hip: Add Linux hip build - #159
Merged
Merged
Conversation
Owner
|
@asikul81316-star do you have comments for this PR? |
Owner
|
@IIIIIllllIIIIIlllll I'm going to merge the PR if @asikul81316-star doesn't have any comments in the next few hours... |
Owner
|
@IIIIIllllIIIIIlllll One minor issue. Is AUDIOCPP_DEPLOYMENT_BUILD=ON a requirement for HIP? If so, could you scope it to HIP instead of changing the global default? |
IIIIIllllIIIIIlllll
force-pushed
the
linux-hip-build
branch
from
July 31, 2026 01:06
562e63d to
df43de9
Compare
- '--backend hip' (alias 'rocm'), plus explicit '--hip on|off' and '--gpu-targets' (comma or semicolon separated) - ROCm detection prefers ROCM_PATH/HIP_PATH, falls back to /opt/rocm and hipconfig (covers split layouts such as core-7.14) - GPU target auto-detection via amdgpu-arch, with a strict rocminfo 'Name: gfxXXX' fallback (rejects fragments like 'gfx11') - CUDA, Vulkan, and HIP are mutually exclusive; HIP builds select the ROCm clang and clear the sticky GPU_BUILD_TARGETS/AMDGPU_TARGETS cache entries so repeated configures honor new --gpu-targets - Default parallel jobs now follow nproc instead of a hardcoded 8
HIP builds default to embedding package specs so the produced binaries run standalone outside the repo checkout; other backends keep the historical OFF default. --deployment-build / --no-deployment-build override the default explicitly.
IIIIIllllIIIIIlllll
force-pushed
the
linux-hip-build
branch
from
July 31, 2026 01:12
df43de9 to
d1296c6
Compare
Contributor
Author
done |
Owner
|
Thanks @IIIIIllllIIIIIlllll! Merged. |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Hi! Following up on the Windows HIP work (#153 ), this adds HIP/ROCm support to the Linux helper script so Linux AMD users get the same one-command build experience:
--backend hip (alias rocm), with --hip on|off and --gpu-targets flags; CUDA/Vulkan/HIP are mutually exclusive
ROCm detection via ROCM_PATH/HIP_PATH, /opt/rocm, or hipconfig (handles split layouts like core-7.14 where clang lives under lib/llvm/bin)
GPU target auto-detection via amdgpu-arch, with a strict rocminfo fallback(The code for this feature might be quite extensive)
Default -j now follows nproc
About --deployment-build : I enable it by default and provide --no-deployment-build to turn it off. This way, users can just copy the built outputs after compiling, instead of having to depend on the model_specs settings in the source code.
Tested: ROCm 7.14 on Linux (gfx1151 / Strix Halo, full build of audiocpp_cli passes) .
No behavior change for existing cuda/vulkan/cpu invocations.
AI usage: Kimi k3 for translate, I reviewed and tested this script