spack/var/spack/repos/builtin/packages/vtk-m/package.py
Cory Bloor 78facf0761
rocmlibs: relax rocm-cmake version requirements (#29857)
* rocmlibs: relax rocm-cmake version requirements

The rocm-cmake modules tend to be backwards-compatible, to the extent
that most ROCm math libraries were built using rocm-cmake@master
for a long while without anybody noticing. (That was fixed in
97f0c3ccd9f0a40896998a7580150a514ec3bc37.)

Some packages, like comgr, barely use rocm-cmake for anything, and
we can easily set a very minimal version requirement. For most
packages, however, it would be a lot of effort to determine the
minimum rocm-cmake version required for each release. For those
packages, I just turned the exact version requirement into a
minimum version requirement.

Since I was looking through the CMakeLists.txt for a large number of
libraries, I also took note of the cmake_minimum_required and adjusted
the cmake minimum requirements to match.

* Add rocblas build dependency to hipblas

The rocblas library is required both for both building and linking
hipblas.

* Remove rocm-cmake from vtk-m dependency list

The rocm-cmake package provides CMake scripts that facilitate common
build configuration tasks in the ROCm libraries. It is never needed at
link-time. Also, there are no calls to find_package(ROCM) or
include(ROCM.*)in vtk-m, so this dependency will never be used.
2022-04-19 07:26:40 -06:00

