mirror/eigen
2
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2024-12-27 07:29:52 +08:00
Code Issues Packages Projects Releases Wiki Activity
1fd5ce1002
eigen/unsupported/test/CMakeLists.txt

420 lines
15 KiB
CMake
Raw Normal View History

Simplify handling and non-splitted tests and include split_test_helper.h instead of re-generating it. This also allows us to modify it without breaking existing build folder.
2018-07-17 00:55:40 +08:00
# The file split_test_helper.h was generated at first run,
# it is now included in test/
if(EXISTS ${CMAKE_CURRENT_BINARY_DIR}/split_test_helper.h)
file(REMOVE ${CMAKE_CURRENT_BINARY_DIR}/split_test_helper.h)
Fix Eigen's building of sharded tests that use CUDA & more igamma/igammac bugfixes. 0. Prior to this PR, not a single sharded CUDA test was actually being *run*. Fixed that. GPU tests are still failing for igamma/igammac. 1. Add calls for igamma/igammac to TensorBase 2. Fix up CUDA-specific calls of igamma/igammac 3. Add unit tests for digamma, igamma, igammac in CUDA.
2016-03-08 06:08:56 +08:00
endif()
add the possibility to disable OpenGL testing
2010-10-13 02:23:52 +08:00
Add Official/Unsupported labels to unit tests and add a ctest driver to submit subprojects to cdash
2013-03-20 15:40:13 +08:00
set_property(GLOBAL PROPERTY EIGEN_CURRENT_SUBPROJECT "Unsupported")
add_custom_target(BuildUnsupported)
Removed the sycl include from Eigen/Core and moved it to Unsupported/Eigen/CXX11/Tensor; added TensorReduction for sycl (full reduction and partial reduction); added TensorReduction test case for sycl (full reduction and partial reduction); fixed the tile size on TensorSyclRun.h based on the device max work group size;
2016-11-05 02:18:19 +08:00
include_directories(../../test ../../unsupported ../../Eigen
Add directory containing split_test_helper.h to include path.
2011-03-29 21:17:49 +08:00
${CMAKE_CURRENT_BINARY_DIR}/../../test)
add a proof of concept autodiff jacobian helper class based on adolc with unit test and FindAdolc cmake module
2009-02-28 00:19:13 +08:00
Use full packet size for Dynamic-sized objects (otherwise, the unalignedcount unit test fails with AVX enabled)
2015-09-03 04:51:43 +08:00
find_package (Threads)
Removed the sycl include from Eigen/Core and moved it to Unsupported/Eigen/CXX11/Tensor; added TensorReduction for sycl (full reduction and partial reduction); added TensorReduction test case for sycl (full reduction and partial reduction); fixed the tile size on TensorSyclRun.h based on the device max work group size;
2016-11-05 02:18:19 +08:00
split the Sparse module into multiple ones, and move non stable parts to unsupported/ (see the ML for details)
2010-06-18 17:28:30 +08:00
find_package(GoogleHash)
Eliminate CMake FindPackageHandleStandardArgs warnings. CMake complains that the package name does not match when the case differs, e.g.: ``` CMake Warning (dev) at /usr/share/cmake-3.18/Modules/FindPackageHandleStandardArgs.cmake:273 (message): The package name passed to `find_package_handle_standard_args` (UMFPACK) does not match the name of the calling package (Umfpack). This can lead to problems in calling code that expects `find_package` result variables (e.g., `_FOUND`) to follow a certain pattern. Call Stack (most recent call first): cmake/FindUmfpack.cmake:50 (find_package_handle_standard_args) bench/spbench/CMakeLists.txt:24 (find_package) This warning is for project developers. Use -Wno-dev to suppress it. ``` Here we rename the libraries to match their true cases.
2021-02-24 12:16:52 +08:00
if(GoogleHash_FOUND)
split the Sparse module into multiple ones, and move non stable parts to unsupported/ (see the ML for details)
2010-06-18 17:28:30 +08:00
add_definitions("-DEIGEN_GOOGLEHASH_SUPPORT")
include_directories(${GOOGLEHASH_INCLUDES})
ei_add_property(EIGEN_TESTED_BACKENDS "GoogleHash, ")
STYLE: Remove CMake-language block-end command arguments Ancient versions of CMake required else(), endif(), and similar block termination commands to have arguments matching the command starting the block. This is no longer the preferred style.
2019-11-01 00:36:27 +08:00
else()
split the Sparse module into multiple ones, and move non stable parts to unsupported/ (see the ML for details)
2010-06-18 17:28:30 +08:00
ei_add_property(EIGEN_MISSING_BACKENDS "GoogleHash, ")
STYLE: Remove CMake-language block-end command arguments Ancient versions of CMake required else(), endif(), and similar block termination commands to have arguments matching the command starting the block. This is no longer the preferred style.
2019-11-01 00:36:27 +08:00
endif()
split the Sparse module into multiple ones, and move non stable parts to unsupported/ (see the ML for details)
2010-06-18 17:28:30 +08:00
Recent Adolc versions require C++11
2018-04-14 01:10:23 +08:00
add the possibility to disable OpenGL testing
2010-10-13 02:23:52 +08:00
find_package(Adolc)
Add missing adolc isinf/isnan. Also modified cmake/FindAdolc.cmake to eliminate warnings, and added search paths to match install layout. Fixed: #2157
2021-02-20 06:17:10 +08:00
if(Adolc_FOUND)
add a proof of concept autodiff jacobian helper class based on adolc with unit test and FindAdolc cmake module
2009-02-28 00:19:13 +08:00
include_directories(${ADOLC_INCLUDES})
add a support module for MPFR C++ with basic unit testing
2010-07-15 22:29:04 +08:00
ei_add_property(EIGEN_TESTED_BACKENDS "Adolc, ")
Recent Adolc versions require C++11
2018-04-14 01:10:23 +08:00
if(EIGEN_TEST_CXX11)
ei_add_test(forward_adolc "" ${ADOLC_LIBRARIES})
else()
message(STATUS "Adolc found, but tests require C++11 mode")
endif()
STYLE: Remove CMake-language block-end command arguments Ancient versions of CMake required else(), endif(), and similar block termination commands to have arguments matching the command starting the block. This is no longer the preferred style.
2019-11-01 00:36:27 +08:00
else()
add a support module for MPFR C++ with basic unit testing
2010-07-15 22:29:04 +08:00
ei_add_property(EIGEN_MISSING_BACKENDS "Adolc, ")
STYLE: Remove CMake-language block-end command arguments Ancient versions of CMake required else(), endif(), and similar block termination commands to have arguments matching the command starting the block. This is no longer the preferred style.
2019-11-01 00:36:27 +08:00
endif()
Add BVH module in unsupported (patch from Ilya Baran) (I thought I committed it a week ago but it seems the command failed)
2009-03-19 04:06:06 +08:00
disable NonLinearOptimization test until it's fixed
2010-08-18 21:11:01 +08:00
# this test seems to never have been successful on x87, so is considered to contain a FP-related bug.
# see thread: "non-linear optimization test summary"
Backed out changeset 363e506776babe4fdfb8e670fe6d91942fc61d7c
2012-12-08 03:53:19 +08:00
ei_add_test(NonLinearOptimization)
starting work on a Numerical differenciation module
2009-09-28 08:43:07 +08:00
ei_add_test(NumericalDiff)
Fixed bug #702 and added unit test. Thanks to Alexander Werner for the report.
