switch to C++17

* Require C++17 in CMake * Remove nvcc version below 11 from GitHub CI * Remove CUDA SDK installer below 9.2 from install scripts * Update supported CUDA versions in README.md * Eliminate obsolete C++ version checks * Remove support for nvcc + clang-5 * Update documentation * Fix warnings * Merge GitLab CI CUDA job description files * Add a GitLab CI test with clang-12 as CUDA compiler and cuda11.0 * Add GitHub CI test with clang-12 + nvcc 11.4 * support passing CMAKE_CUDA_FLAGS variable through CI
alpaka-group · Nov 19, 2021 · 4467954 · 4467954
1 parent aa35987
commit 4467954
Show file tree

Hide file tree

Showing 19 changed files with 98 additions and 181 deletions.
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
@@ -28,6 +28,5 @@ variables:
 
 include:
   - local: '/script/gitlabci/job_base.yml'
-  - local: '/script/gitlabci/job_cuda9.2.yml'
-  - local: '/script/gitlabci/job_cuda11.4.yml'
+  - local: '/script/gitlabci/job_cuda.yml'
   - local: '/script/gitlabci/job_hip4.2.yml'
diff --git a/.zenodo.json b/.zenodo.json
@@ -1,6 +1,6 @@
 {
   "title": "alpaka: Abstraction Library for Parallel Kernel Acceleration",
-  "description": "The alpaka library is a header-only C++14 abstraction library for accelerator development. Its aim is to provide performance portability across accelerators through the abstraction (not hiding!) of the underlying levels of parallelism.",
+  "description": "The alpaka library is a header-only C++17 abstraction library for accelerator development. Its aim is to provide performance portability across accelerators through the abstraction (not hiding!) of the underlying levels of parallelism.",
   "creators": [
     {
       "affiliation": "Helmholtz-Zentrum Dresden-Rossendorf, TU Dresden, LogMeIn Inc.",

diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -30,7 +30,7 @@ string(REGEX MATCH "([0-9]+)" ALPAKA_VERSION_PATCH  ${ALPAKA_VERSION_PATCH_HPP})
 set(PACKAGE_VERSION "${ALPAKA_VERSION_MAJOR}.${ALPAKA_VERSION_MINOR}.${ALPAKA_VERSION_PATCH}")
 
 project(alpaka VERSION      ${ALPAKA_VERSION_MAJOR}.${ALPAKA_VERSION_MINOR}.${ALPAKA_VERSION_PATCH}
-               DESCRIPTION  "The alpaka library is a header-only C++14 abstraction library for accelerator development."
+               DESCRIPTION  "The alpaka library is a header-only C++17 abstraction library for accelerator development."
                HOMEPAGE_URL "https://github.com/alpaka-group/alpaka"
                LANGUAGES    CXX)
 

diff --git a/README.md b/README.md
@@ -5,13 +5,13 @@
 [![Continuous Integration](https://github.com/alpaka-group/alpaka/workflows/Continuous%20Integration/badge.svg)](https://github.com/alpaka-group/alpaka/actions?query=workflow%3A%22Continuous+Integration%22)
 [![Documentation Status](https://readthedocs.org/projects/alpaka/badge/?version=latest)](https://alpaka.readthedocs.io)
 [![Doxygen](https://img.shields.io/badge/API-Doxygen-blue.svg)](https://alpaka-group.github.io/alpaka)
-[![Language](https://img.shields.io/badge/language-C%2B%2B14-orange.svg)](https://isocpp.org/)
+[![Language](https://img.shields.io/badge/language-C%2B%2B17-orange.svg)](https://isocpp.org/)
 [![Platforms](https://img.shields.io/badge/platform-linux%20%7C%20windows%20%7C%20mac-lightgrey.svg)](https://github.com/alpaka-group/alpaka)
 [![License](https://img.shields.io/badge/license-MPL--2.0-blue.svg)](https://www.mozilla.org/en-US/MPL/2.0/)
 
 ![alpaka](docs/logo/alpaka_401x135.png)
 
-The **alpaka** library is a header-only C++14 abstraction library for accelerator development.
+The **alpaka** library is a header-only C++17 abstraction library for accelerator development.
 