404 lines
17 KiB
Python

# Copyright 2013-2022 Lawrence Livermore National Security, LLC and other
# Spack Project Developers. See the top-level COPYRIGHT file for details.
#
# SPDX-License-Identifier: (Apache-2.0 OR MIT)
import os
import shutil
import sys
from spack import *
class VtkM(CMakePackage, CudaPackage, ROCmPackage):
"""VTK-m is a toolkit of scientific visualization algorithms for emerging
processor architectures. VTK-m supports the fine-grained concurrency for
data analysis and visualization algorithms required to drive extreme scale
computing by providing abstract models for data and execution that can be
applied to a variety of algorithms across many different processor
architectures."""
homepage = "https://m.vtk.org/"
maintainers = ['robertmaynard', 'kmorel', 'vicentebolea']
url = "https://gitlab.kitware.com/vtk/vtk-m/-/archive/v1.5.1/vtk-m-v1.5.1.tar.gz"
git = "https://gitlab.kitware.com/vtk/vtk-m.git"
tags = ['e4s']
version('master', branch='master')
version('release', branch='release')
version('1.7.1', sha256="7ea3e945110b837a8c2ba49b41e45e1a1d8d0029bb472b291f7674871dbbbb63", preferred=True)
version('1.7.0', sha256="a86667ac22057462fc14495363cfdcc486da125b366cb568ec23c86946439be4")
version('1.6.0', sha256="14e62d306dd33f82eb9ddb1d5cee987b7a0b91bf08a7a02ca3bce3968c95fd76")
version('1.5.5', commit="d2d1c854adc8c0518802f153b48afd17646b6252")
version('1.5.4', commit="bbba2a1967b271cc393abd043716d957bca97972")
version('1.5.3', commit="a3b8525ef97d94996ae843db0dd4f675c38e8b1e")
version('1.5.2', commit="c49390f2537c5ba8cf25bd39aa5c212d6eafcf61")
version('1.5.1', sha256="64c19e66c0d579cfb21bb0df10d649b523b470b0c9a6c2ea5fd979dfeda2c25e")
version('1.5.0', sha256="b1b13715c7fcc8d17f5c7166ff5b3e9025f6865dc33eb9b06a63471c21349aa8")
version('1.4.0', sha256="8d83cca7cd5e204d10da151ce4f1846c1f7414c7c1e579173d15c5ea0631555a")
version('1.3.0', sha256="f88c1b0a1980f695240eeed9bcccfa420cc089e631dc2917c9728a2eb906df2e")
version('1.2.0', sha256="607272992e05f8398d196f0acdcb4af025a4a96cd4f66614c6341f31d4561763")
version('1.1.0', sha256="78618c81ca741b1fbba0853cb5d7af12c51973b514c268fc96dfb36b853cdb18")
# use release, instead of release with debug symbols b/c vtkm libs
# can overwhelm compilers with too many symbols
variant('build_type', default='Release', description='CMake build type',
values=('Debug', 'Release', 'RelWithDebInfo', 'MinSizeRel'))
variant("shared", default=False, description="build shared libs")
variant("doubleprecision", default=True,
description='enable double precision')
variant("logging", default=False, description="build logging support")
variant("ascent_types", default=True, description="build support for ascent types")
variant("virtuals", default=False, description="enable support for deprecated virtual functions")
variant("mpi", default=False, description="build mpi support")
variant("rendering", default=True, description="build rendering support")
variant("64bitids", default=False,
description="enable 64 bits ids")
variant("testlib", default=False, description="build test library")
# Device variants
# CudaPackage provides cuda variant
# ROCmPackage provides rocm variant
variant("kokkos", default=False, when='@1.6:', description="build using Kokkos backend")
variant("cuda_native", default=True, description="build using native cuda backend", when="+cuda")
variant("openmp", default=(sys.platform != 'darwin'), description="build openmp support")
variant("tbb", default=(sys.platform == 'darwin'), description="build TBB support")
depends_on("cmake@3.12:", type="build") # CMake >= 3.12
depends_on("cmake@3.18:", when="+rocm", type="build") # CMake >= 3.18
conflicts('%gcc@:4.10',
msg='vtk-m requires gcc >= 5. Please install a newer version')
depends_on('cuda@10.1.0:', when='+cuda_native')
depends_on("tbb", when="+tbb")
depends_on("mpi", when="+mpi")
# VTK-m uses the default Kokkos backend
depends_on('kokkos', when='+kokkos')
# VTK-m native CUDA and Kokkos CUDA backends are not compatible
depends_on('kokkos ~cuda', when='+kokkos +cuda +cuda_native')
depends_on('kokkos +cuda', when='+kokkos +cuda ~cuda_native')
for cuda_arch in CudaPackage.cuda_arch_values:
depends_on("kokkos cuda_arch=%s" % cuda_arch, when="+kokkos +cuda ~cuda_native cuda_arch=%s" % cuda_arch)
# VTK-m uses the Kokkos HIP backend.
# If Kokkos provides multiple backends, the HIP backend may or
# may not be used for VTK-m depending on the default selected by Kokkos
depends_on('kokkos +rocm', when='+kokkos +rocm')
# Propagate AMD GPU target to kokkos for +rocm
for amdgpu_value in ROCmPackage.amdgpu_targets:
depends_on("kokkos amdgpu_target=%s" % amdgpu_value, when="+kokkos +rocm amdgpu_target=%s" % amdgpu_value)
depends_on("hip@3.7:", when="+rocm")
# The rocm variant is only valid options for >= 1.7. It would be better if
# this could be expressed as a when clause to disable the rocm variant,
# but that is not currently possible since when clauses are stacked,
# not overwritten.
conflicts('+rocm', when='@:1.6')
conflicts("+rocm", when="+cuda")
conflicts("+rocm", when="~kokkos", msg="VTK-m does not support HIP without Kokkos")
# Can build +shared+cuda after @1.7:
conflicts("+shared", when="@:1.6 +cuda_native")
conflicts("+cuda~cuda_native~kokkos", msg="Cannot have +cuda without a cuda device")
conflicts("+cuda", when="cuda_arch=none",
msg="vtk-m +cuda requires that cuda_arch be set")
def cmake_args(self):
spec = self.spec
options = []
gpu_name_table = {'30': 'kepler', '32': 'kepler', '35': 'kepler',
'50': 'maxwell', '52': 'maxwell', '53': 'maxwell',
