Merge pull request #152 from deepmodeling/develop

Merge deepmodeling develop up to v2.2.4

Author: dyzheng

.github/workflows/cuda.yml

@@ -42,14 +42,31 @@ jobs:
         run: |
           nvidia-smi
           cmake -B build -DUSE_CUSOLVER_LCAO=ON
-          cmake --build build -j4
+          cmake --build build -j4 
           cmake --install build
-      - name: Test
+          cmake -B build -DBUILD_TESTING=ON
+          cmake --build build -j4 --target hsolver_diago
+      - name: Test e2e
         run: |
           export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/usr/local/cuda/lib64
           cd tests/integrate
           echo "ks_solver cusolver" >> ./270_NO_MD_2O/INPUT
-          ./Autotest.sh -r 270_NO_MD_2O
+          ./Autotest.sh -r 270_NO_MD_2O 
+      - name: Test UT
+        run: |
+          cd source/src_pdiag/test/
+          cp ../../../build/source/src_pdiag/test/hsolver_diago .
+          ./hsolver_diago
+          bash diago_parallel_test.sh
+      - name: Test performance
+        run: |
+          cd examples/performance
+          ls -d P1*lcao* > allcase
+          sed -i '/ks_solver/d' P1*lcao*/INPUT
+          sed -i '$a ks_solver cusolver' P1*lcao*/INPUT 
+          bash run.sh
+          cat sumall.dat
+          
 
   stop-runner:
     name: Stop self-hosted EC2 runner

.github/workflows/image.yml

@@ -1,10 +1,10 @@
 name: Build Image
 
 on:
+  workflow_dispatch:
   push:
-    branches:
-      - develop
-
+    tags:
+      - 'v*'
 jobs:
   build_container_and_push:
     runs-on: ubuntu-latest
@@ -13,6 +13,15 @@ jobs:
       - name: Checkout
         uses: actions/checkout@v3
 
+      - name: Docker meta
+        id: meta
+        uses: docker/metadata-action@v4
+        with:
+          images: |
+            ghcr.io/${{ github.repository_owner }}
+            ${{ secrets.DP_HARBOR_REGISTRY }}/dplc
+          tags: type=semver,pattern={{version}}
+
       - name: Setup Docker Buildx
         uses: docker/setup-buildx-action@v1
 
@@ -33,10 +42,6 @@ jobs:
       - name: Build and Push Container
         uses: docker/build-push-action@v2
         with:
-          tags: |
-            ghcr.io/${{ github.repository_owner }}/abacus:latest
-            ${{ secrets.DP_HARBOR_REGISTRY }}/dplc/abacus:latest
+          tags: ${{ steps.meta.outputs.tags }}
           file: Dockerfile
-          cache-from: type=registry,ref=ghcr.io/${{ github.repository_owner }}/abacus:latest
-          cache-to: type=inline
-          push: true
+          push: ${{ github.event_name == 'push' }}

.github/workflows/performance.yml

@@ -26,7 +26,7 @@ jobs:
           cmake --install build
       - name: Test
         run: |
-          . /opt/intel/oneapi/setvars.sh 2> /dev/null || :
+          test -e /opt/intel/oneapi/setvars.sh && . /opt/intel/oneapi/setvars.sh
           cd examples/performance/
           bash run.sh
       - name: Show Result

CMakeLists.txt

@@ -5,7 +5,7 @@ cmake_minimum_required(VERSION 3.16)
 ########################################
 
 project(ABACUS
-    VERSION 2.2.0
+    VERSION 2.2.3
     DESCRIPTION "ABACUS is an electronic structure package based on DFT."
     HOMEPAGE_URL "https://github.com/deepmodeling/abacus-develop"
     LANGUAGES CXX
@@ -28,8 +28,10 @@ set(ABACUS_TEST_DIR ${CMAKE_CURRENT_SOURCE_DIR}/tests)
 set(ABACUS_BIN_PATH ${CMAKE_CURRENT_BINARY_DIR}/${ABACUS_BIN_NAME})
 include_directories(${ABACUS_SOURCE_DIR})
 add_executable(${ABACUS_BIN_NAME} source/main.cpp)
+set(CMAKE_CXX_STANDARD 11)
+set(CMAKE_CXX_STANDARD_REQUIRED ON)
 add_compile_options(-O2 -g)
-SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wno-write-strings" )
+add_compile_options(-Wno-write-strings)
 
