Skip to content

Addition of ger batched and strided batched functions#674

Merged
mahmoodw merged 10 commits into
ROCm:developfrom
mahmoodw:gerBatchFunctions
Sep 9, 2019
Merged

Addition of ger batched and strided batched functions#674
mahmoodw merged 10 commits into
ROCm:developfrom
mahmoodw:gerBatchFunctions

Conversation

@mahmoodw
Copy link
Copy Markdown
Contributor

@mahmoodw mahmoodw commented Aug 27, 2019 •
edited
Loading

  • This is the first implementation of the ger batched and strided batched functions.
  • Changed gemv tests to test API for quick return
  • The associated tests add 1 min to the pre_checkin/quick and almost 30 min to nightlys - please advise about coverage if this is an issue. (Before BLIS)

@mahmoodw mahmoodw requested a review from amcamd as a code owner August 27, 2019 18:23
jzuniga-amd
jzuniga-amd reviewed Aug 29, 2019
View reviewed changes
Comment thread library/include/rocblas-functions.h Outdated
Comment thread library/src/blas2/rocblas_ger.cpp
Comment thread library/src/blas2/rocblas_ger_batched.cpp
Comment thread library/src/blas2/rocblas_ger_strided_batched.cpp
Comment thread library/src/blas2/rocblas_ger_batched.hpp Outdated
Comment thread library/src/blas2/rocblas_ger_batched.hpp
Comment thread library/src/blas2/rocblas_ger_batched.hpp
Comment thread library/src/blas2/rocblas_ger_strided_batched.hpp
Comment thread library/src/blas2/rocblas_ger_strided_batched.hpp
@mahmoodw
Copy link
Copy Markdown
Contributor Author

mahmoodw commented Sep 4, 2019

All tests are passing on Vega 10 and 20. Working on reducing nightlys test time.

@mahmoodw
Copy link
Copy Markdown
Contributor Author

mahmoodw commented Sep 5, 2019

Vega20 times
Ger_batched test times:

  • quick: 8.2 s
  • pre_checkin: 34.4 s
  • nightly: 511.4 s
    Ger_strided_batched test times:
  • quick: 9 s
  • pre_checkin: 74.4 s
  • nightly: 534

These times are now within the suggested timing ranges. Please review and advise.

amcamd
amcamd approved these changes Sep 5, 2019
View reviewed changes
Copy link
Copy Markdown
Contributor

@amcamd amcamd left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

TorreZuk
TorreZuk reviewed Sep 5, 2019
View reviewed changes
Comment thread clients/include/testing_ger_strided_batched.hpp Outdated
@mahmoodw mahmoodw merged commit 47b5d45 into ROCm:develop Sep 9, 2019
mlse-lib-jenkins pushed a commit that referenced this pull request May 14, 2021
@eidenyoshida
Add address sanitizer build option (#674)
8e1015c
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@TorreZuk TorreZuk TorreZuk left review comments

@jzuniga-amd jzuniga-amd jzuniga-amd left review comments

@amcamd amcamd amcamd approved these changes

@leekillough leekillough Awaiting requested review from leekillough

@zaliu zaliu Awaiting requested review from zaliu

@daineAMD daineAMD Awaiting requested review from daineAMD

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

5 participants

@mahmoodw @leekillough @amcamd @TorreZuk @jzuniga-amd