TBNilles/LEANN
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: Add ARM64 Linux wheel support for leann-backend-hnsw (#83)

Browse Source 
* feat: Add ARM64 Linux wheel support for leann-backend-hnsw

* fix: Use OpenBLAS for ARM64 Linux builds instead of Intel MKL

* fix: Configure Faiss with SVE optimization for ARM64 builds

- Set FAISS_OPT_LEVEL to "sve" for ARM64 architecture
- Disable x86-specific SIMD instructions (AVX2, AVX512, SSE4.1)
- Use ARM64-native SVE optimization as per Faiss conda build scripts
- Add architecture detection and proper configuration messages

Fixes compilation error: "xmmintrin.h: No such file or directory"
on ubuntu-24.04-arm runners.

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

* fix: Apply ARM64 compatibility fix directly to Faiss submodule

- Modify faiss/impl/pq.cpp to use x86-specific preprocessor conditions
- Remove patch file approach in favor of direct submodule modification
- Update CMakeLists.txt to reflect the submodule changes
- Fixes ARM64 Linux compilation by preventing x86 SIMD header inclusion

This resolves the "xmmintrin.h: No such file or directory" error
when building ARM64 Linux wheels for Docker compatibility.

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

* chore: Update Faiss submodule to include ARM64 compatibility fix

- Points to commit ed96ff7d with x86-specific preprocessor conditions
- Enables successful ARM64 Linux wheel builds

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

* retrigger ci

* fix: Use different optimization levels for ARM64 based on platform

- Use SVE optimization only for ARM64 Linux
- Use generic optimization for ARM64 macOS to avoid clang SVE issues
- Fixes macOS ARM64 compilation errors with SVE instructions

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

* feat: Update DiskANN submodule with OpenBLAS fallback support

- Points to commit 5c396c4 with ARM64 Linux OpenBLAS support
- Enables DiskANN to build on ARM64 Linux using standard BLAS libraries
- Resolves Intel MKL dependency issues for Docker ARM64 deployments

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

* fix: Update DiskANN submodule with ZeroMQ polling configuration

- Points to commit 3a1016e with explicit polling method setup
- Resolves ZeroMQ autodetection issues on ARM64 Linux
- Ensures stable cross-platform ZeroMQ builds

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

* retrigger ci

* fix: Update DiskANN submodule with ARM64 compiler flags fix

- Points to commit a0dc600 with architecture-specific compiler flags
- Removes x86 SIMD flags on ARM64 Linux to fix compilation errors
- Enables successful ARM64 Linux wheel builds

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

* fix: Update DiskANN submodule with ARM64 compiler flags fix

- Points to commit 0921664 with architecture-specific compiler flags
- Removes x86 SIMD flags on ARM64 Linux to fix compilation errors
- Enables successful ARM64 Linux wheel builds

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

* retrigger ci

* fix: Update DiskANN submodule with cross-platform prefetch support

- Points to commit 39192d6 with unified prefetch macros
- Replaces all Intel-specific _mm_prefetch calls with cross-platform macros
- Enables ARM64 Linux compatibility while maintaining x86 performance
- Resolves all remaining compilation errors for ARM64 builds

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

* fix: Update DiskANN submodule with corrected ARM64 compatibility fixes

- Points to commit 3cb87a8 with proper x86 platform detection
- Includes ARM64 fallback for AVXDistanceInnerProductFloat function
- Resolves all remaining '__m256 was not declared' compilation errors
- Enables successful ARM64 Linux wheel builds for Docker compatibility

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

* fix: Update DiskANN submodule with template type handling fix

- Points to commit d396bc3 with corrected template type handling
- Fixes DistanceInnerProduct template instantiation for int8_t/uint8_t types
- Resolves 'cannot convert const signed char* to const float*' error
- Completes ARM64 Linux compilation compatibility

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

* fix: Update DiskANN submodule with DistanceFastL2::norm template fix

- Points to commit 69d9a99 with corrected template type handling
- Fixes DistanceFastL2::norm template instantiation for int8_t/uint8_t types
- Resolves another 'cannot convert const signed char* to const float*' error
- Continues ARM64 Linux compilation compatibility improvements

