Enable max-mixed multicg_offset_QUDA in specific situations.#93
Open
SaltyChiang wants to merge 1 commit intomilc-qcd:developfrom
Open
Enable max-mixed multicg_offset_QUDA in specific situations.#93SaltyChiang wants to merge 1 commit intomilc-qcd:developfrom
SaltyChiang wants to merge 1 commit intomilc-qcd:developfrom
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR enables the max-mixed multicg_offset_QUDA solver in specific situations. This shouldn't affect the solver precision, as it just changes the sloppy precision.
The first pseudo-fermion used for the RHMC/RHMD algorithm usually has a low precision tolerance, and residues for the MD step for all shifts are not widely distributed. In such a situation, using half-precision as the sloppy precision in QUDA has a shorter wall time to solve the solution. Because the solver doesn't need to spend too many iterations in the refinement step, the faster multishift step will cover the refinement cost, and even faster.
We perform some 512-node tests on Vista on a 192^3x384 lattice:
NOTICE This change has not been checked in production. Although theoretically it shouldn't affect the Monte Carlo Markov Chain, we still need a production test for this.