Repository navigation
tf32 add origami kernels - #854
Merged
Merged
Conversation
carsonbrownlee
requested review from
AlexBrownAMD,
b-shi,
msujon-AMD and
ryanswann-amd
July 25, 2025 23:29
Contributor
Author
|
This commit sets origami kernels as only tf32 kernels. I'm not sure if there is a way to keep grid kernels but prefer origami by default. Performance appears to be within target (4.0x bf16), but still getting this warning message: |
dsaffars
pushed a commit
to dsaffars/rocm-libraries
that referenced
this pull request
Jul 28, 2025
* Move activity logging to user manual * Remove reference to logging.rst file * Add link to rocm examples github repository on main page of documentation * Add note about what version of rocprim/rocblas must be used when building rocsparse from source on windows * Add new line on mainpage to separate links to rocsparse repo and rocm examples * Add new line on mainpage to separate links to rocsparse repo and rocm examples * Add new line on mainpage to separate links to rocsparse repo and rocm examples * Add more instructions to linux installation page * Fix rocm documentation links --------- Co-authored-by: jsandham <james.sandham@amd.com> [ROCm/rocSPARSEcommit: d35758d]
AlexBrownAMD
approved these changes
Jul 28, 2025
assistant-librarian Bot
pushed a commit
to ROCm/hipBLASLt
that referenced
this pull request
Jul 28, 2025
tf32 add origami kernels removing tf32 grid based kernels adding tf32 origami kernels
carsonbrownlee
added a commit
that referenced
this pull request
Jul 29, 2025
removing tf32 grid based kernels adding tf32 origami kernels
yenong-amd
pushed a commit
that referenced
this pull request
Jul 29, 2025
removing tf32 grid based kernels adding tf32 origami kernels (cherry picked from commit 61e1136)
yenong-amd
pushed a commit
that referenced
this pull request
Aug 7, 2025
removing tf32 grid based kernels adding tf32 origami kernels (cherry picked from commit 61e1136)
ammallya
pushed a commit
that referenced
this pull request
Sep 18, 2025
* enable hash checking for sygvdx * address feedback * add support for getrf, potrf, syevx, sygvx * amend comments, update sygvdx [ROCm/rocSOLVER commit: 38c7a3a]
ammallya
pushed a commit
that referenced
this pull request
Sep 18, 2025
* Improve robustness of SYGVDX/HEGVDX tests * Preliminary changes to sygvdx's error bound computation * Enable hashing output for sygvdx (#854) * enable hash checking for sygvdx * address feedback * add support for getrf, potrf, syevx, sygvx * amend comments, update sygvdx (cherry picked from commit d8dd2d6) * Update sygvdx test to handle case in which number of eigenvalues differ This commit updates sygvdx's test to allow it to gracefully handle the case in which the number of eigenvalues computed in rocSOLVER differ from the number of reference eigenvalues. * Refactor sygvdx-hegvdx tests This commit updates the error bounds and fixes bugs in the sygvdx-hegvdx tests. It also adds code to allow selecting whether the tests should enforce equality on the number of reference and computed eivenvalues (default is yes); and whether the tests should use new or legacy error bounds (default is to use new error bounds). * Add documentation and small improvements to code * Update comments * Update `clss::print_debug()` method * Make sygvdx-hegvdx tests slightly more strict * Apply clang-format * Implement suggestions and improve comments * Fix typos * Fix typo * Add minor fixes and improvements * Add minor clarifications * Add missing license and update dates on modified files * Address review comments --------- Co-authored-by: Jonah Quist <jonquist@amd.com> [ROCm/rocSOLVER commit: b94dba0]
mgates3
pushed a commit
to mgates3/rocm-libraries
that referenced
this pull request
Mar 6, 2026
* enable hash checking for sygvdx * address feedback * add support for getrf, potrf, syevx, sygvx * amend comments, update sygvdx
mgates3
pushed a commit
to mgates3/rocm-libraries
that referenced
this pull request
Mar 6, 2026
* Improve robustness of SYGVDX/HEGVDX tests * Preliminary changes to sygvdx's error bound computation * Enable hashing output for sygvdx (ROCm#854) * enable hash checking for sygvdx * address feedback * add support for getrf, potrf, syevx, sygvx * amend comments, update sygvdx (cherry picked from commit 161dc64) * Update sygvdx test to handle case in which number of eigenvalues differ This commit updates sygvdx's test to allow it to gracefully handle the case in which the number of eigenvalues computed in rocSOLVER differ from the number of reference eigenvalues. * Refactor sygvdx-hegvdx tests This commit updates the error bounds and fixes bugs in the sygvdx-hegvdx tests. It also adds code to allow selecting whether the tests should enforce equality on the number of reference and computed eivenvalues (default is yes); and whether the tests should use new or legacy error bounds (default is to use new error bounds). * Add documentation and small improvements to code * Update comments * Update `clss::print_debug()` method * Make sygvdx-hegvdx tests slightly more strict * Apply clang-format * Implement suggestions and improve comments * Fix typos * Fix typo * Add minor fixes and improvements * Add minor clarifications * Add missing license and update dates on modified files * Address review comments --------- Co-authored-by: Jonah Quist <jonquist@amd.com>
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
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.
removing tf32 grid based kernels
adding tf32 origami kernels