 Its aim is to provide performance portability across accelerators through the abstraction (not hiding!) of the underlying levels of parallelism.
 
@@ -68,20 +68,20 @@ Accelerator Back-ends
 Supported Compilers
 -------------------
 
-This library uses C++14 (or newer when available).
-
-|Accelerator Back-end|gcc 7.5 <br/> (Linux)|gcc 8.5 <br/> (Linux)|gcc 9.4 <br/> (Linux)|gcc 10.3 <br/> (Linux)|gcc 11.1 <br/> (Linux)|clang 5/6/7/8 <br/> (Linux)|clang 9 <br/> (Linux)|clang 10 <br/> (Linux)|clang 11 <br/> (Linux)|clang 12 <br/> (Linux)|Apple LLVM 11.3.1/12.4.0/12.5.1/13.0.0 <br /> (macOS)|MSVC 2019 <br/> (Windows)|
-|---|---|---|---|---|---|---|---|---|---|---|---|---|
-|Serial|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|
-|OpenMP 2.0+ blocks|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:x:|:white_check_mark:|
-|OpenMP 2.0+ threads|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:x:|:white_check_mark:|
-|OpenMP 5.0 (CPU)|:x:|:x:|:x:|:x:|:x:|:x:|:x:|:x:|:white_check_mark:|:white_check_mark:|:x:|:x:|
-| std::thread |:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|
-| Boost.Fiber |:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:x:|:white_check_mark:|
-|TBB|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:x:|
-|CUDA (nvcc)|:white_check_mark: <br/> (CUDA 9.2-11.4) |:white_check_mark: <br/> (CUDA 10.1-11.4) |:white_check_mark: <br/> (CUDA 11.0-11.4)|:x:|:x:|:white_check_mark: <br/> (CUDA 10.1-11.4)|:white_check_mark: <br/> (CUDA 11.0-11.4)|:white_check_mark: <br/> (CUDA 11.1-11.4)|:white_check_mark: <br/> (CUDA 11.4)| - |:x:|:white_check_mark: <br/> (CUDA 10.1,10.2,11.2,11.3, 11.4)|
-|CUDA (clang) | - | - | - | - | - | - | :white_check_mark: <br/> (CUDA 9.2-10.1) | :white_check_mark: <br/> (CUDA 9.2-10.1) | :white_check_mark: <br/> (CUDA 10.0-10.2) | - | - | - |
-|[HIP-4.0.1](https://alpaka.readthedocs.io/en/latest/install/HIP.html) (clang)|:x:|:x:|:x:|:x:|:x:|:x:|:x:|:x:|:x:|:white_check_mark:| - | - |
+This library uses C++17 (or newer when available).
+
+|Accelerator Back-end|gcc 7.5 <br/> (Linux)|gcc 8.5 <br/> (Linux)|gcc 9.4 <br/> (Linux)|gcc 10.3 <br/> (Linux)|gcc 11.1 <br/> (Linux)|clang 5-9 <br/> (Linux)|clang 10 <br/> (Linux)|clang 11 <br/> (Linux)|clang 12 <br/> (Linux)|Apple LLVM 11.3.1/12.4.0/12.5.1/13.0.0 <br /> (macOS)|MSVC 2019 <br/> (Windows)|
+|---|---|---|---|---|---|---|---|---|---|---|---|
+|Serial|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|
+|OpenMP 2.0+ blocks|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:x:|:white_check_mark:|
+|OpenMP 2.0+ threads|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:x:|:white_check_mark:|
+|OpenMP 5.0 (CPU)|:x:|:x:|:x:|:x:|:x:|:x:|:x:|:white_check_mark:|:white_check_mark:|:x:|:x:|
+| std::thread |:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|
+| Boost.Fiber |:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:x:|:white_check_mark:|
+|TBB|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:white_check_mark:|:x:|
+|CUDA (nvcc)|:white_check_mark: <br/> (CUDA 11.0-11.4) |:white_check_mark: <br/> (CUDA 11.0-11.4) |:white_check_mark: <br/> (CUDA 11.0-11.4)|:x:|:x:|:white_check_mark: <br/> (CUDA 11.0-11.4)|:white_check_mark: <br/> (CUDA 11.1-11.4)|:white_check_mark: <br/> (CUDA 11.4)|:white_check_mark: <br/> (CUDA 11.4)|:x:|:white_check_mark: <br/> (CUDA 11.2-11.4)|
+|CUDA (clang) | - | - | - | - | - | - | :x: | :white_check_mark: <br/> (CUDA 9.2-10.1) | :white_check_mark: <br/> (CUDA 10.0-10.2) | - | - |
+|[HIP-4.0.1](https://alpaka.readthedocs.io/en/latest/install/HIP.html) (clang)|:x:|:x:|:x:|:x:|:x:|:x:|:x:|:x:|:white_check_mark:| - | - |
 
 Other compilers or combinations marked with :x: in the table above may work but are not tested in CI and are therefore not explicitly supported.
 
@@ -92,10 +92,10 @@ Dependencies
 The **alpaka** library itself just requires header-only libraries.
 However some of the accelerator back-end implementations require different boost libraries to be built.
 
-When an accelerator back-end using *Boost.Fiber* is enabled, `boost-fiber` and all of its dependencies are required to be built in C++14 mode `./b2 cxxflags="-std=c++14"`.
-When *Boost.Fiber* is enabled and alpaka is built in C++17 mode with clang and libstc++, Boost >= 1.67.0 is required.
+When an accelerator back-end using *Boost.Fiber* is enabled, `boost-fiber` and all of its dependencies are required to be built in C++17 mode `./b2 cxxflags="-std=c++17"`.
+When *Boost.Fiber* is enabled and alpaka is built with clang and libstc++, Boost >= 1.67.0 is required.
 
-When an accelerator back-end using *CUDA* is enabled, version *9.0* of the *CUDA SDK* is the minimum requirement.
+When an accelerator back-end using *CUDA* is enabled, version *11.0* (with nvcc as CUDA compiler) or version *9.2* (with clang as CUDA compiler) of the *CUDA SDK* is the minimum requirement.
 *NOTE*: When using nvcc as *CUDA* compiler, the *CUDA accelerator back-end* can not be enabled together with the *Boost.Fiber accelerator back-end* due to bugs in the nvcc compiler.
 *NOTE*: When using clang as a native *CUDA* compiler, the *CUDA accelerator back-end* can not be enabled together with the *Boost.Fiber accelerator back-end* or any *OpenMP accelerator back-end* because this combination is currently unsupported.
 *NOTE*: Separable compilation is disabled by default and can be enabled via the CMake flag `CMAKE_CUDA_SEPARABLE_COMPILATION`.
@@ -109,7 +109,7 @@ Usage
 -----
 
 The library is header only so nothing has to be built.
-CMake 3.15+ is required to provide the correct defines and include paths.
+CMake 3.18+ is required to provide the correct defines and include paths.
 Just call `ALPAKA_ADD_EXECUTABLE` instead of `CUDA_ADD_EXECUTABLE` or `ADD_EXECUTABLE` and the difficulties of the CUDA nvcc compiler in handling `.cu` and `.cpp` files are automatically taken care of.
 Source files do not need any special file ending.
 Examples of how to utilize alpaka within CMake can be found in the `example` folder.

diff --git a/cmake/alpakaCommon.cmake b/cmake/alpakaCommon.cmake
@@ -125,14 +125,14 @@ endif()
 set(ALPAKA_DEBUG "0" CACHE STRING "Debug level")
 set_property(CACHE ALPAKA_DEBUG PROPERTY STRINGS "0;1;2")
 
-set(ALPAKA_CXX_STANDARD_DEFAULT "14")
+set(ALPAKA_CXX_STANDARD_DEFAULT "17")
 # Check whether ALPAKA_CXX_STANDARD has already been defined as a non-cached variable.
 if(DEFINED ALPAKA_CXX_STANDARD)
     set(ALPAKA_CXX_STANDARD_DEFAULT ${ALPAKA_CXX_STANDARD})
 endif()
 
 set(ALPAKA_CXX_STANDARD ${ALPAKA_CXX_STANDARD_DEFAULT} CACHE STRING "C++ standard version")
-set_property(CACHE ALPAKA_CXX_STANDARD PROPERTY STRINGS "14;17;20")
+set_property(CACHE ALPAKA_CXX_STANDARD PROPERTY STRINGS "17;20")
 
 if(NOT TARGET alpaka)
     add_library(alpaka INTERFACE)
@@ -415,11 +415,10 @@ if(ALPAKA_ACC_GPU_CUDA_ENABLE)
                 endif()
             endif()
 
-            # NOTE: Since CUDA 10.2 this option is also alternatively called '--extended-lambda'
             if(ALPAKA_CUDA_EXPT_EXTENDED_LAMBDA STREQUAL ON)
-                alpaka_set_compiler_options(DEVICE target alpaka $<$<COMPILE_LANGUAGE:CUDA>:--expt-extended-lambda>)
+                alpaka_set_compiler_options(DEVICE target alpaka $<$<COMPILE_LANGUAGE:CUDA>:--extended-lambda>)
             endif()
-            # This is mandatory because with c++14 many standard library functions we rely on are constexpr (std::min, std::multiplies, ...)
+            # This is mandatory because with c++17 many standard library functions we rely on are constexpr (std::min, std::multiplies, ...)
             alpaka_set_compiler_options(DEVICE target alpaka $<$<COMPILE_LANGUAGE:CUDA>:--expt-relaxed-constexpr>)
 
             if((CMAKE_BUILD_TYPE STREQUAL "Debug") OR (CMAKE_BUILD_TYPE STREQUAL "RelWithDebInfo"))
@@ -455,15 +454,6 @@ if(ALPAKA_ACC_GPU_CUDA_ENABLE)
             # avoids warnings on host-device signatured, default constructors/destructors
             alpaka_set_compiler_options(DEVICE target alpaka $<$<COMPILE_LANGUAGE:CUDA>:-Xcudafe=--diag_suppress=esa_on_defaulted_function_ignored>)
 
-            # avoids warnings on host-device signature of 'std::__shared_count<>'
-            if(CUDAToolkit_VERSION VERSION_EQUAL 10.0)
-                alpaka_set_compiler_options(DEVICE target alpaka $<$<COMPILE_LANGUAGE:CUDA>:-Xcudafe=--diag_suppress=2905>)
-            elseif(CUDAToolkit_VERSION VERSION_EQUAL 10.1)
-                alpaka_set_compiler_options(DEVICE target alpaka $<$<COMPILE_LANGUAGE:CUDA>:-Xcudafe=--diag_suppress=2912>)
-            elseif(CUDAToolkit_VERSION VERSION_EQUAL 10.2)
-                alpaka_set_compiler_options(DEVICE target alpaka $<$<COMPILE_LANGUAGE:CUDA>:-Xcudafe=--diag_suppress=2976>)
-            endif()
-
             if(ALPAKA_CUDA_KEEP_FILES STREQUAL ON)
                 alpaka_set_compiler_options(DEVICE target alpaka $<$<COMPILE_LANGUAGE:CUDA>:--keep>)
             endif()

diff --git a/docs/source/basic/library.rst b/docs/source/basic/library.rst
@@ -4,7 +4,7 @@ Library Interface
 As described in the chapter about the :doc:`Abstraction </basic/abstraction>`, the general design of the library is very similar to *CUDA* and *OpenCL* but extends both by some points, while not requiring any language extensions.
 General interface design as well as interface implementation decisions differentiating *alpaka* from those libraries are described in the Rationale section.
 It uses C++ because it is one of the most performant languages available on nearly all systems.
-Furthermore, C++14 allows to describe the concepts in a very abstract way that is not possible with many other languages.
+Furthermore, C++17 allows to describe the concepts in a very abstract way that is not possible with many other languages.
 The *alpaka* library extensively makes use of advanced functional C++ template meta-programming techniques.
 The Implementation Details section discusses the C++ library and the way it provides extensibility and optimizability.
 

diff --git a/docs/source/conf.py b/docs/source/conf.py
@@ -126,7 +126,7 @@
     (master_doc, 'alpaka', u'alpaka Documentation',
      author, 'alpaka', 'Abstraction Library for Parallel Kernel Acceleration',
      """
-     The alpaka library is a header-only C++14 abstraction library for
+     The alpaka library is a header-only C++17 abstraction library for
      accelerator development. Its aim is to provide performance portability
      across accelerators through the abstraction (not hiding!) of the underlying
      levels of parallelism.

diff --git a/include/alpaka/core/UniformCudaHip.hpp b/include/alpaka/core/UniformCudaHip.hpp
@@ -12,7 +12,7 @@
 #if defined(ALPAKA_ACC_GPU_CUDA_ENABLED) || defined(ALPAKA_ACC_GPU_HIP_ENABLED)
 
 #    include <alpaka/core/BoostPredef.hpp>
-
+#    include <alpaka/core/Unused.hpp>
 
 #    if defined(ALPAKA_ACC_GPU_CUDA_ENABLED) && !BOOST_LANG_CUDA
 #        error If ALPAKA_ACC_GPU_CUDA_ENABLED is set, the compiler has to support CUDA!
@@ -63,7 +63,7 @@ namespace alpaka
 #    endif
                     ALPAKA_DEBUG_BREAK;
                     // reset the last error to allow user side error handling
-                    ALPAKA_API_PREFIX(GetLastError)();
+                    ignore_unused(ALPAKA_API_PREFIX(GetLastError)());
                     throw std::runtime_error(sError);
                 }
             }
@@ -90,7 +90,7 @@ namespace alpaka
                     else
                     {
                         // reset the last error to avoid propagation to the next CUDA/HIP API call
-                        ALPAKA_API_PREFIX(GetLastError)();
+                        ignore_unused(ALPAKA_API_PREFIX(GetLastError)());
                     }
                 }
             }

diff --git a/include/alpaka/kernel/TaskKernelGpuUniformCudaHipRt.hpp b/include/alpaka/kernel/TaskKernelGpuUniformCudaHipRt.hpp
@@ -39,6 +39,7 @@
 // Implementation details.
 #    include <alpaka/acc/AccGpuUniformCudaHipRt.hpp>
 #    include <alpaka/core/Decay.hpp>
+#    include <alpaka/core/Unused.hpp>
 #    include <alpaka/dev/DevUniformCudaHipRt.hpp>
 #    include <alpaka/kernel/Traits.hpp>
 #    include <alpaka/queue/QueueUniformCudaHipRtBlocking.hpp>
@@ -423,7 +424,7 @@ namespace alpaka
                 // Wait for the kernel execution to finish but do not check error return of this call.
                 // Do not use the alpaka::wait method because it checks the error itself but we want to give a custom
                 // error message.
-                ALPAKA_API_PREFIX(StreamSynchronize)(queue.m_spQueueImpl->m_UniformCudaHipQueue);
+                ignore_unused(ALPAKA_API_PREFIX(StreamSynchronize)(queue.m_spQueueImpl->m_UniformCudaHipQueue));
 #    if ALPAKA_DEBUG >= ALPAKA_DEBUG_MINIMAL
                 std::string const msg(
                     "'execution of kernel: '" + std::string(typeid(TKernelFnObj).name()) + "' failed with");

diff --git a/include/alpaka/pltf/PltfUniformCudaHipRt.hpp b/include/alpaka/pltf/PltfUniformCudaHipRt.hpp
@@ -12,6 +12,7 @@
 #if defined(ALPAKA_ACC_GPU_CUDA_ENABLED) || defined(ALPAKA_ACC_GPU_HIP_ENABLED)
 
 #    include <alpaka/core/BoostPredef.hpp>
+#    include <alpaka/core/Unused.hpp>
 
 #    if defined(ALPAKA_ACC_GPU_CUDA_ENABLED) && !BOOST_LANG_CUDA
 #        error If ALPAKA_ACC_GPU_CUDA_ENABLED is set, the compiler has to support CUDA!
@@ -148,7 +149,7 @@ namespace alpaka
                     // Return the previous error from cudaStreamCreate.
                     ALPAKA_UNIFORM_CUDA_HIP_RT_CHECK(rc);
                     // Reset the Error state.
-                    ALPAKA_API_PREFIX(GetLastError)();
+                    ignore_unused(ALPAKA_API_PREFIX(GetLastError)());
                     return false;
                 }
             }

diff --git a/script/before_install.sh b/script/before_install.sh
@@ -119,16 +119,13 @@ then
     then
         if [ ! -z "${ALPAKA_CXX_STANDARD+x}" ]
         then
-            if (( "${ALPAKA_CXX_STANDARD}" >= 17 ))
+            if [ "${ALPAKA_CI_INSTALL_FIBERS}" == "ON" ]
             then
-                if [ "${ALPAKA_CI_INSTALL_FIBERS}" == "ON" ]
+                if (( ( ( "${ALPAKA_CI_BOOST_BRANCH_MAJOR}" == 1 ) && ( "${ALPAKA_CI_BOOST_BRANCH_MINOR}" < 67 ) ) || ( "${ALPAKA_CI_BOOST_BRANCH_MAJOR}" < 1 ) ))
                 then
-                    if (( ( ( "${ALPAKA_CI_BOOST_BRANCH_MAJOR}" == 1 ) && ( "${ALPAKA_CI_BOOST_BRANCH_MINOR}" < 67 ) ) || ( "${ALPAKA_CI_BOOST_BRANCH_MAJOR}" < 1 ) ))
-                    then
-                        # https://github.com/boostorg/coroutine2/issues/26
-                        echo "libstdc++ in c++17 mode is not compatible with boost.fibers in boost-1.66 and below."
-                        exit 1
-                    fi
+                    # https://github.com/boostorg/coroutine2/issues/26
+                    echo "libstdc++ in c++17 mode is not compatible with boost.fibers in boost-1.66 and below."
+                    exit 1
                 fi
             fi
         fi

diff --git a/script/docker_ci.sh b/script/docker_ci.sh
@@ -119,6 +119,10 @@ then
     then
         ALPAKA_DOCKER_ENV_LIST+=("--env" "CMAKE_CUDA_ARCHITECTURES=${CMAKE_CUDA_ARCHITECTURES}")
     fi
+    if [ ! -z "${CMAKE_CUDA_FLAGS+x}" ]
+    then
+        ALPAKA_DOCKER_ENV_LIST+=("--env" "CMAKE_CUDA_FLAGS=${CMAKE_CUDA_FLAGS}")
+    fi
 fi
 ALPAKA_DOCKER_ENV_LIST+=("--env" "ALPAKA_CI_INSTALL_HIP=${ALPAKA_CI_INSTALL_HIP}")
 if [ "${ALPAKA_CI_INSTALL_HIP}" == "ON" ]