Project

General

Profile

Bug #3231

simulationWorkload incorrect for some of the propagators

Added by Szilárd Páll about 2 months ago. Updated about 1 month ago.

Status:
Closed
Priority:
Normal
Assignee:
-
Category:
mdrun
Target version:
Affected version - extra info:
Affected version:
Difficulty:
uncategorized
Close

Description

As a side-effect of trying to improved the stateGpu management, I found that rerun runs have incorrect simulationWorkload with useGpuUpdate = true.

Associated revisions

Revision ec272c75 (diff)
Added by Artem Zhmurov about 2 months ago

Set GPU update to false in case of a rerun

Rerun does not require the numerical integration of equation of
motion, hence the useGpuUpdate should be false.

Fixes #3231.

Change-Id: I9027641e9dc32fb464ebbb6148c11c530c0eb2cd

History

#1 Updated by Mark Abraham about 2 months ago

Does that cause a problem? do_rerun doesn't call any update code, although of course it calls do_force which acts upon the simulationWorkload flags. do_rerun does need to arrange for data transfers so it can write the expected output, but that should be managed explicitly, and not as a side effect of whether useGpuUpdate has any particular value.

#2 Updated by Artem Zhmurov about 2 months ago

  • Status changed from New to Resolved

#3 Updated by Paul Bauer about 1 month ago

  • Target version set to 2020-rc1

#4 Updated by Paul Bauer about 1 month ago

  • Status changed from Resolved to Closed

Also available in: Atom PDF