2013-11-08 01:32:24 +08:00
ei_add_test(autodiff_scalar)
add an auto-diff module in unsupported. it is similar to adolc's forward mode but the advantage of using Eigen's expression template to compute the derivatives (unless you nest an AutoDiffScalar into an Eigen's matrix).
2009-04-01 22:43:37 +08:00
ei_add_test(autodiff)
disable BVH test on Clang++. Looks like there's a good reason why BVH is unsupported. It seems to have a very weird usage pattern, relying on an externally defined bounding_box function in a naive way.
2011-02-28 10:37:34 +08:00
Add BVH module in unsupported (patch from Ilya Baran) (I thought I committed it a week ago but it seems the command failed)
2009-03-19 04:06:06 +08:00
ei_add_test(BVH)
disable BVH test on Clang++. Looks like there's a good reason why BVH is unsupported. It seems to have a very weird usage pattern, relying on an externally defined bounding_box function in a naive way.
2011-02-28 10:37:34 +08:00
Reverted unintended changes introduced by a bad merge
2016-02-19 14:20:50 +08:00
ei_add_test(matrix_exponential)
Add support for matrix sine, cosine, sinh and cosh.
2010-01-12 02:05:30 +08:00
ei_add_test(matrix_function)
Reverted unintended changes introduced by a bad merge
2016-02-19 14:20:50 +08:00
ei_add_test(matrix_power)
Implement matrix square root for complex matrices. I hope to implement the real case soon, but it's a bit more complicated due to the 2-by-2 blocks in the real Schur decomposition.
2011-05-08 05:57:46 +08:00
ei_add_test(matrix_square_root)
add an AlignedVector3 module suitable for vectorization of 3D vectors
2009-05-04 01:19:19 +08:00
ei_add_test(alignedvector3)
Disable a few unit tests in unsupported
2014-09-01 23:35:58 +08:00
initial pass of FFT module -- includes complex 1-d case only
2009-05-19 12:21:04 +08:00
ei_add_test(FFT)
* fix include for case-sensitive filesystem * on GNU, clock_gettime requires linking -lrt
2009-12-02 23:03:16 +08:00
Initial fork of unsupported module EulerAngles.
2015-09-27 21:51:24 +08:00
ei_add_test(EulerAngles)
Require at least MPFR version 2.3.0, because we use mpfr_signbit. Code in FindMPFR.cmake is taken from FindEigen2.cmake .
2010-07-19 19:26:52 +08:00
find_package(MPFR 2.3.0)
Fix general linking issue for tests linking to multiple libs, and explicitly link mpfr_real test to GMP.
2010-10-17 23:04:43 +08:00
find_package(GMP)
bug #1606: Explicitly set the standard before find_package(StandardMathLibrary). Also replace EIGEN_COMPILER_SUPPORT_CXX11 in favor of EIGEN_COMPILER_SUPPORT_CPP11. Grafted manually from a4afa90d161faab385a77f0e2764fb13ff3b9484
2018-10-19 23:20:51 +08:00
if(MPFR_FOUND AND EIGEN_COMPILER_SUPPORT_CPP11)
mv the mpreal copy in its own folder
2011-08-19 21:08:29 +08:00
include_directories(${MPFR_INCLUDES} ./mpreal)
add a support module for MPFR C++ with basic unit testing
2010-07-15 22:29:04 +08:00
ei_add_property(EIGEN_TESTED_BACKENDS "MPFR C++, ")
Fix general linking issue for tests linking to multiple libs, and explicitly link mpfr_real test to GMP.
2010-10-17 23:04:43 +08:00
set(EIGEN_MPFR_TEST_LIBRARIES ${MPFR_LIBRARIES} ${GMP_LIBRARIES})
Enable mpreal unit test for C++11 compiler only
2015-10-28 00:35:54 +08:00
ei_add_test(mpreal_support "-std=c++11" "${EIGEN_MPFR_TEST_LIBRARIES}" )
add a support module for MPFR C++ with basic unit testing
2010-07-15 22:29:04 +08:00
else()
ei_add_property(EIGEN_MISSING_BACKENDS "MPFR C++, ")
endif()
fix bug #92 - we were doing stupid things when passing the list of libraries to link to.
2010-10-28 22:44:20 +08:00
ei_add_test(sparse_extra "" "")
split the Sparse module into multiple ones, and move non stable parts to unsupported/ (see the ML for details)
2010-06-18 17:28:30 +08:00
added FindFFTW, but I don't think it's right yet
2009-06-11 10:16:32 +08:00
find_package(FFTW)
if(FFTW_FOUND)
add an OpenGL module simplifying the way you can pass Eigen's objects to GL
2010-07-22 22:08:58 +08:00
ei_add_property(EIGEN_TESTED_BACKENDS "fftw, ")
improve FindFFTW cmake module
2012-03-15 22:18:22 +08:00
include_directories( ${FFTW_INCLUDES} )
still test fftw even if the binary for long double is not available
2011-01-28 23:54:01 +08:00
if(FFTWL_LIB)
fix fftw cmake stuff
2011-11-03 22:33:42 +08:00
ei_add_test(FFTW "-DEIGEN_FFTW_DEFAULT -DEIGEN_HAS_FFTWL" "${FFTW_LIBRARIES}" )
still test fftw even if the binary for long double is not available
2011-01-28 23:54:01 +08:00
else()
improve FindFFTW cmake module
2012-03-15 22:18:22 +08:00
ei_add_test(FFTW "-DEIGEN_FFTW_DEFAULT" "${FFTW_LIBRARIES}" )
still test fftw even if the binary for long double is not available
2011-01-28 23:54:01 +08:00
endif()
add an OpenGL module simplifying the way you can pass Eigen's objects to GL
2010-07-22 22:08:58 +08:00
else()
ei_add_property(EIGEN_MISSING_BACKENDS "fftw, ")
endif()
Disable testing of OpenGL by default. The `OpenGLSupport` module contains mostly deprecated features, and the test is highly GL context-dependent, relies on deprecated GLUT, and requires a display. Until the module is updated to support modern OpenGL and the test to use newer windowing frameworks (e.g. GLFW) it's probably best to disable the test by default. The test can be enabled with `cmake -DEIGEN_TEST_OPENGL=ON`. See #2053 for more details.
2020-11-13 08:07:55 +08:00
option(EIGEN_TEST_OPENGL "Enable OpenGL support in unit tests" OFF)
if(EIGEN_TEST_OPENGL)
more unsupported/ CMake fixes
2010-10-17 23:21:10 +08:00
find_package(OpenGL)
find_package(GLUT)
find_package(GLEW)
if(OPENGL_FOUND AND GLUT_FOUND AND GLEW_FOUND)
Adding missing OPENGL_LIBRARIES for openglsupport test. Also adding OpenGL include directories as a better pratice even though these are system include directories in most systems.
