姓名:梁鑫嵘;学号:200110619

实验内容

  1. 进行HPL(The High-Performance Linpack Benchmark)性能测试
  2. 完成实验报告

实验内容

硬件配置

  1. CPU 硬件配置: Intel(R) Xeon(R) Gold 5218 CPU @ 2.30GHz
  2. 峰值性能计算:
    1. $FLOPS_{\text{cpu, double}}=N_{\text{cores}} \times Freq_{\text{per core}} \times N_{FMA}\times 2_{\text{plus and multiplication}} \times \frac{512_{\text{AVX 512}}}{64}$
    2. $FLOPS_{\text{5218, double}}=16_{\text{cores}} \times 2.30_{\text{per core}} \times 1_{FMA}\times 2_{\text{plus and multiplication}} \times \frac{512_{\text{AVX 512}}}{64}=588.8 \ GFlops$

软件依赖

  1. 运行环境:
    1. Vmmare 虚拟机环境,宿主机为 Windows 10 双 Intel Xeon 5218。
    2. OpenEuler 系统
  2. 编译器: gcc version 9.3.1 (GCC)
  3. MPI: mpirun (Open MPI) 4.1.1
  4. BLAS库: libopenblas_skylakexp-r0.3.17.dev.so
  5. HPL:http://www.netlib.org/benchmark/hpl/hpl-2.3.tar.gz

HPL 编译

  1. 编译安装 OpenBLAS

    git clone <https://github.com/xianyi/OpenBLAS> && cd OpenBLAS
    make -j16
    sudo make install
    
  2. 编译安装 OpenMPI

    wget <https://download.open-mpi.org/release/open-mpi/v4.1/openmpi-4.1.1.tar.gz>
    tar xzf openmpi-4.1.1.tar.gz
    cd openmpi-4.1.1
    ./configure
    make -j16
    sudo make install
    
  3. 编译 HPL

    wget <http://www.netlib.org/benchmark/hpl/hpl-2.3.tar.gz>
    tar xzf hpl-2.3.tar.gz
    cd hpl-2.3
    ./configure
    make -j16
    cd testing
    
  4. 准备运行环境

    mkdir /share/
    chmod 775 /share/
    cd /share/
    mkdir /share/hpc/
    cd hpc
    cp ~/hpl-2.3/testing xhpl .
    

HPL 运行