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

* fix: Update DiskANN submodule with LAPACKE header detection

- Points to commit 64a9e01 with LAPACKE header path configuration
- Adds pkg-config based detection for LAPACKE include directories
- Resolves 'lapacke.h: No such file or directory' compilation error
- Completes OpenBLAS integration for ARM64 Linux builds

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

* fix: Update DiskANN submodule with enhanced LAPACKE header detection

- Points to commit 18d0721 with fallback LAPACKE header search paths
- Checks multiple standard locations for lapacke.h on various systems
- Improves ARM64 Linux compatibility for OpenBLAS builds
- Should resolve 'lapacke.h: No such file or directory' errors

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

* fix: Add liblapacke-dev package for ARM64 Linux builds

- Add liblapacke-dev to ARM64 dependencies alongside libopenblas-dev
- Provides lapacke.h header file needed for LAPACK C interface
- Fixes 'lapacke.h: No such file or directory' compilation error
- Enables complete OpenBLAS + LAPACKE support for ARM64 wheel builds

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

* fix: Update DiskANN submodule with cosine_similarity.h x86 intrinsics fix

- Points to commit dbb17eb with corrected conditional compilation
- Fixes immintrin.h inclusion for ARM64 compatibility in cosine_similarity.h
- Resolves 'immintrin.h: No such file or directory' error
- Continues systematic ARM64 Linux compilation fixes

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

* fix: Update DiskANN submodule with LAPACKE library linking fix

- Points to commit 19f9603 with explicit LAPACKE library discovery and linking
- Resolves 'undefined symbol: LAPACKE_sgesdd' runtime error on ARM64 Linux
- Completes ARM64 Linux wheel build compatibility for Docker deployments

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

---------

Co-authored-by: Claude <noreply@anthropic.com>
This commit is contained in:
Andy Lee
2025-09-02 14:27:06 -07:00
committed by GitHub
parent d41e467df9
commit 4dffdfedbe
4 changed files with 73 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

58
.github/workflows/build-reusable.yml vendored
View File

@@ -54,6 +54,17 @@ jobs:
python: '3.12'
- os: ubuntu-22.04
python: '3.13'
# ARM64 Linux builds
- os: ubuntu-24.04-arm
python: '3.9'
- os: ubuntu-24.04-arm
python: '3.10'
- os: ubuntu-24.04-arm
python: '3.11'
- os: ubuntu-24.04-arm
python: '3.12'
- os: ubuntu-24.04-arm
python: '3.13'
- os: macos-14
python: '3.9'
- os: macos-14
@@ -108,13 +119,46 @@ jobs:
pkg-config libabsl-dev libaio-dev libprotobuf-dev \
patchelf
# Install Intel MKL for DiskANN
wget -q https://registrationcenter-download.intel.com/akdlm/IRC_NAS/79153e0f-74d7-45af-b8c2-258941adf58a/intel-onemkl-2025.0.0.940.sh
sudo sh intel-onemkl-2025.0.0.940.sh -a --components intel.oneapi.lin.mkl.devel --action install --eula accept -s
source /opt/intel/oneapi/setvars.sh
echo "MKLROOT=/opt/intel/oneapi/mkl/latest" >> $GITHUB_ENV
echo "LD_LIBRARY_PATH=/opt/intel/oneapi/compiler/latest/linux/compiler/lib/intel64_lin" >> $GITHUB_ENV
echo "LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/intel/oneapi/mkl/latest/lib/intel64" >> $GITHUB_ENV
# Debug: Show system information
echo "🔍 System Information:"
echo "Architecture: $(uname -m)"
echo "OS: $(uname -a)"
echo "CPU info: $(lscpu | head -5)"
# Install math library based on architecture
ARCH=$(uname -m)
echo "🔍 Setting up math library for architecture: $ARCH"
if [[ "$ARCH" == "x86_64" ]]; then
# Install Intel MKL for DiskANN on x86_64
echo "📦 Installing Intel MKL for x86_64..."
wget -q https://registrationcenter-download.intel.com/akdlm/IRC_NAS/79153e0f-74d7-45af-b8c2-258941adf58a/intel-onemkl-2025.0.0.940.sh
sudo sh intel-onemkl-2025.0.0.940.sh -a --components intel.oneapi.lin.mkl.devel --action install --eula accept -s
source /opt/intel/oneapi/setvars.sh
echo "MKLROOT=/opt/intel/oneapi/mkl/latest" >> $GITHUB_ENV
echo "LD_LIBRARY_PATH=/opt/intel/oneapi/compiler/latest/linux/compiler/lib/intel64_lin" >> $GITHUB_ENV
echo "LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/intel/oneapi/mkl/latest/lib/intel64" >> $GITHUB_ENV
echo "✅ Intel MKL installed for x86_64"
# Debug: Check MKL installation
echo "🔍 MKL Installation Check:"
ls -la /opt/intel/oneapi/mkl/latest/ || echo "MKL directory not found"
ls -la /opt/intel/oneapi/mkl/latest/lib/ || echo "MKL lib directory not found"
elif [[ "$ARCH" == "aarch64" ]]; then
# Use OpenBLAS for ARM64 (MKL installer not compatible with ARM64)
echo "📦 Installing OpenBLAS for ARM64..."
sudo apt-get install -y libopenblas-dev liblapack-dev liblapacke-dev
echo "✅ OpenBLAS installed for ARM64"
# Debug: Check OpenBLAS installation
echo "🔍 OpenBLAS Installation Check:"
dpkg -l | grep openblas || echo "OpenBLAS package not found"
ls -la /usr/lib/aarch64-linux-gnu/openblas/ || echo "OpenBLAS directory not found"
fi
# Debug: Show final library paths
echo "🔍 Final LD_LIBRARY_PATH: $LD_LIBRARY_PATH"
- name: Install system dependencies (macOS)
if: runner.os == 'macOS'

