Test: Configure performance tests for math libraries

.github/workflows/test.yml

@@ -29,7 +29,7 @@ jobs:
 
       - name: Build
         run: |
-          cmake -B build -DBUILD_TESTING=ON -DENABLE_DEEPKS=ON -DENABLE_LIBXC=ON -DENABLE_LIBRI=ON -DENABLE_PAW=ON
+          cmake -B build -DBUILD_TESTING=ON -DENABLE_DEEPKS=ON -DENABLE_LIBXC=ON -DENABLE_LIBRI=ON -DENABLE_PAW=ON -DENABLE_GOOGLEBENCH=ON
           cmake --build build -j8
           cmake --install build
 

CMakeLists.txt

@@ -34,6 +34,7 @@ option(DEBUG_INFO "Print message for developers to debug." OFF)
 option(ENABLE_NATIVE_OPTIMIZATION "Enable compilation optimization for the native machine's CPU type" OFF)
 option(COMMIT_INFO "Print commit information in log" ON)
 option(ENABLE_FFT_TWO_CENTER "Enable FFT-based two-center integral method." ON)
+option(ENABLE_GOOGLEBENCH "Enable GOOGLE-benchmark usage." OFF)
 option(ENABLE_RAPIDJSON "Enable rapid-json usage." OFF)
 
 
@@ -594,6 +595,25 @@ if(INFO)
   # modifications on blas_connector and lapack_connector
 endif()
 
+#  Add performance test in abacus
+IF (ENABLE_GOOGLEBENCH)
+  set(BUILD_TESTING ON)
+  find_package(benchmark HINTS ${BENCHMARK_DIR})
+  if(NOT ${benchmark_FOUND})
+    set(BENCHMARK_USE_BUNDLED_GTEST OFF)
+    include(FetchContent)
+    FetchContent_Declare(
+      benchmark
+      GIT_REPOSITORY https://github.com/google/benchmark.git
+      GIT_TAG "origin/main"
+      GIT_SHALLOW TRUE
+      GIT_PROGRESS TRUE 
+    )
+    set(BENCHMARK_ENABLE_TESTING OFF)
+    FetchContent_MakeAvailable(benchmark)
+  endif()
+endif()
+
 IF (BUILD_TESTING)
   set_if_higher(CMAKE_CXX_STANDARD 14) # Required in orbital
   include(CTest)
@@ -623,8 +643,14 @@ IF (BUILD_TESTING)
     endif()
 
     #dependencies & link library
-    target_link_libraries(${UT_TARGET} ${UT_LIBS}
-      Threads::Threads GTest::gtest_main GTest::gmock_main)
+    if(ENABLE_GOOGLEBENCH)
+      target_link_libraries(${UT_TARGET} ${UT_LIBS}
+        Threads::Threads GTest::gtest_main GTest::gmock_main benchmark::benchmark)
+      else()
+        target_link_libraries(${UT_TARGET} ${UT_LIBS}
+          Threads::Threads GTest::gtest_main GTest::gmock_main)
+    endif()
+
     if(USE_OPENMP)
       target_link_libraries(${UT_TARGET} OpenMP::OpenMP_CXX)
     endif()
@@ -634,6 +660,7 @@ IF (BUILD_TESTING)
       WORKING_DIRECTORY $<TARGET_FILE_DIR:${UT_TARGET}>
     )
   endfunction(AddTest)
+
 endif()
 
 add_subdirectory(source)

Dockerfile.cuda

@@ -2,7 +2,7 @@ FROM nvidia/cuda:12.2.0-devel-ubuntu22.04
 
 RUN apt update && apt install -y --no-install-recommends \
     libopenblas-openmp-dev liblapack-dev libscalapack-mpi-dev libelpa-dev libfftw3-dev libcereal-dev \
-    libxc-dev libgtest-dev libgmock-dev python3-numpy \
+    libxc-dev libgtest-dev libgmock-dev libbenchmark-dev python3-numpy \
     bc cmake git g++ make bc time sudo unzip vim wget
 
 ENV GIT_SSL_NO_VERIFY=true TERM=xterm-256color \

Dockerfile.gnu

@@ -1,7 +1,7 @@
 FROM ubuntu:22.04
 RUN apt update && apt install -y --no-install-recommends \
     libopenblas-openmp-dev liblapack-dev libscalapack-mpi-dev libelpa-dev libfftw3-dev libcereal-dev \
-    libxc-dev libgtest-dev libgmock-dev python3-numpy \
+    libxc-dev libgtest-dev libgmock-dev libbenchmark-dev python3-numpy \
     bc cmake git g++ make bc time sudo unzip vim wget gfortran
 
 ENV GIT_SSL_NO_VERIFY=true TERM=xterm-256color \

Dockerfile.intel

@@ -2,7 +2,7 @@ FROM ubuntu:22.04
 
 RUN apt-get update && apt-get install -y \
     bc cmake git gnupg gcc g++ python3-numpy sudo wget vim unzip \
-    libcereal-dev libxc-dev libgtest-dev libgmock-dev
+    libcereal-dev libxc-dev libgtest-dev libgmock-dev libbenchmark-dev
 
 # Following steps by https://software.intel.com/content/www/us/en/develop/documentation/installation-guide-for-intel-oneapi-toolkits-linux/top/installation/install-using-package-managers/apt.html .
 RUN wget -O- https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB \

docs/advanced/install.md

@@ -69,6 +69,16 @@ After building and installing, unit tests can be performed with `ctest`.
 
 To run a subset of unit test, use `ctest -R <test-match-pattern>` to perform tests with name matched by given pattern.
 
+## Build Performance Tests
+
+To build performance tests for ABACUS, define `ENABLE_GOOGLEBENCH` flag. You can also specify the path to a local installation of [Google Benchmark](https://github.com/google/benchmark.git) by setting `BENCHMARK_DIR` flags. If not found locally, the configuration process will try to download it automatically.
+
+```bash
+cmake -B build -DENABLE_GOOGLEBENCH=1
+```
+
+Google Benchmark requires Google Test to build and run the tests. When setting `ENABLE_GOOGLEBENCH` to ON, `BUILD_TESTING` is automatically enabled. After building and installing, performance tests can be executed with `ctest`.
+
 ## Build with CUDA support
 
 ### Extra prerequisites

docs/quick_start/easy_install.md

@@ -126,6 +126,7 @@ Here, 'build' is the path for building ABACUS; and '-D' is used for setting up s
   - `ENABLE_LIBRI=OFF`: [Enable LibRI](../advanced/install.md#add-libri-support) to suppport variety of functionals. If `LIBRI_DIR` and `LIBCOMM_DIR` is defined, `ENABLE_LIBRI` will set to 'ON'.
   - `USE_OPENMP=ON`: Enable OpenMP support. Building ABACUS without OpenMP is not fully tested yet.
   - `BUILD_TESTING=OFF`: [Build unit tests](../advanced/install.md#build-unit-tests).
+  - `ENABLE_GOOGLEBENCH=OFF`: [Build performance tests](../advanced/install.md#build-performance-tests)
   - `ENABLE_MPI=ON`: Enable MPI parallel compilation. If set to `OFF`, a serial version of ABACUS with PW basis only will be compiled. Currently serial version of ABACUS with LCAO basis is not supported yet, so `ENABLE_LCAO` will be automatically set to `OFF`.
   - `ENABLE_COVERAGE=OFF`: Build ABACUS executable supporting [coverage analysis](../CONTRIBUTING.md#generating-code-coverage-report). This feature has a drastic impact on performance.
   - `ENABLE_ASAN=OFF`: Build with Address Sanitizer. This feature would help detecting memory problems.

source/module_base/test/CMakeLists.txt

@@ -217,6 +217,13 @@ AddTest(
   SOURCES assoc_laguerre_test.cpp ../assoc_laguerre.cpp ../tool_quit.cpp ../global_variable.cpp ../global_file.cpp ../global_function.cpp ../memory.cpp ../timer.cpp
   LIBS ${math_libs} formatter
 )
+if(ENABLE_GOOGLEBENCH)
+  AddTest(
+    TARGET perf_sphbes
+    LIBS formatter
+    SOURCES perf_sphbes_test.cpp ../math_sphbes.cpp ../timer.cpp 
+  )
+endif()
 
 if(ENABLE_RAPIDJSON)
   AddTest(

source/module_base/test/perf_sphbes_test.cpp

@@ -0,0 +1,72 @@
+#include"../math_sphbes.h"
+#include<fstream>
+#include <benchmark/benchmark.h>
+#include <iostream>
+#include <cstring>
+#include <cmath>
+
+/************************************************
+*  performace test of class Sphbes
+***********************************************/
+
+/**
+ * Tested function: 
+ *      - sphbesj
+ *      - Spherical_Bessel
+ */
+
+class PerfSphbes : public benchmark::Fixture {
+public:
+    const double q = 1;
+    const int n = 1000;
+    double stop = 1000.0;
+    double dr = 0.0;
+    double* rc, *rinf, *jc, *jinf;
+    void SetUp(const benchmark::State& state){
+        const double rcut = state.range(0) + 0.5;
+        rc = new double[n + 10]; 
+        rinf = new double[n + 10];
+        jc = new double[n + 10];
+        jinf = new double[n + 10];
+
+        // generate data points in (0, rcut] in log scale
+        double rmin = 0.0001;
+        double log_rmin = std::log(rmin);
+        double log_rcut = std::log(rcut);
+        dr = (log_rcut - log_rmin) / (n-1);
+        memset(rc, 0, (n+10) * sizeof(double));
+        for (int i = 0; i < n; i++)
+            rc[i] = std::exp(log_rmin + i * dr);
+
+        // generate data points in [rcut, stop] in linear scale
+        memset(rinf, 0, (n+10) * sizeof(double));
+        rinf[0] = rcut;
+        dr = (stop - rcut) / (n-1);
+        for (int i = 1; i < n; i++)
+            rinf[i] += rinf[i-1] + dr;
+    }
+    void TearDown(const benchmark::State& state){
+        delete[] rc;
+        delete[] rinf;
+        delete[] jc;
+        delete[]  jinf;
+    }
+};    
+
+BENCHMARK_DEFINE_F(PerfSphbes, BM_Spherical_Bessel)(benchmark::State& state) {
+    for (auto _ : state) {
+        ModuleBase::Sphbes::Spherical_Bessel(n, rc, q, state.range(0), jc);
+        ModuleBase::Sphbes::Spherical_Bessel(n, rinf, q, state.range(0), jinf);
+    }
+}
+
+BENCHMARK_DEFINE_F(PerfSphbes, BM_sphbesj)(benchmark::State& state) {
+    for (auto _ : state) {
+        ModuleBase::Sphbes::sphbesj(n, rc, q, state.range(0), jc);
+        ModuleBase::Sphbes::sphbesj(n, rinf, q, state.range(0), jinf);
+    }
+}
+
+BENCHMARK_REGISTER_F(PerfSphbes, BM_sphbesj)->DenseRange(0, 11, 1)->Unit(benchmark::kMicrosecond);
+BENCHMARK_REGISTER_F(PerfSphbes, BM_Spherical_Bessel)->DenseRange(0, 11, 1)->Unit(benchmark::kMicrosecond);
+BENCHMARK_MAIN();