Fixes to support HexKL 1.0.beta.2 on Kaanapali - #100
Open
arunrangasa wants to merge 9 commits into
Open
arunrangasa wants to merge 9 commits into
arunrangasa wants to merge 9 commits into
Conversation
arunrangasa
commented
Sep 21, 2026
arunrangasa
commented
Contributor
- Update HexKL Lib Path; bail out if libraries are not found in the expected location
- Use new names for Macro API Functions
- Set RPC timeout=1800s conservatively to give a chance for long running tests to complete
arunrangasa
force-pushed
the
UpdateHexKLTorchVision
branch
from
September 21, 2026 09:50
66fb290 to
e45f347
Compare
…lcomm#89) Bump dependencies: - triton: df38505e -> a9ced836 (Feb 6 -> Apr 24, 2026) - triton_shared: 06147630 -> f2eb8c5e (Feb 23 -> Aug 4, 2026) - LLVM: follows triton's cmake/llvm-hash.txt (87717bf9) Changes: - Replace 7 obsolete patches with 5 new ones for the updated triton/triton_shared - Remove deprecated usePropertiesForAttributes from dialect .td files - Adapt to MLIR API changes (VectorMultiReduction split, ConvertToLLVMPatternInterface) - Fix AlwaysInliner pass registry initialization in LLVMIRTranslation - Update test CHECK patterns for new LLVM IR codegen - Add clang to LLVM_ENABLE_PROJECTS in build script - Pass inbounds-fast-path=false to rewrite_tensor_descriptor_to_pointer Signed-off-by: mabsar <mabsar@qti.qualcomm.com> Signed-off-by: Arun Rangasamy <arangasa@qti.qualcomm.com>
…exKL beta1→beta2 - Update torch (2.10.0→2.12.0) and torchvision (0.25.0→0.27.0) in torch-requirements.txt - Update HexKL from 1.0.0-beta1 to 1.0.0-beta2 in setup_tools.sh and build_hexagon_mlir.sh - Remove obsolete add_rewrite_tensor_pointer pass from compiler.py Signed-off-by: mabsar <mabsar@qti.qualcommm.com> Signed-off-by: Arun Rangasamy <arangasa@qti.qualcomm.com>
…ate: 1. Update library path location 2. Use RPC timeout=30 mins (for tests running for more than 10s) 3. Use correct HexKL macro API function Signed-off-by: Arun Rangasamy <arangasa@qti.qualcomm.com>
Signed-off-by: Arun Rangasamy <arangasa@qti.qualcomm.com>
…lir 20260529→20260915 (qualcomm#98) Update torch (2.10.0→2.13.0), torchvision (0.25.0→0.28.0), and torch-mlir (20260529.826→20260915.877) in torch-requirements.txt. Remove obsolete add_rewrite_tensor_pointer pass from compiler.py, which was removed in newer triton and replaced by add_rewrite_tensor_descriptor_to_pointer. Tested on v75 and v81: 65 passed, 4 failed (known HexKL,next update). Signed-off-by: mabsar <mabsar@qti.qualcomm.com> Signed-off-by: Arun Rangasamy <arangasa@qti.qualcomm.com>
…1 skel resolution (qualcomm#99) - Upgrade Hexagon SDK from 6.4.0.2 to 6.6.0.0 - Upgrade HexKL from 1.0.0-beta1 to 1.0-beta.2 - Fix _sdk_tool_version to return toolv19 for v81+ (SDK 6.6.0.0 has no toolv88_v81 skel) - Add SDK version subdirectory to HexKL lib path to match SDK 6.6.0.0 layout - Add -L{hexkl_dir} to linker search path for HexKL libs Signed-off-by: mabsar <mabsar@qti.qualcomm.com> Signed-off-by: Arun Rangasamy <arangasa@qti.qualcomm.com>
[hexagon-mlir] Update triton/triton_shared and adapt to new LLVM (qualcomm#89) Bump dependencies: - triton: df38505e -> a9ced836 (Feb 6 -> Apr 24, 2026) - triton_shared: 06147630 -> f2eb8c5e (Feb 23 -> Aug 4, 2026) - LLVM: follows triton's cmake/llvm-hash.txt (87717bf9) Changes: - Replace 7 obsolete patches with 5 new ones for the updated triton/triton_shared - Remove deprecated usePropertiesForAttributes from dialect .td files - Adapt to MLIR API changes (VectorMultiReduction split, ConvertToLLVMPatternInterface) - Fix AlwaysInliner pass registry initialization in LLVMIRTranslation - Update test CHECK patterns for new LLVM IR codegen - Add clang to LLVM_ENABLE_PROJECTS in build script - Pass inbounds-fast-path=false to rewrite_tensor_descriptor_to_pointer Signed-off-by: mabsar <mabsar@qti.qualcomm.com>
…exKL beta1→beta2 - Update torch (2.10.0→2.12.0) and torchvision (0.25.0→0.27.0) in torch-requirements.txt - Update HexKL from 1.0.0-beta1 to 1.0.0-beta2 in setup_tools.sh and build_hexagon_mlir.sh - Remove obsolete add_rewrite_tensor_pointer pass from compiler.py Signed-off-by: mabsar <mabsar@qti.qualcommm.com>
arunrangasa
force-pushed
the
UpdateHexKLTorchVision
branch
from
September 21, 2026 10:01
d399fad to
6074f0a
Compare
Signed-off-by: arunrangasa <arangasa@qti.qualcomm.com>
Javed Absar (javedabsar1)
requested changes
Sep 22, 2026
| export HEXKL_ROOT=${BASE_DIR}/HEXKL_DIR/hexkl_addon | ||
|
|
||
| export HEXAGON_ARCH_VERSION=75 | ||
| export HEXAGON_ARCH_VERSION=81 |
Contributor
There was a problem hiding this comment.
change default to 81?
| torch==2.12.0+cpu | ||
| torchvision==0.27.0+cpu | ||
| torch-mlir==20260529.826 | ||
| torch==2.13.0+cpu |
Contributor
There was a problem hiding this comment.
This PR probably needs rebase as main https://github.com/qualcomm/hexagon-mlir/blob/main/ci/torch-requirements.txt shows already torch==2.13.0+cpu
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.