 find_package(Cereal REQUIRED)
 include_directories(${Cereal_INCLUDE_DIR})
@@ -52,10 +54,9 @@ if(USE_OPENMP)
   find_package(OpenMP REQUIRED)
   target_link_libraries(${ABACUS_BIN_NAME} OpenMP::OpenMP_CXX)
   add_compile_options(${OpenMP_CXX_FLAGS})
-  #add_compile_definitions(_OPENMP)
+  add_link_options(${OpenMP_CXX_LIBRARIES})
 endif()
 
-set(CMAKE_CXX_STANDARD 11)
 include(CheckLanguage)
 check_language(CUDA)
 if(CMAKE_CUDA_COMPILER)
@@ -71,7 +72,7 @@ if(CMAKE_CUDA_COMPILER)
     endif()
   else()
     if(NOT USE_CUDA AND NOT USE_CUSOLVER_LCAO)
-      message(WARNING "CUDA components detected, but both USE_CUDA and USE_CUSOLVER_LCAO set to OFF. \nNOT building CUDA version of ABACUS.")
+      message(STATUS "CUDA components detected, but both USE_CUDA and USE_CUSOLVER_LCAO set to OFF. NOT building CUDA version of ABACUS.")
     elseif (USE_CUDA AND USE_CUSOLVER_LCAO)
       message(FATAL_ERROR "USE_CUDA and USE_CUSOLVER_LCAO set, but now they not allowed to coexist.")
     endif()
@@ -88,8 +89,7 @@ if(USE_CUDA OR USE_CUSOLVER_LCAO)
   cmake_minimum_required(VERSION 3.18) # required by `CUDA_ARCHITECTURES` below
   set(CMAKE_CXX_STANDARD 14)
   set(CMAKE_CXX_EXTENSIONS ON)
-  set(CMAKE_CXX_STANDARD_REQUIRED ON)
-  set(CMAKE_CUDA_STANDARD 14)
+  set(CMAKE_CUDA_STANDARD ${CMAKE_CXX_STANDARD})
   set(CMAKE_CUDA_STANDARD_REQUIRED ON)
   set(CMAKE_CUDA_HOST_COMPILER ${CMAKE_CXX_COMPILER})
   enable_language(CUDA)
@@ -120,9 +120,7 @@ endif()
 # Warning: CMake add support to HIP in version 3.21. This is rather a new version.
 # Use cmake with AMD-ROCm: https://rocmdocs.amd.com/en/latest/Installation_Guide/Using-CMake-with-AMD-ROCm.html
 if(USE_ROCM)
-  set(CMAKE_CXX_STANDARD 11)
-  SET(CMAKE_HIP_STANDARD 11)
-  # set(CMAKE_CXX_STANDARD_REQUIRED True)
+  SET(CMAKE_HIP_STANDARD ${CMAKE_CXX_STANDARD})
   list(APPEND CMAKE_PREFIX_PATH /opt/rocm/hip /opt/rocm)
   list(APPEND CMAKE_SYSTEM_PREFIX_PATH "/opt/rocm/hip")
 
@@ -145,6 +143,9 @@ if(USE_ROCM)
 endif()
 
 if(ENABLE_ASAN)
+  if (CMAKE_CXX_COMPILER_ID STREQUAL "Intel")
+    message(FATAL_ERROR "Address Sanitizer is not supported on Intel compiler.")
+  endif()
   add_compile_options(
     -fsanitize=address
     -fno-omit-frame-pointer
@@ -206,13 +207,16 @@ if(ENABLE_DEEPKS)
     FetchContent_Declare(
       libnpy
       GIT_REPOSITORY https://github.com/llohse/libnpy.git
+      GIT_SHALLOW TRUE
+      GIT_PROGRESS TRUE
     )
     FetchContent_MakeAvailable(libnpy)
   endif()
   include_directories(${libnpy_SOURCE_DIR}/include)
   add_compile_definitions(__DEEPKS)
 endif()
 