2014-12-04 14:18:47 +08:00
include_directories(${OPENGL_INCLUDE_DIR} ${GLUT_INCLUDE_DIR} ${GLEW_INCLUDE_DIRS})
more unsupported/ CMake fixes
2010-10-17 23:21:10 +08:00
ei_add_property(EIGEN_TESTED_BACKENDS "OpenGL, ")
Adding missing OPENGL_LIBRARIES for openglsupport test. Also adding OpenGL include directories as a better pratice even though these are system include directories in most systems.
2014-12-04 14:18:47 +08:00
set(EIGEN_GL_LIB ${GLUT_LIBRARIES} ${GLEW_LIBRARIES} ${OPENGL_LIBRARIES})
more unsupported/ CMake fixes
2010-10-17 23:21:10 +08:00
ei_add_test(openglsupport "" "${EIGEN_GL_LIB}" )
else()
ei_add_property(EIGEN_MISSING_BACKENDS "OpenGL, ")
endif()
properly report OpenGL as a disabled backend
2011-02-17 01:01:06 +08:00
else()
ei_add_property(EIGEN_MISSING_BACKENDS "OpenGL, ")
endif()
added FindFFTW, but I don't think it's right yet
2009-06-11 10:16:32 +08:00
remove use of GSL in polynomialsolver unit test
2012-04-11 15:48:01 +08:00
ei_add_test(polynomialsolver)
Creation of the Polynomials module with the following features: * convenient functions: - Horner and stabilized Horner evaluation - polynomial coefficients from a set of given roots - Cauchy bounds * a QR based polynomial solver
2010-03-25 10:15:05 +08:00
ei_add_test(polynomialutils)
Integrated spline class and simple spline fitting
2011-11-25 21:53:40 +08:00
ei_add_test(splines)
Adding a householder-GMRES implementation from Kolja Brix
2012-03-29 21:00:55 +08:00
ei_add_test(gmres)
Activate dgmres unit test
2018-07-02 18:54:14 +08:00
ei_add_test(dgmres)
moved MINRES to unsupported. Made unit test.
2012-09-24 22:47:38 +08:00
ei_add_test(minres)
Idrs iterative linear solver
2021-02-27 20:09:33 +08:00
ei_add_test(idrs)
Move work in progress Levenberg Marquardt module in unsupported
2012-12-09 01:22:23 +08:00
ei_add_test(levenberg_marquardt)
Update KroneckerProduct wrt evaluator changes
2014-09-19 04:08:49 +08:00
ei_add_test(kronecker_product)
Add Bessel functions to SpecialFunctions. - Split SpecialFunctions files in to a separate BesselFunctions file. In particular add: - Modified bessel functions of the second kind k0, k1, k0e, k1e - Bessel functions of the first kind j0, j1 - Bessel functions of the second kind y0, y1
2019-09-15 00:16:47 +08:00
ei_add_test(bessel_functions)
bug #1232: refactor special functions as a new SpecialFunctions module, currently in unsupported/.
2016-07-08 17:13:55 +08:00
ei_add_test(special_functions)
Added special_packetmath test and tweaked bounds on tests. Refactor shared packetmath code to header file. (Squashed from PR !38)
2020-01-11 18:31:21 +08:00
ei_add_test(special_packetmath "-DEIGEN_FAST_MATH=1")
C++11: add template metaprogramming helpers Create a new directory CXX11 under unsupported/Eigen that contains code that requires C++11. In that directory, add a few generic templates useful for any module relying on C++11. These templates may be included with #include <[unsupported/]Eigen/CXX11/Core>. At the moment, this will only provide templates in the Eigen::internal namespace.
2013-11-15 05:27:06 +08:00
if(EIGEN_TEST_CXX11)
Reducing the code by generalising sycl backend functions/structs.
2016-10-14 19:09:55 +08:00
if(EIGEN_TEST_SYCL)
[SYCL] Rebasing the SYCL support branch on top of the Einge upstream master branch. * Unifying all loadLocalTile from lhs and rhs to an extract_block function. * Adding get_tensor operation which was missing in TensorContractionMapper. * Adding the -D method missing from cmake for Disable_Skinny Contraction operation. * Wrapping all the indices in TensorScanSycl into Scan parameter struct. * Fixing typo in Device SYCL * Unifying load to private register for tall/skinny no shared * Unifying load to vector tile for tensor-vector/vector-tensor operation * Removing all the LHS/RHS class for extracting data from global * Removing Outputfunction from TensorContractionSkinnyNoshared. * Combining the local memory version of tall/skinny and normal tensor contraction into one kernel. * Combining the no-local memory version of tall/skinny and normal tensor contraction into one kernel. * Combining General Tensor-Vector and VectorTensor contraction into one kernel. * Making double buffering optional for Tensor contraction when local memory is version is used. * Modifying benchmark to accept custom Reduction Sizes * Disabling AVX optimization for SYCL backend on the host to allow SSE optimization to the host * Adding Test for SYCL * Modifying SYCL CMake
2019-11-28 18:08:54 +08:00
set(EIGEN_SYCL ON)
# Forward CMake options as preprocessor definitions
if(EIGEN_SYCL_USE_DEFAULT_SELECTOR)
add_definitions(-DEIGEN_SYCL_USE_DEFAULT_SELECTOR=${EIGEN_SYCL_USE_DEFAULT_SELECTOR})
endif()
if(EIGEN_SYCL_NO_LOCAL_MEM)
add_definitions(-DEIGEN_SYCL_NO_LOCAL_MEM=${EIGEN_SYCL_NO_LOCAL_MEM})
endif()
if(EIGEN_SYCL_LOCAL_MEM)
add_definitions(-DEIGEN_SYCL_LOCAL_MEM=${EIGEN_SYCL_LOCAL_MEM})
endif()
if(EIGEN_SYCL_MAX_GLOBAL_RANGE)
add_definitions(-DEIGEN_SYCL_MAX_GLOBAL_RANGE=${EIGEN_SYCL_MAX_GLOBAL_RANGE})
endif()
if(EIGEN_SYCL_LOCAL_THREAD_DIM0)
add_definitions(-DEIGEN_SYCL_LOCAL_THREAD_DIM0=${EIGEN_SYCL_LOCAL_THREAD_DIM0})
endif()
if(EIGEN_SYCL_LOCAL_THREAD_DIM1)
add_definitions(-DEIGEN_SYCL_LOCAL_THREAD_DIM1=${EIGEN_SYCL_LOCAL_THREAD_DIM1})
endif()
if(EIGEN_SYCL_REG_M)
add_definitions(-DEIGEN_SYCL_REG_M=${EIGEN_SYCL_REG_M})
endif()
if(EIGEN_SYCL_REG_N)
add_definitions(-DEIGEN_SYCL_REG_N=${EIGEN_SYCL_REG_N})
endif()
if(EIGEN_SYCL_USE_PROGRAM_CLASS)
add_definitions(-DEIGEN_SYCL_USE_PROGRAM_CLASS=${EIGEN_SYCL_USE_PROGRAM_CLASS})
endif()
if(EIGEN_SYCL_ASYNC_EXECUTION)
add_definitions(-DEIGEN_SYCL_ASYNC_EXECUTION=${EIGEN_SYCL_ASYNC_EXECUTION})
endif()
if(EIGEN_SYCL_DISABLE_SKINNY)
add_definitions(-DEIGEN_SYCL_DISABLE_SKINNY=${EIGEN_SYCL_DISABLE_SKINNY})
endif()
if(EIGEN_SYCL_DISABLE_DOUBLE_BUFFER)
add_definitions(-DEIGEN_SYCL_DISABLE_DOUBLE_BUFFER=${EIGEN_SYCL_DISABLE_DOUBLE_BUFFER})
endif()
if(EIGEN_SYCL_DISABLE_RANK1)
add_definitions(-DEIGEN_SYCL_DISABLE_RANK1=${EIGEN_SYCL_DISABLE_RANK1})
endif()
if(EIGEN_SYCL_DISABLE_SCALAR)
add_definitions(-DEIGEN_SYCL_DISABLE_SCALAR=${EIGEN_SYCL_DISABLE_SCALAR})
endif()
if(EIGEN_SYCL_DISABLE_GEMV)
add_definitions(-DEIGEN_SYCL_DISABLE_GEMV=${EIGEN_SYCL_DISABLE_GEMV})
endif()
if(EIGEN_SYCL_DISABLE_ARM_GPU_CACHE_OPTIMISATION)
add_definitions(-DEIGEN_SYCL_DISABLE_ARM_GPU_CACHE_OPTIMISATION=${EIGEN_SYCL_DISABLE_ARM_GPU_CACHE_OPTIMISATION})
endif()
Add support for triSYCL Eigen is now able to use triSYCL with EIGEN_SYCL_TRISYCL and TRISYCL_INCLUDE_DIR options Fix contraction kernel with correct nd_item dimension
2017-05-06 02:26:27 +08:00
if(EIGEN_SYCL_TRISYCL)
set(CMAKE_CXX_STANDARD 14)
set(STD_CXX_FLAG "-std=c++1z")
STYLE: Remove CMake-language block-end command arguments Ancient versions of CMake required else(), endif(), and similar block termination commands to have arguments matching the command starting the block. This is no longer the preferred style.
2019-11-01 00:36:27 +08:00
else()
[SYCL] Rebasing the SYCL support branch on top of the Einge upstream master branch. * Unifying all loadLocalTile from lhs and rhs to an extract_block function. * Adding get_tensor operation which was missing in TensorContractionMapper. * Adding the -D method missing from cmake for Disable_Skinny Contraction operation. * Wrapping all the indices in TensorScanSycl into Scan parameter struct. * Fixing typo in Device SYCL * Unifying load to private register for tall/skinny no shared * Unifying load to vector tile for tensor-vector/vector-tensor operation * Removing all the LHS/RHS class for extracting data from global * Removing Outputfunction from TensorContractionSkinnyNoshared. * Combining the local memory version of tall/skinny and normal tensor contraction into one kernel. * Combining the no-local memory version of tall/skinny and normal tensor contraction into one kernel. * Combining General Tensor-Vector and VectorTensor contraction into one kernel. * Making double buffering optional for Tensor contraction when local memory is version is used. * Modifying benchmark to accept custom Reduction Sizes * Disabling AVX optimization for SYCL backend on the host to allow SSE optimization to the host * Adding Test for SYCL * Modifying SYCL CMake
2019-11-28 18:08:54 +08:00
if(MSVC)
# Set the host and device compilers C++ standard to C++14. On Windows setting this to C++11
# can cause issues with the ComputeCpp device compiler parsing Visual Studio Headers.
set(CMAKE_CXX_STANDARD 14)
list(APPEND COMPUTECPP_USER_FLAGS -DWIN32)
else()
set(CMAKE_CXX_STANDARD 11)
list(APPEND COMPUTECPP_USER_FLAGS -Wall)
endif()
# The following flags are not supported by Clang and can cause warnings
# if used with -Werror so they are removed here.
if(COMPUTECPP_USE_COMPILER_DRIVER)
set(CMAKE_CXX_COMPILER ${ComputeCpp_DEVICE_COMPILER_EXECUTABLE})
string(REPLACE "-Wlogical-op" "" CMAKE_CXX_FLAGS ${CMAKE_CXX_FLAGS})
string(REPLACE "-Wno-psabi" "" CMAKE_CXX_FLAGS ${CMAKE_CXX_FLAGS})
string(REPLACE "-ansi" "" CMAKE_CXX_FLAGS ${CMAKE_CXX_FLAGS})
endif()
list(APPEND COMPUTECPP_USER_FLAGS
-DEIGEN_NO_ASSERTION_CHECKING=1
-no-serial-memop
-Xclang
-cl-mad-enable)
endif()
ei_add_test(cxx11_tensor_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_image_op_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_math_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_forced_eval_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_broadcast_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_device_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_reduction_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_morphing_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_shuffling_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_padding_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_builtins_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_contract_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_concatenation_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_reverse_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_convolution_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_striding_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_chipping_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_layout_swap_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_inflation_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_random_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_generator_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_patch_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_image_patch_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_volume_patch_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_argmax_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_custom_op_sycl ${STD_CXX_FLAG})
ei_add_test(cxx11_tensor_scan_sycl ${STD_CXX_FLAG})
set(EIGEN_SYCL OFF)
STYLE: Remove CMake-language block-end command arguments Ancient versions of CMake required else(), endif(), and similar block termination commands to have arguments matching the command starting the block. This is no longer the preferred style.
2019-11-01 00:36:27 +08:00
endif()
Use the CMAKE_CXX_STANDARD variable to turn on cxx11
2016-03-05 12:18:01 +08:00
Enabled the new threadpool tests
2016-04-15 07:44:10 +08:00
ei_add_test(cxx11_eventcount "-pthread" "${CMAKE_THREAD_LIBS_INIT}")
ei_add_test(cxx11_runqueue "-pthread" "${CMAKE_THREAD_LIBS_INIT}")
Added a test to validate the new non blocking thread pool
2016-05-11 01:49:34 +08:00
ei_add_test(cxx11_non_blocking_thread_pool "-pthread" "${CMAKE_THREAD_LIBS_INIT}")
Use the CMAKE_CXX_STANDARD variable to turn on cxx11
2016-03-05 12:18:01 +08:00
ei_add_test(cxx11_meta)
Drop support for c++03 in Eigen tensor. Get rid of some code used to emulate c++11 functionality with older compilers.
