Cherry-pick: [ROCm] Fix bf16 upcast handling for libdevice calls.#793
Open
magaonka-amd wants to merge 1 commit intoROCm:rocm-jaxlib-v0.9.2from
Open
Cherry-pick: [ROCm] Fix bf16 upcast handling for libdevice calls.#793magaonka-amd wants to merge 1 commit intoROCm:rocm-jaxlib-v0.9.2from
magaonka-amd wants to merge 1 commit intoROCm:rocm-jaxlib-v0.9.2from
Conversation
Imported from GitHub PR openxla#39871 📝 Summary of Changes arguments always up/down casted for bf16 - there are no native libdevice functions for bf16 🎯 Justification with current implementation up/down cast are omitted for bf16, which is not correct 🚀 Kind of Contribution Please remove what does not apply: 🐛 Bug Fix 🧪 Tests 📊 Benchmark (for Performance Improvements) Please measure and include speedups for one of the public HLOs in `compiler/xla/tools/benchmarks/hlo/`. 🧪 Unit Tests: Added new test: //xla/backends/gpu/codegen/triton/transforms/tests:triton_xla_math_to_libdevice_rocm.mlir 🧪 Execution Tests: What execution tests were added? For example, a new optimization should be tested with an end-to-end execution test triggering the optimization and asserting correctness. Please provide test cases running with at most 2 GPUs. Copybara import of the project: -- b197577 by Zoran Jovanovic <zjovanov@amd.com>: [ROCm] Fix bf16 upcast handling for libdevice calls. -- e537018 by Zoran Jovanovic <zjovanov@amd.com>: Review comments. -- 76e617e by Zoran Jovanovic <zjovanov@amd.com>: Rmoved unnecessary file inclusion. Merging this change closes openxla#39871 COPYBARA_INTEGRATE_REVIEW=openxla#39871 from ROCm:rocm-fix-bf16-libdevice-upcast-3 76e617e PiperOrigin-RevId: 892211779
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
2 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.
Imported from GitHub PR openxla#39871
📝 Summary of Changes
arguments always up/down casted for bf16 - there are no native libdevice functions for bf16
🎯 Justification
with current implementation up/down cast are omitted for bf16, which is not correct
🚀 Kind of Contribution
Please remove what does not apply: 🐛 Bug Fix 🧪 Tests