pytorch/.circleci/scripts/binary_ios_upload.sh
Tao Xu 6de6041585 [iOS] Disable NNPACK on iOS builds (#39868)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/39868

### Summary

why disable NNPACK on iOS

- To stay consistency with our internal version
- It's currently blocking some external users due to its lack support of x86 architecture
    - https://github.com/pytorch/pytorch/issues/32040
    - https://discuss.pytorch.org/t/undefined-symbols-for-architecture-x86-64-for-libtorch-in-swift-unit-test/84552/6
- NNPACK uses fast convolution algorithms (FFT, winograd) to reduce the computational complexity of convolutions with large kernel size. The algorithmic speedup is limited to specific conv params which are unlikely to appear in mobile networks.
- Since XNNPACK has been enabled, it performs much better than NNPACK on depthwise-separable convolutions which is the algorithm being used by most of mobile computer vision networks.

### Test Plan

- CI Checks

Test Plan: Imported from OSS

Differential Revision: D22087365

Pulled By: xta0

fbshipit-source-id: 89a959b0736c1f8703eff10723a8fbd02357fd4a
2020-06-17 01:39:56 -07:00

44 lines
1.5 KiB
Bash

#!/bin/bash
set -ex -o pipefail
echo ""
echo "DIR: $(pwd)"
WORKSPACE=/Users/distiller/workspace
PROJ_ROOT=/Users/distiller/project
ARTIFACTS_DIR=${WORKSPACE}/ios
ls ${ARTIFACTS_DIR}
ZIP_DIR=${WORKSPACE}/zip
mkdir -p ${ZIP_DIR}/install/lib
mkdir -p ${ZIP_DIR}/src
# copy header files
cp -R ${ARTIFACTS_DIR}/arm64/include ${ZIP_DIR}/install/
# build a FAT bianry
cd ${ZIP_DIR}/install/lib
target_libs=(libc10.a libclog.a libcpuinfo.a libeigen_blas.a libpytorch_qnnpack.a libtorch_cpu.a libtorch.a libXNNPACK.a)
for lib in ${target_libs[*]}
do
if [ -f "${ARTIFACTS_DIR}/x86_64/lib/${lib}" ] && [ -f "${ARTIFACTS_DIR}/arm64/lib/${lib}" ]; then
libs=("${ARTIFACTS_DIR}/x86_64/lib/${lib}" "${ARTIFACTS_DIR}/arm64/lib/${lib}")
lipo -create "${libs[@]}" -o ${ZIP_DIR}/install/lib/${lib}
fi
done
lipo -i ${ZIP_DIR}/install/lib/*.a
# copy the umbrella header and license
cp ${PROJ_ROOT}/ios/LibTorch.h ${ZIP_DIR}/src/
cp ${PROJ_ROOT}/LICENSE ${ZIP_DIR}/
# zip the library
ZIPFILE=libtorch_ios_nightly_build.zip
cd ${ZIP_DIR}
#for testing
touch version.txt
echo $(date +%s) > version.txt
zip -r ${ZIPFILE} install src version.txt LICENSE
# upload to aws
brew install awscli
set +x
export AWS_ACCESS_KEY_ID=${AWS_S3_ACCESS_KEY_FOR_PYTORCH_BINARY_UPLOAD}
export AWS_SECRET_ACCESS_KEY=${AWS_S3_ACCESS_SECRET_FOR_PYTORCH_BINARY_UPLOAD}
set +x
# echo "AWS KEY: ${AWS_ACCESS_KEY_ID}"
# echo "AWS SECRET: ${AWS_SECRET_ACCESS_KEY}"
aws s3 cp ${ZIPFILE} s3://ossci-ios-build/ --acl public-read