+list(APPEND math_libs m)
 target_link_libraries(${ABACUS_BIN_NAME} ${math_libs})
 
 if(DEFINED Libxc_DIR)
@@ -244,17 +248,18 @@ add_compile_definitions(
 )
 
 IF (BUILD_TESTING)
+  set(CMAKE_CXX_STANDARD 14) # Required in orbital
   include(CTest)
   enable_testing()
-  find_package(Threads)
-  set(CMAKE_CXX_STANDARD_REQUIRED ON)
   find_package(GTest HINTS /usr/local/lib/ ${GTEST_DIR})
   if(NOT ${GTest_FOUND})
     include(FetchContent)
     FetchContent_Declare(
       googletest
       GIT_REPOSITORY https://github.com/google/googletest.git
       GIT_TAG "origin/main"
+      GIT_SHALLOW TRUE
+      GIT_PROGRESS TRUE
     )
     FetchContent_MakeAvailable(googletest)
   endif()
@@ -267,8 +272,11 @@ IF (BUILD_TESTING)
     add_executable(${UT_TARGET} ${UT_SOURCES})
     #dependencies & link library
     target_link_libraries(${UT_TARGET} ${UT_LIBS}
-      OpenMP::OpenMP_CXX pthread GTest::gtest_main GTest::gmock_main)
-    install(TARGETS ${UT_TARGET} DESTINATION ${CMAKE_CURRENT_BINARY_DIR}/../../../tests )
+      Threads::Threads GTest::gtest_main GTest::gmock_main)
+    if(USE_OPENMP)
+      target_link_libraries(${UT_TARGET} OpenMP::OpenMP_CXX)
+    endif()
+    install(TARGETS ${UT_TARGET} DESTINATION ${CMAKE_BINARY_DIR}/tests )
     add_test(NAME ${UT_TARGET}
       COMMAND ${UT_TARGET}
       WORKING_DIRECTORY $<TARGET_FILE_DIR:${UT_TARGET}>
@@ -290,6 +298,7 @@ target_link_libraries(${ABACUS_BIN_NAME}
     io
     ions
     lcao
+    gint
     parallel
     mrrr
     pdiag
@@ -302,7 +311,6 @@ target_link_libraries(${ABACUS_BIN_NAME}
     hamilt
     psi
     esolver
-    -lm
 )
 
 install(PROGRAMS ${ABACUS_BIN_PATH}

Dockerfile.gnu

@@ -1,9 +1,9 @@
 FROM debian:bullseye-slim
 
-RUN apt-get update && apt-get install -y --no-install-recommends git g++ gfortran libssl-dev make cmake vim wget bc unzip \
+RUN apt-get update && apt-get install -y --no-install-recommends git g++ gfortran libssl-dev make cmake vim wget bc unzip python3-numpy\
     && apt-get install -y --no-install-recommends mpich libmpich-dev
 
-ENV GIT_SSL_NO_VERIFY 1
+ENV GIT_SSL_NO_VERIFY=1 TERM=xterm-256color
 
 RUN cd /tmp \
     && git clone https://github.com/USCiLab/cereal.git \

docs/input-main.md

@@ -1114,7 +1114,7 @@ This part of variables are used to control the molecular dynamics calculations.
 
 - **Type**: Double
 - **Description**: This is the time step(fs) used in md simulation .
-- **Default**: 1
+- **Default**: 1.0
 
 #### md_tfirst & md_tlast
 
@@ -1178,31 +1178,31 @@ temperature will fluctuate violently; if it is too small, the temperature will t
 
 - **Type**: Real
 - **Description**: the velocity of shock wave ($\AA$/fs) for MSST.
-- **Default**: 0
+- **Default**: 0.0
 
 #### msst_vis
 
 - **Type**: Real
 - **Description**: artificial viscosity (mass/length/time) for MSST.
-- **Default**: 0
+- **Default**: 0.0
 
 #### msst_tscale
 
 - **Type**: Real
 - **Description**: reduction in initial temperature (0~1) used to compress volume in MSST.
-- **Default**: 0
+- **Default**: 0.01
 
 #### msst_qmass
 
-- **Type**: Double
+- **Type**: Real
 - **Description**: Inertia of extended system variable. Used only when md_type is 4, you should set a number which is larger than 0. Note that Qmass of NHC is set by md_tfreq.
 - **Default**: No default
 
 #### md_damp
 
 - **Type**: Real
 - **Description**: damping parameter (fs) used to add force in Langevin method.
-- **Default**: 1
+- **Default**: 1.0
 
 ### DFT+U correction
 

examples/performance/P105_si512_lcao/result.ref

@@ -1,5 +1,5 @@
-etotref 
-etotperatomref 
-totalforceref 0.0
-totalstressref 0.0
-totaltimeref 
+etotref -57838.0048262096606777
+etotperatomref -112.9648531762
+totalforceref 27.819520
+totalstressref 5537.133126
+totaltimeref +2495.54754

examples/performance/run.sh

@@ -37,8 +37,8 @@ check_out() {
         cal=$(grep "$key" $outfile | awk '{printf "%.'$ca'f\n",$2}')
         ref=$(grep "$key" result.ref | awk '{printf "%.'$ca'f\n",$2}')
         deviation=$(awk 'BEGIN {x='$ref';y='$cal';if (x<y) {a=y-x} else {a=x-y};printf "%.'$ca'f\n",a}')
-        deviation1=$(awk 'BEGIN{print '$deviation'*(10**'$ca')}')
-
+        deviation1=$(awk 'BEGIN{print '$deviation'*(10^'$ca')}')
+        
         if [ $key == "totaltimeref" ]; then
             break
         fi
@@ -80,7 +80,7 @@ run_abacus() {
         lastword=$(tail -1 result.log | awk '{print $1}')
     fi
     if [[ $lastword != "SEE" ]]; then
-        /usr/bin/time -v mpirun -n $1 -env OMP_NUM_THREADS=$2 $abacus 2>time.log ｜ tee result.log
+        OMP_NUM_THREADS=$2 /usr/bin/time -v mpirun -n $1 $abacus 2>time.log | tee result.log
     else
         printf "**result.log is normal end, skip this job** "
     fi

examples/performance/sumdat.sh

@@ -13,9 +13,9 @@ fi
 test -f $outf && rm $outf
 
 #title
-printf "%20s %7s %8s %8s %6s %6s %8s %8s %8s %8s %8s %8s %8s %8s %8s %8s %8s %8s %8s\n" \
-"example" "Natoms" "EneCut" "k-points" "NProc" "Niter" "TotTime" "1stSCF" "SCF/iter" \
-"Run%" "c_bands%" "s_bands%" "h_psi%" "vloc%" "vnl%" "FFT%" "stress%" "force%" "MaxResSize" > $outf
+printf "%20s %15s %7s %8s %8s %6s %6s %8s %8s %8s %8s %8s %8s %8s %8s %8s %8s\n" \
+"example" "ks_solver" "Natoms" "EneCut" "k-points" "NProc" "Niter" "TotTime" "1stSCF" "SCF/iter" \
+"Run%" "h_psi%" "vloc%" "vnl%" "stress%" "force%" "MaxResSize" > $outf
 
 for i in `cat $allcase`;do
     if [[ ! -f $i/result.log ]];then
@@ -25,6 +25,7 @@ for i in `cat $allcase`;do
     fi
 
