This repository aims to benchmark Matrix Multiply (SGEMM) hand-tuned libraries and code generation stacks on a single thread on one CPU core. The focus will be on machine learning workloads so FP32 or smaller and irregular sizes of matrices. The goal is to expose high performance atomic kernels that can then be used to build highly efficient higher level implemenations spanning multiple cores or distributed across systems.
Note: 8GB Mac Mini runs roughly 25% slower than the 16GB version on other tests.
Clone the repo along with submodules.
git clone --recurse-submodules https://github.com/mmperf/mmperf.git
Create a virtual environment and install requirements.
cd mmperf
python3 -m venv ./mmperf_env
source mmperf_env/bin/activate
pip install -r requirements.txt
pip install -r ./external/llvm-project/mlir/python/requirements.txt
Build the project specifying the backend(s) to run matmul. Below is a command to build mmperf with MLIR backend.
cmake -GNinja \
-DCMAKE_CXX_COMPILER=clang++-11 \
-DCMAKE_C_COMPILER=clang-11 \
-DUSE_MLIR=ON \
-B build .
cmake --build build
Another example to build with all available backends. Assumes you have MKL, OpenBLAS, and Halide installed (see below for installation details)
HALIDE_DIR=/home/foo/lokal/halide/ MKL_DIR=/opt/intel/oneapi/mkl/latest/ cmake -GNinja \
-DCMAKE_CXX_COMPILER=clang++-11 \
-DCMAKE_C_COMPILER=clang-11 \
-DMKL_DIR=/opt/intel/oneapi/mkl/latest/ \
-DUSE_MLIR=ON \
-DUSE_MKL=ON \
-DUSE_RUY=ON \
-DUSE_HALIDE=ON \
-DUSE_OPENBLAS=ON \
-DUSE_IREE=ON \
-B build .
cmake --build build
If you want to build mmperf with MLIR-CUDA backend, you need to have NVIDIA CUDA-11.0 installed on your system. Make sure it is installed and environment variables $PATH and $LD_LIBRARY_PATH are correctly configured. Also make sure if you can invoke nvcc compiler from the command line. How to install NIVIDIA CUDA-11.0 toolkit, please refer to this link. In order to compile the MLIR-CUDA backend, you need to specify -DUSE_MLIR_CUDA switch and specify the compiler with -DCMAKE_CUDA_COMPILER. For example, if you want to compile both MLIR-CUDA and CUBLAS backend, the compilation command would look like this:
cmake -GNinja \
-DCMAKE_CXX_COMPILER=clang++-11 \
-DCMAKE_C_COMPILER=clang-11 \
-DCMAKE_CUDA_COMPILER=nvcc
-DUSE_MLIR_CUDA=ON \
-DUSE_CUBLAS=ON \
-DSIZE_FILE=benchmark_sizes/benchmark_small_sizes.txt \
-B build .
cmake --build buildInstall matplotlib to generate performance plot.
pip install matplotlib
We use AOT compilation to generate binaries for matrix multiplication for specified backends and run them to generate the benchmarking numbers. To run all tests and generate performance numbers run:
cmake --build build/matmul --target run_all_tests
results folder will be created in the mmperf top-level directory which will contain GLOPS for every matmul size and every backend. A plot comparing performances of backends will also be generated in matmul.png.
Each generated binary can also be executed individually. To run a specific matrix size (say 24x64x512) for a backend run:
./build/matmul/matmul_<LIBRARY>_24x64x512
Matrix sizes: benchmark_sizes folder has text files containing the matrix sizes that mmperf runs on. You can change the matrix size input file by editing SIZE_FILE option in cmake/common.cmake. Default is benchmark_all_sizes.txt.
Number of iterations: The number of iterations for a matmul to be benchmarked can be set by changing NUM_REPS variable in cmake/common.cmake. Default is 100.
The building of submodule external/llvm-project can be space and time consuming. If you already have your own standalone llvm and don't want to fetch and compile this submodule, you scan specify the llvm on your system with PREBUILT_LLVM_PATH compilation flag:
cmake -GNinja \
-DCMAKE_CXX_COMPILER=clang++-11 \
-DCMAKE_C_COMPILER=clang-11 \
-DPREBUILT_LLVM_PATH=$HOME/opt/llvm \
-DUSE_MLIR=ON \
-B build .
cmake --build buildTo compile llvm from scratch, you might want all of these as well:
echo "deb http://apt.llvm.org/DISTRO_NAME/ llvm-toolchain-DISTRO_NAME main" >> /etc/apt/sources.list
wget -O - https://apt.llvm.org/llvm-snapshot.gpg.key | apt-key add -
apt-get update && apt-get upgrade -y
apt-get install -y clang-11 clang-tools-11 libc++1-11 libc++-11-dev \
libc++abi1-11 libc++abi-11-dev libclang1-11 libclang-11-dev \
libclang-common-11-dev libclang-cpp11 libclang-cpp11-dev liblld-11 \
liblld-11-dev liblldb-11 liblldb-11-dev libllvm11 libomp-11-dev \
libomp5-11 lld-11 lldb-11 llvm-11 llvm-11-dev llvm-11-runtime \
llvm-11-tools libfuzzer-11-devgit clone https://github.com/halide/Halide.git --recurse-submodules
cd Halide/
sudo apt install libclang-11-dev clang-11 liblld-11-dev
LLD_DIR=/usr/lib/llvm-11/lib/cmake/lld cmake . -GNinja \
-DCMAKE_BUILD_TYPE=Release \
-DTARGET_WEBASSEMBLY=OFF \
-DCMAKE_INSTALL_PREFIX=/home/<foo>/lokal/
ninja
ninja install
export HALIDE_DIR=/home/<foo>/lokal/halide
sudo apt install libopenblas-dev
git clone https://github.com/flame/blis
cd blis
./configure --prefix=/home/foo/lokal/ --enable-cblas -c amd64
make -j 16
make install
Download and install from https://software.intel.com/content/www/us/en/develop/articles/installation-guide-for-intel-oneapi-toolkits.html
The linalg codegen pass is in matmul/matmul-compile/matmul-compile.cpp.
This benchmark was run on an Intel Xeon CPU running at 3.1GHz. The machine has 256Kb L1 cache, 8Mb L2 cache and 24.8Mb L3 cache. It supports AVX-512 instructions. The peak performance of the machine is 3.1 x 8 x 2 x 2 = 99.2 GFLOPS for double precision and 198.4 GFLOPS for single precision.