Squashed 'thirdParty/cupla/' changes from a96b293dc..4578ae151

4578ae151 Merge pull request ComputationalRadiationPhysics#128 from psychocoderHPC/topic-updateAlpakaToCurrent0.4.0dev 0c090e536 Merge commit 'dd8afa5cec90a1398b8da3f9cd7d30aa59664cb5' into topic-updateAlpakaToCurrent0.4.0dev dd8afa5ce Squashed 'alpaka/' changes from 0a2b6161..d5e59590 767768a88 rename alpaka queue names 8c227fbb6 Merge pull request ComputationalRadiationPhysics#130 from psychocoderHPC/fix-configHeaderTest a6c3da536 fix `config header` test a439c9d6b Merge pull request ComputationalRadiationPhysics#129 from psychocoderHPC/topic-updateTravisCMakeTo3.15.4 2e0daaf1b update tavis CMake to 3.15.4 04698e9c6 Merge pull request ComputationalRadiationPhysics#123 from SimeonEhrig/ci_enable_backends 25e64b623 time command add to ci test c559bf432 Merge pull request ComputationalRadiationPhysics#107 from SimeonEhrig/init-gitlab-ci 488a2e95f Merge pull request ComputationalRadiationPhysics#117 from tdd11235813/pr-fixes-atomics c2916be68 Adds atomicAnd, atomicXor, atomicOr. 7c330a5a9 Merge pull request ComputationalRadiationPhysics#112 from psychocoderHPC/topic-standaloneHeader d84f2fdea configuration header 6a97c1cf1 Merge pull request ComputationalRadiationPhysics#113 from fwyzard/dev_cuplaGetErrorName d210a37a0 Add cuplaGetErrorName with the same functionality as cuplaGetErrorString d050bf3cd Merge pull request ComputationalRadiationPhysics#111 from tdd11235813/doc-subtree 19cf46913 Describes how to pull subtree as generic git author. 4ecc143f4 Merge pull request ComputationalRadiationPhysics#103 from tdd11235813/dev 16c806e68 Merge pull request ComputationalRadiationPhysics#104 from tdd11235813/dev-targets 43fd52319 Add gitlab-ci.yml for GitLab CI tests fd37e7b7d Alpaka as git subtree instead of git submodule. 16b455107 Squashed 'alpaka/' content from commit 0a2b6161 69b79bd95 Merge commit '16b4551075f93cdd71309433b02936a152dfee9f' as 'alpaka' fa5cc5260 Removes alpaka git module. 67485d510 Some stylistic changes. 4f2b9c019 CUPLA_ADD_EXECUTABLE links cupla target. Target provides c++11. git-subtree-dir: thirdParty/cupla git-subtree-split: 4578ae151ee6ac5c7810f76ff0c3a4506158807d
psychocoderHPC · Oct 25, 2019 · 9c93042 · 9c93042
1 parent c98c0a0
commit 9c93042
Show file tree

Hide file tree

Showing 487 changed files with 91,278 additions and 163 deletions.
diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
@@ -0,0 +1,113 @@
+.base_job:
+  script:
+    # the default build type is Release
+    # if neccesary, you can rerun the pipeline with another build type-> https://docs.gitlab.com/ee/ci/pipelines.html#manually-executing-pipelines
+    # to change the build type, you must set the environment variable CUPLA_BUILD_TYPE
+    - if [[ ! -v CUPLA_BUILD_TYPE ]] ; then
+        CUPLA_BUILD_TYPE=Release ;
+      fi
+    - echo "number of processor threads $(nproc)"
+    - $CXX --version
+    - cmake --version
+    # print boost version
+    - echo -e "#include <boost/version.hpp>\n#include <iostream>\nint main() { std::cout << BOOST_VERSION << std::endl; return 0; }" | $CXX -x c++ - -o boost_version >/dev/null || { echo 0; }
+    - echo "Boost version $(./boost_version)"
+    - export cupla_DIR=$CI_PROJECT_DIR
+    # use one build directory for all build configurations
+    - mkdir build
+    - cd build
+    - echo "Build type-> $CUPLA_BUILD_TYPE"
+    # ALPAKA_ACCS contains the backends, which are used for each build
+    # the backends are set in the sepcialized base jobs .base_gcc,.base_clang and.base_cuda
+    - for CMAKE_FLAGS in $ALPAKA_ACCS ; do
+        echo "###################################################"
+        && echo "# Example Matrix Multiplication (adapted original)"
+        && echo "###################################################"
+        && echo "can not run with CPU_B_SEQ_T_SEQ due to missing elements layer in original SDK example"
+        && echo "CPU_B_SEQ_T_OMP2/THREADS too many threads necessary (256)"
+        && if [[ $CMAKE_FLAGS =~ -*DALPAKA_ACC_GPU_CUDA_ENABLE=ON.* ]]; then
+          cmake $cupla_DIR/example/CUDASamples/matrixMul/ $CMAKE_FLAGS -DCMAKE_BUILD_TYPE=$CUPLA_BUILD_TYPE
+          && make -j
+          && time ./matrixMul -wA=64 -wB=64 -hA=64 -hB=64
+          && rm -r * ;
+        fi
+        && echo "###################################################"
+        && echo "# Example Async API (adapted original)"
+        && echo "###################################################"
+        && echo "can not run with CPU_B_SEQ_T_SEQ due to missing elements layer in original SDK example"
+        && echo "CPU_B_SEQ_T_OMP2/THREADS too many threads necessary (512)"
+        && if [[ $CMAKE_FLAGS =~ -*DALPAKA_ACC_GPU_CUDA_ENABLE=ON.* ]]; then
+          cmake $cupla_DIR/example/CUDASamples/asyncAPI/ $CMAKE_FLAGS -DCMAKE_BUILD_TYPE=$CUPLA_BUILD_TYPE
+          && make -j
+          && time ./asyncAPI
+          && rm -r * ;
+        fi
+        && echo "###################################################"
+        && echo "# Example Async API (added elements layer)"
+        && echo "###################################################"
+        && cmake $cupla_DIR/example/CUDASamples/asyncAPI_tuned/ $CMAKE_FLAGS -DCMAKE_BUILD_TYPE=$CUPLA_BUILD_TYPE
+        && make -j
+        && time ./asyncAPI_tuned
+        && rm -r *
+        && echo "###################################################"
+        && echo "Example vectorAdd (added elements layer)"
+        && echo "###################################################"
+        && cmake $cupla_DIR/example/CUDASamples/vectorAdd/ $CMAKE_FLAGS -DCMAKE_BUILD_TYPE=$CUPLA_BUILD_TYPE
+        && make -j
+        && time ./vectorAdd 100000
+        && rm -r * ;
+      done
+
+.base_gcc:
+  variables:
+    GIT_SUBMODULE_STRATEGY: normal
+    CXX: g++
+    CC: gcc
+    ALPAKA_ACCS: "-DALPAKA_ACC_CPU_B_SEQ_T_SEQ_ENABLE=ON
+                  -DALPAKA_ACC_CPU_B_SEQ_T_OMP2_ENABLE=ON
+                  -DALPAKA_ACC_CPU_B_OMP2_T_SEQ_ENABLE=ON"
+                  # -DALPAKA_ACC_CPU_B_SEQ_T_THREADS_ENABLE=ON
+  extends: .base_job
+  # x86_64 tag is used to get a multi-core CPU for the tests
+  tags:
+    - x86_64
+
+.base_clang:
+  variables:
+    GIT_SUBMODULE_STRATEGY: normal
+    CXX: clang++
+    CC: clang
+    ALPAKA_ACCS: "-DALPAKA_ACC_CPU_B_SEQ_T_SEQ_ENABLE=ON
+                  -DALPAKA_ACC_CPU_B_OMP2_T_SEQ_ENABLE=ON"
+                  # -DALPAKA_ACC_CPU_B_SEQ_T_OMP2_ENABLE=ON
+                  # -DALPAKA_ACC_CPU_B_SEQ_T_THREADS_ENABLE=ON
+  extends: .base_job
+  # x86_64 tag is used to get a multi-core CPU for the tests
+  tags:
+    - x86_64
+
+.base_cuda:
+  variables:
+    GIT_SUBMODULE_STRATEGY: normal
+    CXX: g++
+    CC: gcc
+    ALPAKA_ACCS: "-DALPAKA_ACC_GPU_CUDA_ENABLE=ON"
+  before_script:
+    - nvidia-smi
+    - nvcc --version
+  extends: .base_job
+  tags:
+    - cuda
+    - intel
+
+gcc7:
+  image: registry.gitlab.com/hzdr/cupla-docker/gcc7:latest
+  extends: .base_gcc
+
+clang7:
+  image: registry.gitlab.com/hzdr/cupla-docker/clang7:latest
+  extends: .base_clang
+
+cuda9:
+  image: registry.gitlab.com/hzdr/cupla-docker/cuda9:latest
+  extends: .base_cuda
diff --git a/.gitmodules b/.gitmodules
diff --git a/.travis.yml b/.travis.yml
@@ -39,6 +39,7 @@ before_install:
   - mkdir -p $HOME/asyncAPI
   - mkdir -p $HOME/asyncAPI_tuned
   - mkdir -p $HOME/vectorAdd
+  - mkdir -p $HOME/test/config
   - export CMAKE_FLAGS="-DALPAKA_ACC_"$STRATEGY"_ENABLE=ON"
   - if [ "$COMPILER" == "gcc" ]; then
         echo "Using g++-4.9 and sequential OpenMP2 threads ...";
@@ -66,15 +67,15 @@ install:
   # CMAKE                                                                     #
   #############################################################################
   - export PATH=$CMAKE_ROOT/bin:$PATH
-  - CMAKE_311_FOUND=$(cmake --version | grep " 3\.11\." >/dev/null && { echo 0; } || { echo 1; })
-  - if [ $CMAKE_311_FOUND -ne 0 ]; then
+  - CMAKE_315_FOUND=$(cmake --version | grep " 3\.15\." >/dev/null && { echo 0; } || { echo 1; })
+  - if [ $CMAKE_315_FOUND -ne 0 ]; then
       mkdir -p $CMAKE_ROOT &&
       cd $CMAKE_ROOT &&
       rm -rf $CMAKE_ROOT/* &&
-      travis_retry wget --no-check-certificate http://cmake.org/files/v3.11/cmake-3.11.0-Linux-x86_64.tar.gz &&
-      tar -xzf cmake-3.11.0-Linux-x86_64.tar.gz &&
-      mv cmake-3.11.0-Linux-x86_64/* . &&
-      rm -rf cmake-3.11.0-Linux-x86_64.tar.gz cmake-3.11.0-Linux-x86_64 &&
+      travis_retry wget --no-check-certificate http://cmake.org/files/v3.15/cmake-3.15.4-Linux-x86_64.tar.gz &&
+      tar -xzf cmake-3.15.4-Linux-x86_64.tar.gz &&
+      mv cmake-3.15.4-Linux-x86_64/* . &&
+      rm -rf cmake-3.15.4-Linux-x86_64.tar.gz cmake-3.15.4-Linux-x86_64 &&
       cd -;
     fi
   - cmake --version
@@ -155,5 +156,9 @@ script:
       ./vectorAdd 100000;
     fi
   #############################################################################
-  # Test: (To do: add tests in $TRAVIS_BUILD_DIR/test/                        #
+  # Test: additional tests                                                    #
   #############################################################################
+  - cd $HOME/test/config
+  - if [[ $CXX =~ "^g\+\+" ]] || [[ "$COMPILER" == "nvcc" ]] ; then
+      $TRAVIS_BUILD_DIR/test/system/config/test.sh $CXX;
+    fi
diff --git a/INSTALL.md b/INSTALL.md
@@ -16,22 +16,19 @@ Requirements
     - `mkdir -p $HOME/src`
     - `git clone git://github.com/ComputationalRadiationPhysics/cupla.git $HOME/src/cupla`
     - `cd $HOME/src/cupla`
-    - `git submodule init`
-    - `git submodule update`
     - `export CUPLA_ROOT=$HOME/src/cupla`
-  - use cupla without the submodule alpaka: 
-    Set the advanced CMake variable `cupla_ALPAKA_PROVIDER` to `extern` and
+  - use a different alpaka installation:
     set environment variable `ALPAKA_ROOT` or extend `CMAKE_PREFIX_PATH` with the
     path to alpaka.
-
 
-compile an example
------------------
+
+Compile an example
+------------------
 
 - create build directory `mkdir -p buildCuplaExample`
 - `cd buildCuplaExample`
 - `cmake $CUPLA_ROOT/example/CUDASamples/matrixMul -D<ACC_TYPE>=ON`
-    - list of supported ACC_TYPES
+    - list of supported `ACC_TYPE`s
         - `ALPAKA_ACC_CPU_B_SEQ_T_OMP2_ENABLE`
         - `ALPAKA_ACC_CPU_B_SEQ_T_THREADS_ENABLE`
         - `ALPAKA_ACC_CPU_B_TBB_T_SEQ_ENABLE`
@@ -41,3 +38,29 @@ compile an example
           see [TuningGuide.md](doc/TuningGuide.md)
 - `make -j`
 - `./matrixMul -wA=320 -wB=320 -hA=320 -hB=320` (parameters must be a multiple of 32!)
+
+
+How to update alpaka as git subtree?
+------------------------------------
+
+```zsh
+# git author is generic to not mess up contribution statistics
+GIT_AUTHOR_NAME="Third Party" GIT_AUTHOR_EMAIL="[email protected]" \
+ git subtree pull --prefix alpaka \
+ https://github.com/ComputationalRadiationPhysics/alpaka.git develop --squash
+```
+
+**How to commit local changes to alpaka upstream?**
+
+If your local alpaka version contains changes you want to contribute back upstream via fork, then you can use `git subtree push`:
+
+```zsh
+# Add your fork of alpaka to git remotes
+git remote add alpaka-fork [email protected]:YOUR_NAME/alpaka.git
+
+# Push your changes to your fork
+git subtree push --prefix=alpaka alpaka-fork
+```
+Then check your github page of your fork to open a pull request upstream.
+
+More information can be found in this [git subtree guide](https://www.atlassian.com/blog/git/alternatives-to-git-submodule-git-subtree).
diff --git a/README.md b/README.md
@@ -37,9 +37,8 @@ Dependencies
 ------------
 
 - **cmake 3.11.0**
-- **alpaka** 
-  - is loaded as `git submodule` within **cupla** (see [INSTALL.md](INSTALL.md))
-  - for more information please read [README.md](https://github.com/ComputationalRadiationPhysics/alpaka/blob/master/README.md)
+- **[alpaka](https://github.com/ComputationalRadiationPhysics/alpaka/)**
+  - alpaka is loaded as `git subtree` within **cupla**, see [INSTALL.md](INSTALL.md)
 
 Usage
 -----
@@ -49,13 +48,15 @@ Usage
 - Checkout the [tuning guide](doc/TuningGuide.md) for a step further to performance
   portable code.
 
+[cupla can be used as a header-only library and without the CMake build system](doc/ConfigurationHeader.md)
 
 Authors
 -------
 
 ### Maintainers and core developers
 
 - Rene Widera
+- Matthias Werner
 
 ### Former Members, Contributions and Thanks
 

diff --git a/alpaka b/alpaka
diff --git a/alpaka/.dockerignore b/alpaka/.dockerignore
@@ -0,0 +1 @@
+.git
diff --git a/alpaka/.gitignore b/alpaka/.gitignore
@@ -0,0 +1,18 @@
+/doc/doxygen/*
+!/doc/doxygen/Doxyfile
+!/doc/doxygen/alpaka_doxygen.png
+/doc/latex/*
+**/build
+
+# tmp files
+*~
+
+# netbeans project files
+/nbproject/
+
+# Code::Blocks project files
+/*.cbp
+/*.layout
+
+# original backup files
+*.orig