     basis=`awk '$1=="basis_type"{print $2}' ${i}/INPUT | tr [A-Z] [a-z]`
+    solver=`grep ks_solver ${i}/OUT.*/INPUT | awk '{print $2}'`
     #echo $basis
     if [[ "$basis" == "pw" ]];then
         natoms=`sed -n '/ELEMENT NATOM/,/----/'p ${i}/result.log| sed  '1d;$d' | awk 'BEGIN{a=0}{a+=$2}END{print a}'`
@@ -36,13 +37,13 @@ for i in `cat $allcase`;do
         scf1=`grep -A 1 "ITER   ETOT(eV)" ${i}/result.log | awk 'END{printf"%.2f", $NF}'`
         totalscf=`awk '$2=="Run"{print $3}' ${i}/result.log`
         scfpiter=`awk -v a=$totalscf -v b=$scf1 -v c=$niter 'BEGIN{printf"%.2f",(a-b)/(c-1)}'`
-        fft=`awk '$2=="FFT3D"{printf"%.1f",$6}' ${i}/result.log`
+        #fft=`awk '$2=="FFT3D"{printf"%.1f",$6}' ${i}/result.log`
         hpsi=`awk '$2=="h_psi"{printf"%.1f",$6}' ${i}/result.log`
         vloc=`awk '$2=="vloc"{printf"%.1f",$6}' ${i}/result.log`
         vnl=`awk '$2=="vnl"{printf"%.1f",$6}' ${i}/result.log`
         sc=`awk '$2=="Run"{printf"%.1f",$6}' ${i}/result.log`
-        cbands=`awk '$2=="c_bands"{printf"%.1f",$6}' ${i}/result.log`
-        sbands=`awk '$2=="sum_band"{printf"%.1f",$6}' ${i}/result.log`
+        #cbands=`awk '$2=="c_bands"{printf"%.1f",$6}' ${i}/result.log`
+        #sbands=`awk '$2=="sum_band"{printf"%.1f",$6}' ${i}/result.log`
         stress=`awk '$2=="cal_stress"{printf"%.1f",$6}' ${i}/result.log`
         force=`awk '$2=="cal_force_nl"{printf"%.1f",$6}' ${i}/result.log`
     elif [[ "$basis" == "lcao" ]];then
@@ -55,13 +56,13 @@ for i in `cat $allcase`;do
         scf1=`grep -A 1 "ITER   ETOT(eV)" ${i}/result.log | awk 'END{printf"%.2f", $NF}'`
         totalscf=`awk '$1=="Run"{print $3}' ${i}/result.log`
         scfpiter=`awk -v a=$totalscf -v b=$scf1 -v c=$niter 'BEGIN{printf"%.2f",(a-b)/(c-1)}'`
-        fft="-"
+        #fft="-"
         hpsi="-"
         vloc=`awk '$2=="vlocal"{printf"%.1f",$6}' ${i}/result.log`
         vnl="-"
         sc=`awk '$2=="Run"{printf"%.1f",$6}' ${i}/result.log`
-        cbands=`awk '$2=="cal_bands"{printf"%.1f",$6}' ${i}/result.log`
-        sbands=`awk '$2=="sum_bands"{printf"%.1f",$6}' ${i}/result.log`
+        #cbands=`awk '$2=="cal_bands"{printf"%.1f",$6}' ${i}/result.log`
+        #sbands=`awk '$2=="sum_bands"{printf"%.1f",$6}' ${i}/result.log`
         stress=`awk '$2=="evaluate_vl_stress"{printf"%.1f",$6}' ${i}/result.log`
         force=`awk '$2=="evaluate_vl_force"{printf"%.1f",$6}' ${i}/result.log`
     else
@@ -70,8 +71,8 @@ for i in `cat $allcase`;do
     fi
     maxres=`grep "Maximum resident set size" ${i}/time.log | awk '{print $NF}'`
 
-    printf "%20s %7s %8s %8s %6s %6s %8s %8s %8s %8s %8s %8s %8s %8s %8s %8s %8s %8s %s\n" \
-    $i $natoms $encut $kpt $nproc $niter $tottime $scf1 $scfpiter $sc $cbands $sbands $hpsi $vloc $vnl $fft \
+    printf "%20s %15s %7s %8s %8s %6s %6s %8s %8s %8s %8s %8s %8s %8s %8s %8s %s\n" \
+    $i $solver $natoms $encut $kpt $nproc $niter $tottime $scf1 $scfpiter $sc $hpsi $vloc $vnl \
     $stress $force $maxres >> $outf
 
 done

source/CMakeLists.txt

@@ -13,6 +13,7 @@ add_subdirectory(module_surchem)
 add_subdirectory(module_deepks)
 add_subdirectory(module_xc)
 add_subdirectory(module_esolver)
+add_subdirectory(module_gint)
 add_subdirectory(src_io)
 add_subdirectory(src_ions)
 add_subdirectory(src_lcao)