2019-10-19 07:42:00 +08:00
ei_add_test(cxx11_maxsizevector)
ei_add_test(cxx11_tensor_argmax)
ei_add_test(cxx11_tensor_assign)
ei_add_test(cxx11_tensor_block_access)
ei_add_test(cxx11_tensor_block_eval)
ei_add_test(cxx11_tensor_block_io)
ei_add_test(cxx11_tensor_broadcasting)
ei_add_test(cxx11_tensor_casts)
ei_add_test(cxx11_tensor_chipping)
ei_add_test(cxx11_tensor_comparisons)
ei_add_test(cxx11_tensor_concatenation)
ei_add_test(cxx11_tensor_const)
Use the CMAKE_CXX_STANDARD variable to turn on cxx11
2016-03-05 12:18:01 +08:00
ei_add_test(cxx11_tensor_contraction)
ei_add_test(cxx11_tensor_convolution)
Drop support for c++03 in Eigen tensor. Get rid of some code used to emulate c++11 functionality with older compilers.
2019-10-19 07:42:00 +08:00
ei_add_test(cxx11_tensor_custom_index)
ei_add_test(cxx11_tensor_custom_op)
ei_add_test(cxx11_tensor_dimension)
ei_add_test(cxx11_tensor_empty)
ei_add_test(cxx11_tensor_executor "-pthread" "${CMAKE_THREAD_LIBS_INIT}")
Use the CMAKE_CXX_STANDARD variable to turn on cxx11
2016-03-05 12:18:01 +08:00
ei_add_test(cxx11_tensor_expr)
Drop support for c++03 in Eigen tensor. Get rid of some code used to emulate c++11 functionality with older compilers.
2019-10-19 07:42:00 +08:00
ei_add_test(cxx11_tensor_fft)
Use the CMAKE_CXX_STANDARD variable to turn on cxx11
2016-03-05 12:18:01 +08:00
ei_add_test(cxx11_tensor_fixed_size)
Drop support for c++03 in Eigen tensor. Get rid of some code used to emulate c++11 functionality with older compilers.
2019-10-19 07:42:00 +08:00
ei_add_test(cxx11_tensor_forced_eval)
ei_add_test(cxx11_tensor_generator)
ei_add_test(cxx11_tensor_ifft)
ei_add_test(cxx11_tensor_image_patch)
ei_add_test(cxx11_tensor_index_list)
Use the CMAKE_CXX_STANDARD variable to turn on cxx11
2016-03-05 12:18:01 +08:00
ei_add_test(cxx11_tensor_inflation)
Drop support for c++03 in Eigen tensor. Get rid of some code used to emulate c++11 functionality with older compilers.
2019-10-19 07:42:00 +08:00
ei_add_test(cxx11_tensor_intdiv)
ei_add_test(cxx11_tensor_io)
ei_add_test(cxx11_tensor_layout_swap)
ei_add_test(cxx11_tensor_lvalue)
ei_add_test(cxx11_tensor_map)
ei_add_test(cxx11_tensor_math)
ei_add_test(cxx11_tensor_mixed_indices)
Use the CMAKE_CXX_STANDARD variable to turn on cxx11
2016-03-05 12:18:01 +08:00
ei_add_test(cxx11_tensor_morphing)
Drop support for c++03 in Eigen tensor. Get rid of some code used to emulate c++11 functionality with older compilers.
2019-10-19 07:42:00 +08:00
ei_add_test(cxx11_tensor_move)
ei_add_test(cxx11_tensor_notification "-pthread" "${CMAKE_THREAD_LIBS_INIT}")
ei_add_test(cxx11_tensor_of_complex)
ei_add_test(cxx11_tensor_of_const_values)
ei_add_test(cxx11_tensor_of_strings)
Use the CMAKE_CXX_STANDARD variable to turn on cxx11
2016-03-05 12:18:01 +08:00
ei_add_test(cxx11_tensor_padding)
ei_add_test(cxx11_tensor_patch)
Drop support for c++03 in Eigen tensor. Get rid of some code used to emulate c++11 functionality with older compilers.
2019-10-19 07:42:00 +08:00
ei_add_test(cxx11_tensor_random)
Use the CMAKE_CXX_STANDARD variable to turn on cxx11
2016-03-05 12:18:01 +08:00
ei_add_test(cxx11_tensor_reduction)
Drop support for c++03 in Eigen tensor. Get rid of some code used to emulate c++11 functionality with older compilers.
2019-10-19 07:42:00 +08:00
ei_add_test(cxx11_tensor_ref)
ei_add_test(cxx11_tensor_roundings)
ei_add_test(cxx11_tensor_scan)
Use the CMAKE_CXX_STANDARD variable to turn on cxx11
2016-03-05 12:18:01 +08:00
ei_add_test(cxx11_tensor_shuffling)
Drop support for c++03 in Eigen tensor. Get rid of some code used to emulate c++11 functionality with older compilers.
2019-10-19 07:42:00 +08:00
ei_add_test(cxx11_tensor_simple)
Use the CMAKE_CXX_STANDARD variable to turn on cxx11
2016-03-05 12:18:01 +08:00
ei_add_test(cxx11_tensor_striding)
Drop support for c++03 in Eigen tensor. Get rid of some code used to emulate c++11 functionality with older compilers.
2019-10-19 07:42:00 +08:00
ei_add_test(cxx11_tensor_sugar)
ThreadLocal container that does not rely on thread local storage
2019-09-10 06:18:14 +08:00
ei_add_test(cxx11_tensor_thread_local "-pthread" "${CMAKE_THREAD_LIBS_INIT}")
Use full packet size for Dynamic-sized objects (otherwise, the unalignedcount unit test fails with AVX enabled)
2015-09-03 04:51:43 +08:00
ei_add_test(cxx11_tensor_thread_pool "-pthread" "${CMAKE_THREAD_LIBS_INIT}")
Merged in tntnatbry/eigen (pull request PR-319) Tensor Trace op
2017-07-07 12:18:03 +08:00
ei_add_test(cxx11_tensor_trace)
Drop support for c++03 in Eigen tensor. Get rid of some code used to emulate c++11 functionality with older compilers.
2019-10-19 07:42:00 +08:00
ei_add_test(cxx11_tensor_volume_patch)
# ei_add_test(cxx11_tensor_symmetry)
Move cxx11_tensor_uint128 test under an EIGEN_TEST_CXX11 guarded block Builds configured without the -DEIGEN_TEST_CXX11=ON flag would fail right away without this, as this test seems to rely on those language features. The skip under compilation with MSVC was kept.
2018-07-21 07:08:40 +08:00
if("${CMAKE_SIZEOF_VOID_P}" EQUAL "8" AND NOT CMAKE_CXX_COMPILER_ID STREQUAL "MSVC")
# This test requires __uint128_t which is only available on 64bit systems
ei_add_test(cxx11_tensor_uint128)
endif()
Only enable the cxx11_tensor_uint128 test on 64 bit machines since 32 bit systems don't support the __uin128_t type
2016-02-06 10:14:23 +08:00
C++11: add template metaprogramming helpers Create a new directory CXX11 under unsupported/Eigen that contains code that requires C++11. In that directory, add a few generic templates useful for any module relying on C++11. These templates may be included with #include <[unsupported/]Eigen/CXX11/Core>. At the moment, this will only provide templates in the Eigen::internal namespace.
2013-11-15 05:27:06 +08:00
endif()
Made the cuda tests compile using make check
2016-01-28 04:22:17 +08:00
# These tests needs nvcc
Fix unit test filename.
2016-01-28 16:35:37 +08:00
find_package(CUDA 7.0)
Renamed the EIGEN_TEST_NVCC cmake option into EIGEN_TEST_CUDA per the discussion in bug #1173.
2016-04-07 00:35:23 +08:00
if(CUDA_FOUND AND EIGEN_TEST_CUDA)
Filter some compilation flags that nvcc warns about.
2016-03-23 05:26:50 +08:00
# Make sure to compile without the -pedantic, -Wundef, -Wnon-virtual-dtor
# and -fno-check-new flags since they trigger thousands of compilation warnings
# in the CUDA runtime
Fix configuration of CUDA: - preserve user defined CUDA_NVCC_FLAGS - remove the -ansi flag that conflicts with -std=c++11 - do not add -std=c++11 if already there
2016-07-12 00:09:04 +08:00
# Also remove -ansi that is incompatible with std=c++11.
Fixed compilation warnings in the cuda tests
2016-03-18 22:04:54 +08:00
string(REPLACE "-pedantic" "" CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS}")
string(REPLACE "-Wundef" "" CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS}")
Filter some compilation flags that nvcc warns about.
2016-03-23 05:26:50 +08:00
string(REPLACE "-Wnon-virtual-dtor" "" CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS}")
string(REPLACE "-fno-check-new" "" CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS}")
Fix configuration of CUDA: - preserve user defined CUDA_NVCC_FLAGS - remove the -ansi flag that conflicts with -std=c++11 - do not add -std=c++11 if already there
2016-07-12 00:09:04 +08:00
string(REPLACE "-ansi" "" CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS}")
Filter some compilation flags that nvcc warns about.
2016-03-23 05:26:50 +08:00
Fixed compilation warnings in the cuda tests
2016-03-18 22:04:54 +08:00
message(STATUS "Flags used to compile cuda code: " ${CMAKE_CXX_FLAGS})
Made the cuda tests compile using make check
2016-01-28 04:22:17 +08:00
if("${CMAKE_CXX_COMPILER_ID}" STREQUAL "Clang")
Improved support for compiling CUDA code with clang as the host compiler
2016-10-04 08:09:33 +08:00
set(CUDA_NVCC_FLAGS "-ccbin ${CMAKE_C_COMPILER}" CACHE STRING "nvcc flags" FORCE)
Made the cuda tests compile using make check
2016-01-28 04:22:17 +08:00
endif()
Created the new EIGEN_TEST_CUDA_CLANG option to compile the CUDA tests using clang instead of nvcc
2016-04-09 04:16:08 +08:00
if(EIGEN_TEST_CUDA_CLANG)
Minor build improvements * Allow specifying multiple GPU architectures. E.g.: cmake -DEIGEN_CUDA_COMPUTE_ARCH="60;70" * Pass CUDA SDK path to clang. Without it it will default to /usr/local/cuda which may not be the right location, if cmake was invoked with -DCUDA_TOOLKIT_ROOT_DIR=/some/other/CUDA/path
2019-06-01 05:08:34 +08:00
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
string(APPEND CMAKE_CXX_FLAGS " --cuda-path=${CUDA_TOOLKIT_ROOT_DIR}")
foreach(ARCH IN LISTS EIGEN_CUDA_COMPUTE_ARCH)
string(APPEND CMAKE_CXX_FLAGS " --cuda-gpu-arch=sm_${ARCH}")
endforeach()
Created the new EIGEN_TEST_CUDA_CLANG option to compile the CUDA tests using clang instead of nvcc
2016-04-09 04:16:08 +08:00
endif()
Fixed option '--relaxed-constexpr' has been deprecated and replaced by option '--expt-relaxed-constexpr' warning generated by nvcc 7.5
2016-05-28 01:47:53 +08:00
set(EIGEN_CUDA_RELAXED_CONSTEXPR "--expt-relaxed-constexpr")
if (${CUDA_VERSION} STREQUAL "7.0")
set(EIGEN_CUDA_RELAXED_CONSTEXPR "--relaxed-constexpr")
endif()
Minor build improvements * Allow specifying multiple GPU architectures. E.g.: cmake -DEIGEN_CUDA_COMPUTE_ARCH="60;70" * Pass CUDA SDK path to clang. Without it it will default to /usr/local/cuda which may not be the right location, if cmake was invoked with -DCUDA_TOOLKIT_ROOT_DIR=/some/other/CUDA/path
2019-06-01 05:08:34 +08:00
set(NVCC_ARCH_FLAGS)
foreach(ARCH IN LISTS EIGEN_CUDA_COMPUTE_ARCH)
string(APPEND NVCC_ARCH_FLAGS " -gencode arch=compute_${ARCH},code=sm_${ARCH}")
endforeach()
Remove code checking for CMake < 3.5 As the CMake version is at least 3.5 the code checking for earlier versions can be removed.
2020-12-14 17:57:44 +08:00
set(CUDA_NVCC_FLAGS "${EIGEN_CUDA_RELAXED_CONSTEXPR} -Xcudafe \"--display_error_number\" ${NVCC_ARCH_FLAGS} ${CUDA_NVCC_FLAGS}")
Made the cuda tests compile using make check
2016-01-28 04:22:17 +08:00
cuda_include_directories("${CMAKE_CURRENT_BINARY_DIR}" "${CUDA_TOOLKIT_ROOT_DIR}/include")
Fixed the flags passed to nvcc to compile the tensor code.
2016-01-28 06:46:34 +08:00
set(EIGEN_ADD_TEST_FILENAME_EXTENSION "cu")
merging the CUDA and HIP implementation for the Tensor directory and the unit tests
2018-06-21 04:44:58 +08:00
ei_add_test(cxx11_tensor_complex_gpu)
ei_add_test(cxx11_tensor_complex_cwise_ops_gpu)
ei_add_test(cxx11_tensor_reduction_gpu)
ei_add_test(cxx11_tensor_argmax_gpu)
ei_add_test(cxx11_tensor_cast_float16_gpu)
ei_add_test(cxx11_tensor_scan_gpu)
Made the cuda tests compile using make check
2016-01-28 04:22:17 +08:00
Minor build improvements * Allow specifying multiple GPU architectures. E.g.: cmake -DEIGEN_CUDA_COMPUTE_ARCH="60;70" * Pass CUDA SDK path to clang. Without it it will default to /usr/local/cuda which may not be the right location, if cmake was invoked with -DCUDA_TOOLKIT_ROOT_DIR=/some/other/CUDA/path
2019-06-01 05:08:34 +08:00
set(EIGEN_CUDA_OLDEST_COMPUTE_ARCH 9999)
foreach(ARCH IN LISTS EIGEN_CUDA_COMPUTE_ARCH)
if(${ARCH} LESS ${EIGEN_CUDA_OLDEST_COMPUTE_ARCH})
set(EIGEN_CUDA_OLDEST_COMPUTE_ARCH ${ARCH})
endif()
endforeach()
CUDA contractions require arch >= 3.0: don't compile the cuda contraction tests on older architectures.
2016-09-13 09:49:01 +08:00
# Contractions require arch 3.0 or higher
Minor build improvements * Allow specifying multiple GPU architectures. E.g.: cmake -DEIGEN_CUDA_COMPUTE_ARCH="60;70" * Pass CUDA SDK path to clang. Without it it will default to /usr/local/cuda which may not be the right location, if cmake was invoked with -DCUDA_TOOLKIT_ROOT_DIR=/some/other/CUDA/path
2019-06-01 05:08:34 +08:00
if (${EIGEN_CUDA_OLDEST_COMPUTE_ARCH} GREATER 29)
Register the cxx11_tensor_device only for recent cuda architectures (i.e. >= 3.0) since the test instantiate contractions that require a modern gpu.
2016-09-13 10:01:52 +08:00
ei_add_test(cxx11_tensor_device)
merging the CUDA and HIP implementation for the Tensor directory and the unit tests
2018-06-21 04:44:58 +08:00
ei_add_test(cxx11_tensor_gpu)
ei_add_test(cxx11_tensor_contract_gpu)
ei_add_test(cxx11_tensor_of_float16_gpu)
CUDA contractions require arch >= 3.0: don't compile the cuda contraction tests on older architectures.
2016-09-13 09:49:01 +08:00
endif()
Added a test to validate the conversion of half floats into floats on Kepler GPUs. Restricted the testing of the random number generation code to GPU architecture greater than or equal to 3.5.
2016-03-04 02:37:25 +08:00
# The random number generation code requires arch 3.5 or greater.
Minor build improvements * Allow specifying multiple GPU architectures. E.g.: cmake -DEIGEN_CUDA_COMPUTE_ARCH="60;70" * Pass CUDA SDK path to clang. Without it it will default to /usr/local/cuda which may not be the right location, if cmake was invoked with -DCUDA_TOOLKIT_ROOT_DIR=/some/other/CUDA/path
2019-06-01 05:08:34 +08:00
if (${EIGEN_CUDA_OLDEST_COMPUTE_ARCH} GREATER 34)
merging the CUDA and HIP implementation for the Tensor directory and the unit tests
2018-06-21 04:44:58 +08:00
ei_add_test(cxx11_tensor_random_gpu)
Added a test to validate the conversion of half floats into floats on Kepler GPUs. Restricted the testing of the random number generation code to GPU architecture greater than or equal to 3.5.
2016-03-04 02:37:25 +08:00
endif()
Added preliminary support for half floats on CUDA GPU. For now we can simply convert floats into half floats and vice versa
2016-02-19 14:16:07 +08:00
Made the cuda tests compile using make check
2016-01-28 04:22:17 +08:00
unset(EIGEN_ADD_TEST_FILENAME_EXTENSION)
Fix unit test filename.
2016-01-28 16:35:37 +08:00
endif()
Adding support for using Eigen in HIP kernels. This commit enables the use of Eigen on HIP kernels / AMD GPUs. Support has been added along the same lines as what already exists for using Eigen in CUDA kernels / NVidia GPUs. Application code needs to explicitly define EIGEN_USE_HIP when using Eigen in HIP kernels. This is because some of the CUDA headers get picked up by default during Eigen compile (irrespective of whether or not the underlying compiler is CUDACC/NVCC, for e.g. Eigen/src/Core/arch/CUDA/Half.h). In order to maintain this behavior, the EIGEN_USE_HIP macro is used to switch to using the HIP version of those header files (see Eigen/Core and unsupported/Eigen/CXX11/Tensor) Use the "-DEIGEN_TEST_HIP" cmake option to enable the HIP specific unit tests.
2018-06-06 22:12:58 +08:00
Initial support of TensorBlock
2018-07-21 08:37:20 +08:00
# Add HIP specific tests
Adding support for using Eigen in HIP kernels. This commit enables the use of Eigen on HIP kernels / AMD GPUs. Support has been added along the same lines as what already exists for using Eigen in CUDA kernels / NVidia GPUs. Application code needs to explicitly define EIGEN_USE_HIP when using Eigen in HIP kernels. This is because some of the CUDA headers get picked up by default during Eigen compile (irrespective of whether or not the underlying compiler is CUDACC/NVCC, for e.g. Eigen/src/Core/arch/CUDA/Half.h). In order to maintain this behavior, the EIGEN_USE_HIP macro is used to switch to using the HIP version of those header files (see Eigen/Core and unsupported/Eigen/CXX11/Tensor) Use the "-DEIGEN_TEST_HIP" cmake option to enable the HIP specific unit tests.
2018-06-06 22:12:58 +08:00
if (EIGEN_TEST_HIP)
set(HIP_PATH "/opt/rocm/hip" CACHE STRING "Path to the HIP installation.")
if (EXISTS ${HIP_PATH})
Initial support of TensorBlock
2018-07-21 08:37:20 +08:00
list(APPEND CMAKE_MODULE_PATH ${HIP_PATH}/cmake)
Adding support for using Eigen in HIP kernels. This commit enables the use of Eigen on HIP kernels / AMD GPUs. Support has been added along the same lines as what already exists for using Eigen in CUDA kernels / NVidia GPUs. Application code needs to explicitly define EIGEN_USE_HIP when using Eigen in HIP kernels. This is because some of the CUDA headers get picked up by default during Eigen compile (irrespective of whether or not the underlying compiler is CUDACC/NVCC, for e.g. Eigen/src/Core/arch/CUDA/Half.h). In order to maintain this behavior, the EIGEN_USE_HIP macro is used to switch to using the HIP version of those header files (see Eigen/Core and unsupported/Eigen/CXX11/Tensor) Use the "-DEIGEN_TEST_HIP" cmake option to enable the HIP specific unit tests.
2018-06-06 22:12:58 +08:00
find_package(HIP REQUIRED)
if (HIP_FOUND)
execute_process(COMMAND ${HIP_PATH}/bin/hipconfig --platform OUTPUT_VARIABLE HIP_PLATFORM)
CMakefile update for ROCm 4.0 Starting with ROCm 4.0, the `hipconfig --platform` command will return `amd` (prior return value was `hcc`). Updating the CMakeLists.txt files in the test dirs to account for this change.
2020-10-30 01:00:04 +08:00
if ((${HIP_PLATFORM} STREQUAL "hcc") OR (${HIP_PLATFORM} STREQUAL "amd"))
Adding support for using Eigen in HIP kernels. This commit enables the use of Eigen on HIP kernels / AMD GPUs. Support has been added along the same lines as what already exists for using Eigen in CUDA kernels / NVidia GPUs. Application code needs to explicitly define EIGEN_USE_HIP when using Eigen in HIP kernels. This is because some of the CUDA headers get picked up by default during Eigen compile (irrespective of whether or not the underlying compiler is CUDACC/NVCC, for e.g. Eigen/src/Core/arch/CUDA/Half.h). In order to maintain this behavior, the EIGEN_USE_HIP macro is used to switch to using the HIP version of those header files (see Eigen/Core and unsupported/Eigen/CXX11/Tensor) Use the "-DEIGEN_TEST_HIP" cmake option to enable the HIP specific unit tests.
