

Feature #3029: GPU force buffer ops + reduction

do not fall back to CPU path on energy-only steps

Added by Szilárd Páll about 1 month ago. Updated 19 days ago.

Status:
Closed
Priority:
High
Assignee:
Category:
-
Target version:
Difficulty:
uncategorized
Close

Description

Current code does:

useGpuFBufferOps = simulationWork.useGpuBufferOps && simulationWork.useGpuNonbonded &&
        !(flags.computeVirial || flags.computeEnergy);

This energy steps have no reason to fall back to CPU buffer ops so that condition needs to be removed.

Associated revisions

Revision 1b346f03 (diff)
Added by Szilárd Páll about 1 month ago

Make the wait on nonbonded GPU results conditional

When the force reduction is done on the GPU and there are no energy or
shift force results required, there is no need to block and wait on the
CPU until the GPU nonbonded kernels complete.
This change makes the wait conditional on whether there are nonbonded
force, energy or shift force outputs so the blocking wait is now skipped
with GPU buffer ops on force-only steps.

Also removed the now unnecessary boolean argument passed to
gpu_launch_cpyback().

Refs #3128

Change-Id: Ic1285f5a00ac910cd1d6c4358f41f2c7c41dea4c

Revision a202c765 (diff)
Added by Alan Gray 19 days ago

Remove energy step dependence on GPU force buffer ops

Energy steps have no reason to fall back to CPU buffer ops. This
change removes the energy step flag from the calculation of the GPU
force buffer op flag.

Addresses #3128

Change-Id: Ie65a919f3bae4aec7374627bbb7e510d5a09b6c5

History

#1 Updated by Szilárd Páll 20 days ago

  • Status changed from New to In Progress

#2 Updated by Paul Bauer 19 days ago

  • Status changed from In Progress to Resolved

#3 Updated by Paul Bauer 19 days ago

  • Status changed from Resolved to Closed

Also available in: Atom PDF