2
packages/leann-backend-diskann/third_party/DiskANN vendored

Submodule packages/leann-backend-diskann/third_party/DiskANN updated: c593831474...19f9603c72

21
packages/leann-backend-hnsw/CMakeLists.txt
View File

@@ -49,9 +49,28 @@ set(BUILD_TESTING OFF CACHE BOOL "" FORCE)
set(FAISS_ENABLE_C_API OFF CACHE BOOL "" FORCE)
set(FAISS_OPT_LEVEL "generic" CACHE STRING "" FORCE)
# Disable additional SIMD versions to speed up compilation
# Disable x86-specific SIMD optimizations (important for ARM64 compatibility)
set(FAISS_ENABLE_AVX2 OFF CACHE BOOL "" FORCE)
set(FAISS_ENABLE_AVX512 OFF CACHE BOOL "" FORCE)
set(FAISS_ENABLE_SSE4_1 OFF CACHE BOOL "" FORCE)
# ARM64-specific configuration
if(CMAKE_SYSTEM_PROCESSOR MATCHES "aarch64|arm64")
message(STATUS "Configuring Faiss for ARM64 architecture")
if(CMAKE_SYSTEM_NAME STREQUAL "Linux")
# Use SVE optimization level for ARM64 Linux (as seen in Faiss conda build)
set(FAISS_OPT_LEVEL "sve" CACHE STRING "" FORCE)
message(STATUS "Setting FAISS_OPT_LEVEL to 'sve' for ARM64 Linux")
else()
# Use generic optimization for other ARM64 platforms (like macOS)
set(FAISS_OPT_LEVEL "generic" CACHE STRING "" FORCE)
message(STATUS "Setting FAISS_OPT_LEVEL to 'generic' for ARM64 ${CMAKE_SYSTEM_NAME}")
endif()
# ARM64 compatibility: Faiss submodule has been modified to fix x86 header inclusion
message(STATUS "Using ARM64-compatible Faiss submodule")
endif()
# Additional optimization options from INSTALL.md
set(CMAKE_BUILD_TYPE "Release" CACHE STRING "" FORCE)

2
packages/leann-backend-hnsw/third_party/faiss vendored

Submodule packages/leann-backend-hnsw/third_party/faiss updated: a0361858fc...ed96ff7dba