2018-06-06 22:12:58 +08:00
include_directories(${CMAKE_CURRENT_BINARY_DIR})
include_directories(${HIP_PATH}/include)
set(EIGEN_ADD_TEST_FILENAME_EXTENSION "cu")
merging the CUDA and HIP implementation for the Tensor directory and the unit tests
2018-06-21 04:44:58 +08:00
#
# complex datatype is not yet supported by HIP
# so leaving out those tests for now
#
# ei_add_test(cxx11_tensor_complex_gpu)
# ei_add_test(cxx11_tensor_complex_cwise_ops_gpu)
#
ei_add_test(cxx11_tensor_reduction_gpu)
ei_add_test(cxx11_tensor_argmax_gpu)
ei_add_test(cxx11_tensor_cast_float16_gpu)
ei_add_test(cxx11_tensor_scan_gpu)
ei_add_test(cxx11_tensor_device)
ei_add_test(cxx11_tensor_gpu)
ei_add_test(cxx11_tensor_contract_gpu)
ei_add_test(cxx11_tensor_of_float16_gpu)
ei_add_test(cxx11_tensor_random_gpu)
Initial support of TensorBlock
2018-07-21 08:37:20 +08:00
Adding support for using Eigen in HIP kernels. This commit enables the use of Eigen on HIP kernels / AMD GPUs. Support has been added along the same lines as what already exists for using Eigen in CUDA kernels / NVidia GPUs. Application code needs to explicitly define EIGEN_USE_HIP when using Eigen in HIP kernels. This is because some of the CUDA headers get picked up by default during Eigen compile (irrespective of whether or not the underlying compiler is CUDACC/NVCC, for e.g. Eigen/src/Core/arch/CUDA/Half.h). In order to maintain this behavior, the EIGEN_USE_HIP macro is used to switch to using the HIP version of those header files (see Eigen/Core and unsupported/Eigen/CXX11/Tensor) Use the "-DEIGEN_TEST_HIP" cmake option to enable the HIP specific unit tests.
2018-06-06 22:12:58 +08:00
unset(EIGEN_ADD_TEST_FILENAME_EXTENSION)
Initial support of TensorBlock
2018-07-21 08:37:20 +08:00
CMakefile update for ROCm 4.0 Starting with ROCm 4.0, the `hipconfig --platform` command will return `amd` (prior return value was `hcc`). Updating the CMakeLists.txt files in the test dirs to account for this change.
2020-10-30 01:00:04 +08:00
elseif ((${HIP_PLATFORM} STREQUAL "nvcc") OR (${HIP_PLATFORM} STREQUAL "nvidia"))
Adding support for using Eigen in HIP kernels. This commit enables the use of Eigen on HIP kernels / AMD GPUs. Support has been added along the same lines as what already exists for using Eigen in CUDA kernels / NVidia GPUs. Application code needs to explicitly define EIGEN_USE_HIP when using Eigen in HIP kernels. This is because some of the CUDA headers get picked up by default during Eigen compile (irrespective of whether or not the underlying compiler is CUDACC/NVCC, for e.g. Eigen/src/Core/arch/CUDA/Half.h). In order to maintain this behavior, the EIGEN_USE_HIP macro is used to switch to using the HIP version of those header files (see Eigen/Core and unsupported/Eigen/CXX11/Tensor) Use the "-DEIGEN_TEST_HIP" cmake option to enable the HIP specific unit tests.
2018-06-06 22:12:58 +08:00
message(FATAL_ERROR "HIP_PLATFORM = nvcc is not supported within Eigen")
else ()
message(FATAL_ERROR "Unknown HIP_PLATFORM = ${HIP_PLATFORM}")
endif()
Initial support of TensorBlock
2018-07-21 08:37:20 +08:00
STYLE: Remove CMake-language block-end command arguments Ancient versions of CMake required else(), endif(), and similar block termination commands to have arguments matching the command starting the block. This is no longer the preferred style.
2019-11-01 00:36:27 +08:00
endif()
Adding support for using Eigen in HIP kernels. This commit enables the use of Eigen on HIP kernels / AMD GPUs. Support has been added along the same lines as what already exists for using Eigen in CUDA kernels / NVidia GPUs. Application code needs to explicitly define EIGEN_USE_HIP when using Eigen in HIP kernels. This is because some of the CUDA headers get picked up by default during Eigen compile (irrespective of whether or not the underlying compiler is CUDACC/NVCC, for e.g. Eigen/src/Core/arch/CUDA/Half.h). In order to maintain this behavior, the EIGEN_USE_HIP macro is used to switch to using the HIP version of those header files (see Eigen/Core and unsupported/Eigen/CXX11/Tensor) Use the "-DEIGEN_TEST_HIP" cmake option to enable the HIP specific unit tests.
2018-06-06 22:12:58 +08:00
else ()
message(FATAL_ERROR "EIGEN_TEST_HIP is ON, but the specified HIP_PATH (${HIP_PATH}) does not exist")
Initial support of TensorBlock
2018-07-21 08:37:20 +08:00
Adding support for using Eigen in HIP kernels. This commit enables the use of Eigen on HIP kernels / AMD GPUs. Support has been added along the same lines as what already exists for using Eigen in CUDA kernels / NVidia GPUs. Application code needs to explicitly define EIGEN_USE_HIP when using Eigen in HIP kernels. This is because some of the CUDA headers get picked up by default during Eigen compile (irrespective of whether or not the underlying compiler is CUDACC/NVCC, for e.g. Eigen/src/Core/arch/CUDA/Half.h). In order to maintain this behavior, the EIGEN_USE_HIP macro is used to switch to using the HIP version of those header files (see Eigen/Core and unsupported/Eigen/CXX11/Tensor) Use the "-DEIGEN_TEST_HIP" cmake option to enable the HIP specific unit tests.
2018-06-06 22:12:58 +08:00
endif()
Initial support of TensorBlock
2018-07-21 08:37:20 +08:00
STYLE: Remove CMake-language block-end command arguments Ancient versions of CMake required else(), endif(), and similar block termination commands to have arguments matching the command starting the block. This is no longer the preferred style.
2019-11-01 00:36:27 +08:00
endif()
Adding support for using Eigen in HIP kernels. This commit enables the use of Eigen on HIP kernels / AMD GPUs. Support has been added along the same lines as what already exists for using Eigen in CUDA kernels / NVidia GPUs. Application code needs to explicitly define EIGEN_USE_HIP when using Eigen in HIP kernels. This is because some of the CUDA headers get picked up by default during Eigen compile (irrespective of whether or not the underlying compiler is CUDACC/NVCC, for e.g. Eigen/src/Core/arch/CUDA/Half.h). In order to maintain this behavior, the EIGEN_USE_HIP macro is used to switch to using the HIP version of those header files (see Eigen/Core and unsupported/Eigen/CXX11/Tensor) Use the "-DEIGEN_TEST_HIP" cmake option to enable the HIP specific unit tests.
2018-06-06 22:12:58 +08:00
Reference in New Issue Copy Permalink