Skip to content

update yaml for some gfx941 HHS sizes - #372

Merged
KKyang merged 1 commit into
ROCm:developfrom
jichangjichang:update_gfx941_hhs_yaml_1009
Oct 9, 2023
Merged

KKyang merged 1 commit into
ROCm:developfrom
jichangjichang:update_gfx941_hhs_yaml_1009

Conversation

@jichangjichang

Copy link
Copy Markdown
Collaborator

No description provided.

@jichangjichang

Copy link
Copy Markdown
Collaborator Author

tune 2 more sizes for 941

@KKyang
KKyang merged commit d18cae4 into ROCm:develop Oct 9, 2023
assistant-librarian Bot pushed a commit that referenced this pull request Jul 10, 2025
Increase the coverage from 53% to 57% (500+ lines) (#372)

- This time we cover mainly rocblaslt_auxiliary.cpp and auxiliary.cpp
![image (12)
(1)](https://github.com/user-attachments/assets/3217b0e0-3758-40ab-b8d9-8f3bfb1a65a6)

- The total time cost:
**Discard the test code of testing_aux_mat_copy and
testing_aux_auxiliary_func, and both of them show 0 ms as follows.**

[ RUN      ] _/aux_test.conversion/pre_checkin_aux_mat_copy_f16_r
[       OK ] _/aux_test.conversion/pre_checkin_aux_mat_copy_f16_r (0 ms)
[ RUN      ] _/aux_test.conversion/pre_checkin_aux_auxiliary_func_f16_r
[ OK ] _/aux_test.conversion/pre_checkin_aux_auxiliary_func_f16_r (0 ms)
[----------] 32 tests from _/aux_test (9309 ms total)
[----------] 32 tests from _/aux_test (8970 ms total)
[----------] 32 tests from _/aux_test (8746 ms total)
[----------] 32 tests from _/aux_test (8976 ms total)
[----------] 32 tests from _/aux_test (11416 ms total)

**With the test code:**

[ RUN      ] _/aux_test.conversion/pre_checkin_aux_mat_copy_f16_r
[       OK ] _/aux_test.conversion/pre_checkin_aux_mat_copy_f16_r (0 ms)
[ RUN      ] _/aux_test.conversion/pre_checkin_aux_auxiliary_func_f16_r
[ OK ] _/aux_test.conversion/pre_checkin_aux_auxiliary_func_f16_r (0 ms)
[----------] 32 tests from _/aux_test (13088 ms total)
[ RUN      ] _/aux_test.conversion/pre_checkin_aux_mat_copy_f16_r
[       OK ] _/aux_test.conversion/pre_checkin_aux_mat_copy_f16_r (0 ms)
[ RUN      ] _/aux_test.conversion/pre_checkin_aux_auxiliary_func_f16_r
[ OK ] _/aux_test.conversion/pre_checkin_aux_auxiliary_func_f16_r (0 ms)
[----------] 32 tests from _/aux_test (10484 ms total)
[ RUN      ] _/aux_test.conversion/pre_checkin_aux_mat_copy_f16_r
[       OK ] _/aux_test.conversion/pre_checkin_aux_mat_copy_f16_r (0 ms)
[ RUN      ] _/aux_test.conversion/pre_checkin_aux_auxiliary_func_f16_r
[ OK ] _/aux_test.conversion/pre_checkin_aux_auxiliary_func_f16_r (0 ms)
[----------] 32 tests from _/aux_test (10467 ms total)
[ RUN      ] _/aux_test.conversion/pre_checkin_aux_mat_copy_f16_r
[       OK ] _/aux_test.conversion/pre_checkin_aux_mat_copy_f16_r (0 ms)
[ RUN      ] _/aux_test.conversion/pre_checkin_aux_auxiliary_func_f16_r
[ OK ] _/aux_test.conversion/pre_checkin_aux_auxiliary_func_f16_r (0 ms)
[----------] 32 tests from _/aux_test (10965 ms total)

**Conclusion: The overhead is less than 0.xxx + 0.yyy < 2 ms.**

---------

Co-authored-by: tomchengchitang <tom.tang@amd.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants