[release/2.12] Pin torchaudio to ROCm/audio release/2.11.0.1 - #3382
[release/2.12] Pin torchaudio to ROCm/audio release/2.11.0.1#3382ethanwee1 wants to merge 1 commit into
Conversation
|
TheRock test build (Windows, gfx110X-all) for this torchaudio pin: Inputs:
Verifies torchaudio resolves from |
|
Jenkins build for 397caab617f3b3beb91836468259cc663e1210b9 commit finished as NOT_BUILT |
|
This PR now contains two commits:
New TheRock test build (Windows, gfx110X-all) with both changes: Previous build (pin only, failed on the cross-drive bug now fixed): https://github.com/ROCm/TheRock/actions/runs/28466063021 |
|
Jenkins build for 212d846a9952acce94eefcb7b3bcf9a69225e841 commit finished as NOT_BUILT |
Build status updateTheRock test build (Windows, gfx110X-all): https://github.com/ROCm/TheRock/actions/runs/28474577213 — FAILED (in the torchaudio compile), but it got materially further than the prior run. Changes in this PR (3 commits) and what each achieved:
Why the build still fails (out of scope for this PR): This is not a one-line header fix — behind it, Disposition: the GPU port is being handed to the torchaudio/audio team. This PR stands on its own as (1) the correct torchaudio version pin and (2) a general Windows cross-drive HIPIFY fix, both build-validated up to the torchaudio ROCm GPU-port gap. |
|
Jenkins build for 212d846a9952acce94eefcb7b3bcf9a69225e841 commit finished as FAILURE |
|
Jenkins build for 74bce3687d4eacf122f181b438885d907a49c1cb commit finished as NOT_BUILT |
https://amd-hub.atlassian.net/browse/ROCM-24803?focusedCommentId=500051 Interim unblock: pins torchaudio to the stable release/2.11 (2.11.0), which already builds and publishes on Windows, replacing the floating pytorch/audio main (2.11.0a0) pin. The torchaudio 2.11.0.1 ROCm GPU port is tracked separately in #3382 and ROCm/audio#16.
|
Reworked to the HIPIFY-based torchaudio port (ROCm/audio#17, mirrors torchvision) instead of the manual compat shim. New build: https://github.com/ROCm/TheRock/actions/runs/28538325024 |
|
Jenkins build for 74bce3687d4eacf122f181b438885d907a49c1cb commit finished as FAILURE |
|
Repointed Where the HIPIFY fixes live:
Prior passing TheRock run with this pin combo: https://github.com/ROCm/TheRock/actions/runs/28529085509 |
…yle) Mirror torchvision's setup.py approach: run hipify_python.hipify() over libtorchaudio CUDA sources and headers, then compile the generated .hip outputs. Fixes vs the first HIPIFY attempt (TheRock 28538325024): 1. Do not widen `#ifdef USE_CUDA` to `USE_ROCM` in headers — that pulled `cuda_fp16.h` into CPU `.cpp` compiles because PyTorch defines USE_ROCM globally on ROCm extension builds. Keep upstream USE_CUDA guards; pass `-DUSE_CUDA` on hipcc/nvcc only for GPU translation units. 2. Hipify with `os.path.realpath()` paths and depth-aware include globs so headers like `cuda_utils.h` are rewritten on the same inode Windows runners use when B: is a subst alias of the C: checkout (otherwise the compiler reads un-hipified headers from the underlying path). 3. Include `cuctc/` in the hipify scope. Reference: https://github.com/ROCm/vision/blob/release/0.27/setup.py Paired with ROCm/pytorch#3382 (torchaudio pin + Windows cross-drive HIPIFY fix in torch/utils/cpp_extension.py — that fix is in PyTorch, not here).
|
Pin updated to audio HIPIFY v2 |
|
Jenkins build for 20ed9fc17d7e46e02019b8f652ca8037fcac073e commit finished as FAILURE |
|
Jenkins build for 7b07f69b1ff1b8b60426138073ec3163051217e9 commit finished as FAILURE |
|
Pin now on HIPIFY v3 (ROCm/audio#17 |
…header + guards) Addresses review feedback (jithunnair-amd) to not invoke hipify manually in the torchaudio setup file. Instead this ports libtorchaudio to ROCm via fork source updates only, relying on the built-in CUDAExtension hipify for the .cu source bodies: - cuda_compat.h maps the CUDA runtime symbols used in headers to their HIP equivalents under USE_ROCM. - Widen the rnnt/forced_align GPU #ifdef USE_CUDA guards, macros.h, and Options::stream_ to also cover USE_ROCM. - Route cub through hipcub under USE_ROCM. The Windows B:/C: subst cross-drive HIPIFY fix lives in ROCm/pytorch#3382 (torch/utils/cpp_extension.py).
|
Jenkins build for 19ca05d9a4931735adf4fddbbea53de643810edf commit finished as FAILURE |
…header + guards) Addresses review feedback (jithunnair-amd) to not invoke hipify manually in the torchaudio setup file. Instead this ports libtorchaudio to ROCm via fork source updates only, relying on the built-in CUDAExtension hipify for the .cu source bodies: - cuda_compat.h maps the CUDA runtime symbols used in headers to their HIP equivalents under USE_ROCM. - Widen the rnnt/forced_align GPU #ifdef USE_CUDA guards, macros.h, and Options::stream_ to also cover USE_ROCM. - Route cub through hipcub under USE_ROCM. The Windows B:/C: subst cross-drive HIPIFY fix lives in ROCm/pytorch#3382 (torch/utils/cpp_extension.py).
Point related_commits at the source-only HIP port (ROCm/audio#17) on release/2.11.0.1. Uses ethanwee1/audio fork until #17 merges; then flip origin to https://github.com/ROCm/audio. Companion: ROCm/audio#17, ROCm/pytorch cpp_extension fix (separate PR). ROCM-24803
19ca05d to
9aebea9
Compare
|
Rebased stack ready for review. TheRock Windows gfx110X-all validation build (pin + Run: https://github.com/ROCm/TheRock/actions/runs/31709842398
PR split (ROCM-24803):
cc @jithunnair-amd @pablo-garay for review |
|
Build failure analysis for run 31709842398: Not a PyTorch/torchaudio code issue. The build died in Install requirements before checkout/build started: The Redispatch using Pablo's ninja-tolerant workflow branch (same fix as users/garayp/win-ninja-tolerant) + the ROCm version from the July passing build: New run: https://github.com/ROCm/TheRock/actions/runs/31711078734
|
https://amd-hub.atlassian.net/browse/ROCM-24803
Summary
Rebased onto current
release/2.12. Pinsrelated_commitsto the source-only ROCm GPU port for torchaudio 2.11.0.1 (ROCm/audio#17 @a0da46b3).The Windows
CUDAExtensioncross-drive HIPIFY fix is split out to ROCm/pytorch#3562 per review feedback.related_commits
Uses
ethanwee1/audiofork URL until #17 merges; then flip origin tohttps://github.com/ROCm/audio.Validation
TheRock Windows gfx110X-all build dispatched on combined branch
ew/rocm-24803-validation(this pin + #3562).Companion PRs
cuda_compat.h, guard widening, hipcub)cpp_extension.pyrealpath fix