Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Tune Aquavanjaram942 HHS NN TN NT GEMM sizes equality and grid based #1184

Open
wants to merge 2 commits into
base: develop
Choose a base branch
from
Open

Tune Aquavanjaram942 HHS NN TN NT GEMM sizes equality and grid based #1184

wants to merge 2 commits into from
+4,530 −2,012

Conversation

aferoz21
Copy link
Contributor

  1. Tuned HHS TN, NN, NT GEMM sizes with ~17% improvement.
  2. 2 Grid point sizes updated and 8 equality size kernels added.

@aferoz21 aferoz21 added noCI Disable testing on supported CI systems: math libraries CI has this feature enabled.. gfx94x Run CI on gfx94x labels Sep 30, 2024
@aferoz21
Copy link
Contributor Author

hiPBLAS-test results

[----------] 1 test from ExtOpTest/ExtOpAMaxUnsupportedDatatypeTest
[ RUN ] ExtOpTest/ExtOpAMaxUnsupportedDatatypeTest.amaxFailureUnsupportedDatatype/0
[ OK ] ExtOpTest/ExtOpAMaxUnsupportedDatatypeTest.amaxFailureUnsupportedDatatype/0 (0 ms)
[----------] 1 test from ExtOpTest/ExtOpAMaxUnsupportedDatatypeTest (0 ms total)

[----------] 1 test from ExtOpTest/ExtOpAMaxWithScaleUnsupportedDatatypeTest
[ RUN ] ExtOpTest/ExtOpAMaxWithScaleUnsupportedDatatypeTest.amaxWithScaleFailureUnsupportedDatatype/0
[ OK ] ExtOpTest/ExtOpAMaxWithScaleUnsupportedDatatypeTest.amaxWithScaleFailureUnsupportedDatatype/0 (0 ms)
[----------] 1 test from ExtOpTest/ExtOpAMaxWithScaleUnsupportedDatatypeTest (0 ms total)

[----------] Global test environment tear-down
[==========] 48206 tests from 13 test suites ran. (1419771 ms total)
[ PASSED ] 48206 tests.
hipBLASLt version: 1000

KKyang
KKyang previously approved these changes Oct 1, 2024
View reviewed changes
@aferoz21
Copy link
Contributor Author

aferoz21 commented Oct 1, 2024

Thanks Koji ! added 1 more grid size kernel that was missed. Also verify log updated with trig_float init.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@KKyang KKyang KKyang left review comments

@jichangjichang jichangjichang Awaiting requested review from jichangjichang jichangjichang is a code owner

@aazz44ss aazz44ss Awaiting requested review from aazz44ss aazz44ss is a code owner

@vin-huang vin-huang Awaiting requested review from vin-huang vin-huang is a code owner

@imcarsonliao imcarsonliao Awaiting requested review from imcarsonliao imcarsonliao is a code owner

@hcman2 hcman2 Awaiting requested review from hcman2 hcman2 is a code owner

@Serge45 Serge45 Awaiting requested review from Serge45 Serge45 is a code owner

@Jinp800125 Jinp800125 Awaiting requested review from Jinp800125 Jinp800125 is a code owner

@TonyYHsieh TonyYHsieh Awaiting requested review from TonyYHsieh TonyYHsieh is a code owner

@solaslin solaslin Awaiting requested review from solaslin solaslin is a code owner

At least 2 approving reviews are required to merge this pull request.

Assignees
No one assigned
Labels
gfx94x Run CI on gfx94x noCI Disable testing on supported CI systems: math libraries CI has this feature enabled..
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@aferoz21 @KKyang