[executorch][arm] Add bfloat16 support to VGF backend#20755
Closed
SS-JIA wants to merge 2 commits into
Closed
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20755
Note: Links to docs will display an error until the docs builds have been completed. ❌ 9 New Failures, 1 Cancelled JobAs of commit 619fa2e with merge base 5a691cb ( NEW FAILURES - The following jobs have failed:
CANCELLED JOB - The following job was cancelled. Please retry:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
Bot
added
the
CLA Signed
This PR needs a
|
SS-JIA
added a commit
that referenced
this pull request
Jul 6, 2026
Pull Request resolved: #20755 Re-land bf16 support for the Arm VGF backend, restoring the two GitHub PRs that were reverted (#20299 and #20660). bf16 support needs the `VK_FORMAT_R16_SFLOAT_FPENCODING_BFLOAT16_ARM` `VkFormat`, which was added in Vulkan headers 1.4.350. The internal headers were pinned to 1.4.341, so the original diff-train import of #20299 (D110192953) failed to compile internally and both PRs were reverted in OSS. The Vulkan 1.4.350 header import and the `//third-party/khronos` selector wiring are the diffs below this one in the stack; this diff is the backend code that depends on them. Changes: - #20299: handle bf16 in `VGFSetup.cpp` (new `FormatScalarKind::BFloat` plus `VK_FORMAT_R16_SFLOAT_FPENCODING_BFLOAT16_ARM` cases in `get_format_component_count`, `get_format_info`, and `get_format_size`), the VGF test infrastructure (`test_pipeline.py`, `runner_utils.py`, `test_matmul.py`), and the `vulkan_utils.sh` SDK bump for OSS parity. - #20660: extend bf16 test coverage across the VGF op and model tests. All source and test changes are mirrored across the `fbcode/` and `xplat/` copies of the Arm backend. ghstack-source-id: 400367995 @exported-using-ghexport Differential Revision: [D110538232](https://our.internmc.facebook.com/intern/diff/D110538232/)
rascani
approved these changes
Jul 6, 2026
SS-JIA
added a commit
that referenced
this pull request
Jul 7, 2026
Pull Request resolved: #20755 Re-land bf16 support for the Arm VGF backend, restoring the two GitHub PRs that were reverted (#20299 and #20660). bf16 support needs the `VK_FORMAT_R16_SFLOAT_FPENCODING_BFLOAT16_ARM` `VkFormat`, which was added in Vulkan headers 1.4.350. The internal headers were pinned to 1.4.341, so the original diff-train import of #20299 (D110192953) failed to compile internally and both PRs were reverted in OSS. The Vulkan 1.4.350 header import and the `//third-party/khronos` selector wiring are the diffs below this one in the stack; this diff is the backend code that depends on them. Changes: - #20299: handle bf16 in `VGFSetup.cpp` (new `FormatScalarKind::BFloat` plus `VK_FORMAT_R16_SFLOAT_FPENCODING_BFLOAT16_ARM` cases in `get_format_component_count`, `get_format_info`, and `get_format_size`), the VGF test infrastructure (`test_pipeline.py`, `runner_utils.py`, `test_matmul.py`), and the `vulkan_utils.sh` SDK bump for OSS parity. - #20660: extend bf16 test coverage across the VGF op and model tests. All source and test changes are mirrored across the `fbcode/` and `xplat/` copies of the Arm backend. ghstack-source-id: 400790603 @exported-using-ghexport Differential Revision: [D110538232](https://our.internmc.facebook.com/intern/diff/D110538232/)
Contributor
|
Thanks @SS-JIA - seems like a couple of CI issues.. |
digantdesai
approved these changes
Jul 9, 2026
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
3 participants
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.
Stack from ghstack (oldest at bottom):
Re-land bf16 support for the Arm VGF backend, restoring the two GitHub PRs that were reverted (#20299 and #20660).
bf16 support needs the
VK_FORMAT_R16_SFLOAT_FPENCODING_BFLOAT16_ARMVkFormat, which was added in Vulkan headers 1.4.350. The internal headers were pinned to 1.4.341, so the original diff-train import of #20299 (D110192953) failed to compile internally and both PRs were reverted in OSS. The Vulkan 1.4.350 header import and the//third-party/khronosselector wiring are the diffs below this one in the stack; this diff is the backend code that depends on them.Changes:
VGFSetup.cpp(newFormatScalarKind::BFloatplusVK_FORMAT_R16_SFLOAT_FPENCODING_BFLOAT16_ARMcases inget_format_component_count,get_format_info, andget_format_size), the VGF test infrastructure (test_pipeline.py,runner_utils.py,test_matmul.py), and thevulkan_utils.shSDK bump for OSS parity.All source and test changes are mirrored across the
fbcode/andxplat/copies of the Arm backend.Differential Revision: D110538232
cc @digantdesai @freddan80 @per @zingo @oscarandersson8218 @mansnils @Sebastian-Larsson @robell @rascani