TridiagSolver (dist): STEP2b Make rank1 work just on local non-deflated eigenvectors (multi-threaded) #997
Conversation
This file contains 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
albestro
changed the title
albestro
force-pushed
the
alby/trisolver-dist-opt-step2-multi
branch
from
a36baa8
to
05d4760
Compare
albestro
force-pushed
the
alby/trisolver-dist-opt-step2
branch
from
c22b56f
to
42d6681
Compare
albestro
added a commit
that referenced
this pull request
Oct 16, 2023
Develop: TridiagSolver (dist): STEP1 rank-independent sort of eigenvalues by column type for rank1 solver (#967) Develop: TridiagSolver (dist): STEP2 Make rank1 work just on local non-deflated eigenvectors (#996) Develop: TridiagSolver (dist): STEP2b Make rank1 work just on local non-deflated eigenvectors (multi-threaded) (#997) Develop: TridiagSolver (dist): STEP3 reduce GEMM step computational cost (#998)
albestro
force-pushed
the
alby/trisolver-dist-opt-step2
branch
from
42d6681
to
477b2e2
Compare
albestro
force-pushed
the
alby/trisolver-dist-opt-step2-multi
branch
from
05d4760
to
fffad13
Compare
albestro
force-pushed
the
alby/trisolver-dist-opt-step2
branch
from
477b2e2
to
0d17094
Compare
albestro
force-pushed
the
alby/trisolver-dist-opt-step2-multi
branch
from
fffad13
to
a538c2e
Compare
albestro
commented
Dec 12, 2023
Comment on lines
+1279
to
+1288
| const std::size_t nthreads = [dist_sub, k_lc] { | ||
| const std::size_t workload = to_sizet(dist_sub.localSize().rows() * k_lc); | ||
| const std::size_t workload_unit = 2 * to_sizet(dist_sub.tile_size().linear_size()); | ||
|
|
||
| const std::size_t min_workers = 1; | ||
| const std::size_t available_workers = getTridiagRank1NWorkers(); | ||
|
|
||
| const std::size_t ideal_workers = util::ceilDiv(to_sizet(workload), workload_unit); | ||
| return std::clamp(ideal_workers, min_workers, available_workers); | ||
| }(); |
There was a problem hiding this comment.
Just pointing out the logic, which might deserve a comment/note.
Number of "ideal" threads is calculated dynamically as a function of how much work rank1-solver has to do. The work is measured in terms of local problem size and a thread should have at least 2 tiles to work on.
This ideal number of threads is then used to determine how many actual threads can be used, by checking the range defined by configuration parameters (at least 1, at most config parameter).
albestro
force-pushed
the
alby/trisolver-dist-opt-step2-multi
branch
from
6c4a7b5
to
ec9ebf6
Compare
4 tasks
msimberg
approved these changes
Dec 12, 2023
rasolca
approved these changes
Dec 12, 2023
|
cscs-ci run |
Codecov ReportAttention:
❗ Your organization needs to install the Codecov GitHub app to enable full functionality. Additional details and impacted files@@ Coverage Diff @@
## master #997 +/- ##
==========================================
+ Coverage 93.98% 94.04% +0.05%
==========================================
Files 145 145
Lines 9034 9050 +16
Branches 1159 1157 -2
==========================================
+ Hits 8491 8511 +20
Misses 321 321
+ Partials 222 218 -4 ☔ View full report in Codecov by Sentry. |
github-actions bot
pushed a commit
that referenced
this pull request
Dec 12, 2023
…eflated eigenvectors (multi-threaded) (#997)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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 makes the changes from #996 multithreaded.
Main changes:
Suggestion for reviewers: turn on "hide whitespace" feature (due to the different indentation introduced with an additional nesting for dynamic nthreads).
TODO: