Squashed 'thirdParty/alpaka/' changes from 6b42f8920..d5e59590f

d5e59590f Merge pull request ComputationalRadiationPhysics#862 from jkrude/math/abs-trait-fix
a9abefc2b Resolved CI-Errors missing Type
8fc692a8b Merge pull request ComputationalRadiationPhysics#861 from BenjaminW3/topic-clang-9
79fd778e9 Update include/alpaka/math/abs/AbsCudaBuiltIn.hpp
684fe2923 Add support for clang-9.0.0
41e1f3bd7 Merge pull request ComputationalRadiationPhysics#863 from BenjaminW3/topic-mac-update
9f0fd066e Fix MacOS builds and add support for newer versions
2ce0b99a7 Added trait specialisation for double.
f3d363bc5 Merge pull request ComputationalRadiationPhysics#848 from jkrude/KernelExecutionFixture-missing-pragma
bbac90b28 Added missing "#pragma once"
82d3294e0 Merge pull request ComputationalRadiationPhysics#846 from BenjaminW3/topic-cuda_attributes
d68debe99 Merge pull request ComputationalRadiationPhysics#844 from psychocoderHPC/fix-nonCudaExeptionsCanNotBeHandled
f61d20143 Use cudaDeviceGetAttribute instead of slow cudaGetDeviceProperties
4a66dfbaa fix: cuda exeptions
d395aa340 Merge pull request ComputationalRadiationPhysics#841 from tdd11235813/pr-dlr-patch
29e9e718f Fixes GCC/Address Sanitizer false-positive issue.
ed8fe79cd Merge pull request ComputationalRadiationPhysics#839 from BenjaminW3/topic-cudaErrorIllegalInstruction
bd5a5b293 Disable nvstd::function test due to cudaErrorIllegalInstruction crashes
d7c807c73 Merge pull request ComputationalRadiationPhysics#837 from jkrude/overload-math-functions
6f7357e99 changed typo in PowHipBuiltIn
929bca049 Merge pull request ComputationalRadiationPhysics#835 from jkrude/axpy-rnd
974d805b0 fixed RoundHipBuildIn and absHipBuiltIn bugs
8a60f36aa fixed SinCudaBuiltIn bug
808fe1cc3 fixed and added doxygen-comments
84298239f specified float template
a56e6cc05 Update test/integ/axpy/src/axpy.cpp
667c5b7b6 added seed to random number generation
35733682f Merge pull request ComputationalRadiationPhysics#830 from jkrude/warning-fix
5d82440b7 Merge pull request ComputationalRadiationPhysics#831 from jkrude/bug-fix
d7ea75d2a corrected max -> min
d6c4e9fc8 integrated requested changes
cd75e6f3a /math/fmod && /math/remainder incomplete type bug-fix
881766e89 muted cuda-warning
7217755fb Merge pull request ComputationalRadiationPhysics#834 from BenjaminW3/topic-CUDA-MSVC
04789982b Workaround hanging CUDA windows installer removal
b2b2acf28 Merge pull request ComputationalRadiationPhysics#829 from DerWaldschrat/develop
df9f0329e Merge pull request ComputationalRadiationPhysics#832 from tdd11235813/pr-rocm270
62e5f9dd6 Updates travis/HIP to use rocm2.7.0.
ef722be69 Fix ComputationalRadiationPhysics#750 by adding a dummy bBusyWaiting parameter to the EventCpu constructor.
5d942aa02 Merge pull request ComputationalRadiationPhysics#819 from BenjaminW3/topic-cmake-3-15
eb2875052 Use CMake 3.15 in CI
fe28f1b20 Merge pull request ComputationalRadiationPhysics#822 from psychocoderHPC/topic-supportExternQueues
ad4d3cd9e github annotations
1b12af0c7 Merge pull request ComputationalRadiationPhysics#825 from BenjaminW3/topic-CI-versions
41ce889fe Add support for boost-1.71.0
ee43ce45b Add gcc 9 support to readme
b0a40a118 Merge pull request ComputationalRadiationPhysics#824 from BenjaminW3/topic-#823
fc471e161 Enable types with alignemnt higher than defaultAlignment for static shared memory
b4ffe419d Merge pull request ComputationalRadiationPhysics#821 from BenjaminW3/feature/benjamino/topic-clean-cuda-math-includes
ca9d658e8 remove queue shared pointer specializations
61b1b0b80 add test case for a user defined queue
17554f54d support for queues defined outside of alpaka
a903bfd6c Add missing header files to IDE projects
1567c4649 Clean up includes
fce472be5 Remove unnecessary inconsistent use of ALPAKA_NO_HOST_ACC_WARNING for AcosCudaBuiltIn
d9503ef20 Merge pull request ComputationalRadiationPhysics#814 from BenjaminW3/topic-reduce-test-includes
b40a68cbc Suppress MSVC CUDA_ONLY_MODE warning
72b53ba3c Merge pull request ComputationalRadiationPhysics#818 from ax3l/fix-mandelbrotConst
3d963cd81 Merge pull request ComputationalRadiationPhysics#816 from tdd11235813/pr-hip-race-fix
00853f4af Mandelbrot: const inline member
a236d4623 HIP/HCC non-blocking queue destructor waits for tasks.
925b03fc0 Reduces HIP wait-workaround sleep-timer from 100ms to 10ms
e89a0be54 Fixes race-condition HIP/NVCC queue (unneeded workaround removed)
9c8b1760b Merge pull request ComputationalRadiationPhysics#813 from BenjaminW3/topic-gcc-9
b72ebcd1b Add gcc-9 to CI testing
e2b43279a Clean up the unit test includes
b0ca1fc8c Move definition of BOOST_COMP_CLANG_CUDA to BoostPredef.hpp
3779102a4 Merge pull request ComputationalRadiationPhysics#812 from jkelling/gcc9.1noexcept
123530422 Add noexcept to some ctors
c64add43b Merge pull request ComputationalRadiationPhysics#809 from ax3l/topic-defTplDepth
4c30ca85d Template Depth: C++11 Default (1024)
58ea948c5 Merge pull request ComputationalRadiationPhysics#808 from tdd11235813/pr-logofix
653fb5f13 Fixes alpaka logo (removes white color in ear).
fd301ce74 Merge pull request ComputationalRadiationPhysics#807 from BenjaminW3/topic-std-function-kernel
e13547b15 Add unit test for nvstd::function as kernel
5b5676825 Merge pull request ComputationalRadiationPhysics#806 from BenjaminW3/topc_nvcc_flags_compile
f60ad8d5c Add unit tests for std::function and std::bind kernels and remove from examples
0f408376a Merge pull request ComputationalRadiationPhysics#803 from tdd11235813/pr-atomics-accs
91d9e0d18 Fix EXPT_EXTENDED_LAMBDA and EXPT_RELAXED_CONSTEXPR usage
722eff132 Suppresses a strict-overflow warning for gcc7.4.0.
e2523a075 No more disables atomics' tests in the CI.
297298a5a Atomics test reduced to 1D size_t accelerator.
01dfba843 Merge pull request ComputationalRadiationPhysics#804 from tdd11235813/pr-hiprand-fix
548dd8bab Fixes hiprand issue in travis CI.
6f43249c2 Merge pull request ComputationalRadiationPhysics#800 from tdd11235813/pr-hip-fixes
8988328a1 queue::detail ***Blocking -> ***BlockingImpl (CUDA, HIP).
e8e5ccbbf Adds HIP nvcc9.0+ requirement. Fixes typo (build->built).
d80cd6994 Makes enqueue thread-safe for Hip(HCC).
abf6c5310 Renames QueueHipRt*Impl -> QueueHipRt*.
786966160 Fixes HIP_PLATFORM and travis issues, rocRAND install into run.sh.
acdcb00e9 Re-enables use of numeric limits for HIP(HCC).
bdb45186f Adds missing ALPAKA_FN_HOST_ACC attributes.
e65ee1fd7 Updates information on required and tested HIP version.
a9a5518ff Fixes HIP(HCC) warnings.
e2c105bcf Replaces deprecated hipKernelLaunch.
300ae2467 Removes old AMD GPU target to fix HCC issue with time test.
9bdeb8574 Merge pull request ComputationalRadiationPhysics#580 from BenjaminW3/topic-separable-compilation
ca3dc6f1f Add option to enable separable compilation for nvcc via ALPAKA_CUDA_NVCC_SEPARABLE_COMPILATION
d437c3cee Merge pull request ComputationalRadiationPhysics#799 from BenjaminW3/topic-raise-minimum-CMake-version
b647a54b0 Raise minimum supported CMake version
936b59e6c Merge pull request ComputationalRadiationPhysics#797 from psychocoderHPC/fix-blockingEnqueue2
776a66a80 Merge pull request ComputationalRadiationPhysics#798 from psychocoderHPC/fix-travisGobblingWarning
aa5494028 fix BlockingQueue multi-thread support
ab2c0c75f Merge pull request ComputationalRadiationPhysics#796 from psychocoderHPC/fix-cudaWarning
b48df64ec travis: fix gobbling warning
9a944cda8 Merge pull request ComputationalRadiationPhysics#795 from BenjaminW3/topic-CI-version-updates
4c390a8f4 Add clang 8 + nvcc-10.1 support
5f308814f fix cuda warning
5b818aec6 Update CMake 3.14.0 to 3.14.4
4f53b4c05 Remove ESan build option as it has not been used and removed from clang 9.0
fe7b5d88e Update CUDA 10.1 to Update 1 and add clang 8 build
ce5d60ea2 Merge pull request ComputationalRadiationPhysics#793 from psychocoderHPC/topic-queueRenaming
493e1c702 update source code comments
7b2fd9152 rename queues
210ef7239 fix cmake for tests
c08e0df3e Merge pull request ComputationalRadiationPhysics#792 from psychocoderHPC/fix-cmake
064384863 fix cmake for tests

git-subtree-dir: thirdParty/alpaka
git-subtree-split: d5e59590f7649f09575a363d02582424b471aba6

CMakeLists.txt

@@ -11,7 +11,7 @@
 ################################################################################
 # Required CMake version
 
-cmake_minimum_required(VERSION 3.11.0)
+cmake_minimum_required(VERSION 3.11.4)
 
 project("alpakaAll")
 

README.md

@@ -57,26 +57,26 @@ Accelerator Back-ends
 | Boost.Fiber | boost::fibers::fiber |Host CPU (single core)|sequential|parallel (cooperative multitasking)|
 |TBB|TBB 2.2+|Host CPU (multi core)|parallel (preemptive multitasking)|sequential (only 1 thread per block)|
 |CUDA|CUDA 8.0-10.0|NVIDIA GPUs|parallel (undefined)|parallel (lock-step within warps)|
-|HIP(nvcc)|[HIP 1.5.8292+](https://github.com/ROCm-Developer-Tools/HIP)|NVIDIA GPUs SM 2.0+|parallel (undefined)|parallel (lock-step within warps)|
+|HIP(nvcc)|[HIP 1.5+](https://github.com/ROCm-Developer-Tools/HIP)|NVIDIA GPUs SM 2.0+|parallel (undefined)|parallel (lock-step within warps)|
 
 
 Supported Compilers
 -------------------
 
 This library uses C++11 (or newer when available).
 
-|Accelerator Back-end|gcc 4.9.4 <br/> (Linux)|gcc 5.5 <br/> (Linux)|gcc 6.4/7.3 <br/> (Linux)|gcc 8.1 <br/> (Linux)|clang 4 <br/> (Linux)|clang 5 <br/> (Linux)|clang 6 <br/> (Linux)|clang 7 <br/> (Linux)|clang 8 <br/> (Linux)|Apple LLVM 10.0.0 <br/> (macOS)|MSVC 2017.9 <br/> (Windows)|
+|Accelerator Back-end|gcc 4.9.4 <br/> (Linux)|gcc 5.5 <br/> (Linux)|gcc 6.4/7.3 <br/> (Linux)|gcc 8.1/9.1 <br/> (Linux)|clang 4 <br/> (Linux)|clang 5 <br/> (Linux)|clang 6 <br/> (Linux)|clang 7 <br/> (Linux)|clang 8 <br/> (Linux)|clang 9 <br/> (Linux)|Apple LLVM 10.2-11.2 <br/> (macOS)|MSVC 2017.9 <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 4.0+ (CPU)|: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:|: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:|:white_check_mark:|
-|CUDA (nvcc)|:white_check_mark: <br/> (CUDA 8.0-10.1)|:white_check_mark: <br/> (CUDA 9.0-10.1)|:white_check_mark: <br/> (CUDA 9.2-10.1) |:x:|:white_check_mark: <br/> (CUDA 9.1-10.1)|:white_check_mark: <br/> (CUDA 10.1)|:white_check_mark: <br/> (CUDA 10.1)|:white_check_mark: <br/> (CUDA 10.1)|:x:|:x:|:white_check_mark: <br/> (CUDA 10.0-10.1)|
-|CUDA (clang) | - | - | - | - | :white_check_mark: <br/> (CUDA 8.0)| :white_check_mark: <br/> (CUDA 8.0)| :white_check_mark: <br/> (CUDA 8.0-9.0) | :white_check_mark: <br/> (CUDA 8.0-9.2) | :white_check_mark: <br/> (CUDA 8.0-10.0) | - | - |
-|[HIP](doc/markdown/user/implementation/mapping/HIP.md) (nvcc)|:white_check_mark: <br/> (nvcc 8.0)|:x:|:x:|:x:|:x:|:x:|:x:|:x:|:x:|:x:|:x:|
+|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 4.0+ (CPU)|: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:|: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:|:white_check_mark:|
+|CUDA (nvcc)|:white_check_mark: <br/> (CUDA 8.0-10.1)|:white_check_mark: <br/> (CUDA 9.0-10.1)|:white_check_mark: <br/> (CUDA 9.2-10.1) |:x:|:white_check_mark: <br/> (CUDA 9.1-10.1)|:white_check_mark: <br/> (CUDA 10.1)|:white_check_mark: <br/> (CUDA 10.1)|:white_check_mark: <br/> (CUDA 10.1)|:white_check_mark: <br/> (CUDA 10.1)|:x:|:x:|:white_check_mark: <br/> (CUDA 10.0-10.1)|
+|CUDA (clang) | - | - | - | - | :white_check_mark: <br/> (CUDA 8.0)| :white_check_mark: <br/> (CUDA 8.0)| :white_check_mark: <br/> (CUDA 8.0-9.0) | :white_check_mark: <br/> (CUDA 8.0-9.2) | :white_check_mark: <br/> (CUDA 8.0-10.0) | :white_check_mark: <br/> (CUDA 9.2-10.1) | - | - |
+|[HIP](doc/markdown/user/implementation/mapping/HIP.md) (nvcc)|:white_check_mark: <br/> (nvcc 9.0+)|:x:|:x:|:x:|:x:|:x:|:x:|:x:|:x:|:x:|:x:|:x:|
 
 
 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.
@@ -88,12 +88,13 @@ 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 build in C++11 mode `./b2 cxxflags="-std=c++11"`.
-When *Boost.Fiber* is enabled and alpaka is build 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++11 mode `./b2 cxxflags="-std=c++11"`.
+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 *CUDA* is enabled, version *8.0* 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 any *OpenMP accelerator back-end* because this combination is currently unsupported.
+*NOTE*: Separable compilation is only supported when using nvcc, not with clang as native *CUDA* compiler. It is disabled by default and can be enabled via the CMake flag `ALPAKA_CUDA_NVCC_SEPARABLE_COMPILATION`.
 
 When an accelerator back-end using *OpenMP* is enabled, the compiler and the platform have to support the corresponding minimum *OpenMP* version.
 
@@ -103,8 +104,8 @@ When an accelerator back-end using *TBB* is enabled, the compiler and the platfo
 Usage
 -----
 
-The library is header only so nothing has to be build.
-CMake 3.11.0+ is required to provide the correct defines and include paths.
+The library is header only so nothing has to be built.
+CMake 3.11.4+ 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.