spec2017-on-iOS

Minimum iOS version iOS16

This is only an experimental playground to run spec2017 on iOS. It supports all the specRate2017 Interger and FP Benchmarks. Use at your own risk!

How to install

1. Download ipa file on release page
2. Use sideload app to sideload it.(if you don't have a computer, you can use scarlet or flekstore.)

How to compile it yourself

Pre requirement: Xcode Version >=14, llvm-objcopy, make.

1. Copy the input folder from the ipa file and paste it under Spec2017/ folder.
2. Add specbin/rary.a as static lib

TODO: How to compile spec benchmarks

1. Modify all the benchmarks main Function/Program's name to name inside /specbin/specEntry.c file
2. Here is the list of benchmark entry points

intRate

500.perlbench     -perlmainc.c       line 61
502.gcc           -main.c            line 35
505.mcf           -mcf.c             line 134
520.omnetpp_r     -simulator/main.cc line 32
523.xalancbmk_r   -XalanExe.cpp      line 870
525.x264_r        -x264_src/x264.c   line 152
531.deepsjeng_r   -sjeng.cpp         line 44
541.leela_r       -Leela.cpp         line 20
548.exchange2_r   -exchange2.f90     line 1383
557.xz_r          -spec.c            line 25

fpRate

503.bwaves_r     -flow_lam.F                              line 107
507.cactuBSSN_r  -Cactus/main/flesh.cc                    line 57
508.namd_r       -spec_namd.C                             line 35
510.parest_r     -source/me-tomography/me_tomography.cc   line 215
511.povray_r     -povray.cpp                              line 190
519.lbm_r        -main.c                                  line 25
521.wrf_r        -wrf.f90                                 line 6
526.blender_r    -blender/source/creator/creator.c        line 1530
527.cam4_r       -ccsm_driver.f90                         line 25
538.imagick_r    -utilities/convert.c                     line 91
544.nab_r        -nabmd.c                                 line 40
549.fotonik3d_r  -yeemain.f90                             line 37
554.roms_r       -master.f90                              line 1241

How to compile static library(llvm flang) for Fortran benchmark

Note: You can find the pre-built library inside fortran_lib folder(All build with Flang17.0.0).

1. Follow the instruction on offical repo (https://github.com/llvm/llvm-project/tree/main/flang) to build the flang compiler for your own platform.
2. Clone the repo again for building ios lib
3. Next run the following commands(PATH_TO_LLVM_BUILD_FOLDER is the folder of the native platform build)

rm -rf build
mkdir build
rm -rf install
mkdir install
ROOTDIR=`pwd`
INSTALLDIR=$ROOTDIR/install

cd build
cmake \
  -G Ninja \
  -DCMAKE_BUILD_TYPE=Release \
  -DCMAKE_INSTALL_PREFIX=$INSTALLDIR \
  -DCMAKE_CXX_STANDARD=17 \
  -DCMAKE_EXPORT_COMPILE_COMMANDS=ON \
  -DCMAKE_CXX_LINK_FLAGS="-Wl,-rpath,$LD_LIBRARY_PATH" \
  -DLLVM_TABLEGEN=/PATH_TO_LLVM_BUILD_FOLDER/bin/llvm-tblgen \
  -DCLANG_TABLEGEN=/PATH_TO_LLVM_BUILD_FOLDERbin/clang-tblgen \
  -DMLIR_TABLEGEN=/PATH_TO_LLVM_BUILD_FOLDER/bin/mlir-tblgen \
  -DMLIR_TABLEGEN_EXE=/PATH_TO_LLVM_BUILD_FOLDER/bin/mlir-tblgen \
  -DMLIR_LINALG_ODS_YAML_GEN=/PATH_TO_LLVM_BUILD_FOLDER/bin/mlir-linalg-ods-yaml-gen \
  -DMLIR_LINALG_ODS_YAML_GEN_EXE=/PATH_TO_LLVM_BUILD_FOLDER/bin/mlir-linalg-ods-yaml-gen \
  -DFLANG_INCLUDE_TESTS=FALSE \
  -DFLANG_BUILD_EXAMPLES=FALSE \
  -DFLANG_BUILD_TOOLS=OFF \
  -DLLVM_ENABLE_ASSERTIONS=ON \
  -DLLVM_TARGETS_TO_BUILD=AArch64 \
  -DLLVM_LIT_ARGS=-v \
  -DLLVM_ENABLE_PROJECTS="clang;mlir;flang;openmp" \
  -DLLVM_TARGET_ARCH=AArch64 \
  -DLLVM_DEFAULT_TARGET_TRIPLE=arm64-apple-darwin-macho \
  -DLLVM_BUILD_RUNTIME=Off -DLLVM_INCLUDE_TESTS=Off \
  -DLLVM_INCLUDE_EXAMPLES=Off -DLLVM_ENABLE_BACKTRACES=Off \
  -DCMAKE_OSX_ARCHITECTURES="arm64" \
  -DLLVM_ENABLE_PIC=False \
  -DCMAKE_OSX_SYSROOT=/Library/Developer/CommandLineTools/SDKs/iPhoneOS16.2.sdk \
  -DCMAKE_C_FLAGS="-arch arm64 -target arm64-apple-darwin-macho -miphoneos-version-min=14" \
  -DCMAKE_CXX_FLAGS="-arch arm64 -target arm64-apple-darwin-macho -miphoneos-version-min=14" \
  ../llvm-project/llvm
  
  && ninja

5. You will find the static lib under lib/ folder. (Mainly libFortranCommon.a libFortranDecimal.a libFortranEvaluate.a libFortranEvaluateTesting.a libFortranLower.a libFortran_main.a libFortranParser.a libFortranRuntime.a libFortranSemantics.a)