'60': 'pascal', '61': 'pascal', '62': 'pascal',
'70': 'volta', '72': 'turing', '75': 'turing',
'80': 'ampere', '86': 'ampere'}
with working_dir('spack-build', create=True):
options = ["-DVTKm_ENABLE_TESTING:BOOL=OFF"]
# shared vs static libs logic
# force building statically with cuda
if "+shared" in spec:
options.append('-DBUILD_SHARED_LIBS=ON')
else:
options.append('-DBUILD_SHARED_LIBS=OFF')
# double precision
if "+doubleprecision" in spec:
options.append("-DVTKm_USE_DOUBLE_PRECISION:BOOL=ON")
else:
options.append("-DVTKm_USE_DOUBLE_PRECISION:BOOL=OFF")
# logging support
if "+logging" in spec:
if not spec.satisfies('@1.3.0:'):
raise InstallError('logging is not supported for\
vtkm version lower than 1.3')
options.append("-DVTKm_ENABLE_LOGGING:BOOL=ON")
else:
options.append("-DVTKm_ENABLE_LOGGING:BOOL=OFF")
# mpi support
if "+mpi" in spec:
if not spec.satisfies('@1.3.0:'):
raise InstallError('mpi is not supported for\
vtkm version lower than 1.3')
options.append("-DVTKm_ENABLE_MPI:BOOL=ON")
else:
options.append("-DVTKm_ENABLE_MPI:BOOL=OFF")
# rendering support
if "+rendering" in spec:
options.append("-DVTKm_ENABLE_RENDERING:BOOL=ON")
else:
options.append("-DVTKm_ENABLE_RENDERING:BOOL=OFF")
# Support for ascent types
if "+ascent_types" in spec:
options.append("-DVTKm_USE_DEFAULT_TYPES_FOR_ASCENT:BOOL=ON")
else:
options.append("-DVTKm_USE_DEFAULT_TYPES_FOR_ASCENT:BOOL=OFF")
# Support for deprecated virtual functions
if "+virtuals" in spec:
options.append("-DVTKm_NO_DEPRECATED_VIRTUAL:BOOL=OFF")
else:
options.append("-DVTKm_NO_DEPRECATED_VIRTUAL:BOOL=ON")
# 64 bit ids
if "+64bitids" in spec:
options.append("-DVTKm_USE_64BIT_IDS:BOOL=ON")
print("64 bit ids enabled")
else:
options.append("-DVTKm_USE_64BIT_IDS:BOOL=OFF")
# Support for the testing header files
if "+testlib" in spec and spec.satisfies('@1.7.0:'):
options.append("-DVTKm_ENABLE_TESTING_LIBRARY:BOOL=ON")
if spec.variants["build_type"].value != 'Release':
options.append("-DVTKm_NO_ASSERT:BOOL=ON")
# cuda support
if "+cuda_native" in spec:
options.append("-DVTKm_ENABLE_CUDA:BOOL=ON")
options.append("-DCMAKE_CUDA_HOST_COMPILER={0}".format(
env["SPACK_CXX"]))
if 'cuda_arch' in spec.variants:
cuda_value = spec.variants['cuda_arch'].value
cuda_arch = cuda_value[0]
if cuda_arch in gpu_name_table:
vtkm_cuda_arch = gpu_name_table[cuda_arch]
options.append('-DVTKm_CUDA_Architecture={0}'.format(
vtkm_cuda_arch))
else:
# this fix is necessary if compiling platform has cuda, but
# no devices (this is common for front end nodes on hpc
# clusters). We choose volta as a lowest common denominator
options.append("-DVTKm_CUDA_Architecture=volta")
else:
options.append("-DVTKm_ENABLE_CUDA:BOOL=OFF")
# hip support
if "+rocm" in spec:
options.append("-DVTKm_NO_DEPRECATED_VIRTUAL:BOOL=ON")
archs = ",".join(self.spec.variants['amdgpu_target'].value)
options.append(
"-DCMAKE_HIP_ARCHITECTURES:STRING={0}".format(archs))
# openmp support
if "+openmp" in spec:
# openmp is added since version 1.3.0
if not spec.satisfies('@1.3.0:'):
raise InstallError('OpenMP is not supported for\
vtkm version lower than 1.3')
options.append("-DVTKm_ENABLE_OPENMP:BOOL=ON")
else:
options.append("-DVTKm_ENABLE_OPENMP:BOOL=OFF")
if "+kokkos" in spec:
options.append("-DVTKm_ENABLE_KOKKOS:BOOL=ON")
else:
options.append("-DVTKm_ENABLE_KOKKOS:BOOL=OFF")
# tbb support
if "+tbb" in spec:
# vtk-m detectes tbb via TBB_ROOT env var
os.environ["TBB_ROOT"] = spec["tbb"].prefix
options.append("-DVTKm_ENABLE_TBB:BOOL=ON")
else:
options.append("-DVTKm_ENABLE_TBB:BOOL=OFF")
return options
def smoke_test(self):
print("Checking VTK-m installation...")
spec = self.spec
checkdir = "spack-check"
with working_dir(checkdir, create=True):
source = r"""
#include <vtkm/cont/Algorithm.h>
#include <vtkm/cont/ArrayHandle.h>
#include <vtkm/cont/Initialize.h>
#include <iostream>
#include <vector>
struct NoArgKernel {
VTKM_EXEC void operator()(vtkm::Id) const {}
void SetErrorMessageBuffer(
const vtkm::exec::internal::ErrorMessageBuffer &errorMessage) {
this->ErrorMessage = errorMessage;
}
vtkm::exec::internal::ErrorMessageBuffer ErrorMessage;
};
template <typename PortalType> struct FillArrayKernel {
using ValueType = typename PortalType::ValueType;
FillArrayKernel(const PortalType &array, ValueType fill)
: Array(array), FillValue(fill) {}
VTKM_EXEC void operator()(vtkm::Id index) const {
this->Array.Set(index, this->FillValue);
}
void SetErrorMessageBuffer(
const vtkm::exec::internal::ErrorMessageBuffer &) {
}
PortalType Array;
ValueType FillValue;
};
int main() {
vtkm::cont::Initialize();
constexpr vtkm::Id size = 1000000;
#if defined(VTKM_ENABLE_KOKKOS)
constexpr vtkm::cont::DeviceAdapterTagKokkos desired_device;
#elif defined(VTKM_ENABLE_CUDA)
constexpr vtkm::cont::DeviceAdapterTagCuda desired_device;
#elif defined(VTKM_ENABLE_TBB)
constexpr vtkm::cont::DeviceAdapterTagTBB desired_device;
#elif defined(VTKM_ENABLE_OPENMP)
constexpr vtkm::cont::DeviceAdapterTagOpenMP desired_device;
#else
#error "No VTK-m Device Adapter enabled"
#endif
std::cout << "-------------------------------------------\n";
std::cout << "Testing No Argument Kernel" << std::endl;
vtkm::cont::Algorithm::Schedule(desired_device, NoArgKernel(), size);
vtkm::cont::ArrayHandle<vtkm::Id> handle;
{
std::cout << "-------------------------------------------\n";
std::cout << "Testing Kernel + ArrayHandle PrepareForOutput" << std::endl;
vtkm::cont::Token token;
auto portal = handle.PrepareForOutput(size, desired_device, token);
vtkm::cont::Algorithm::Schedule(desired_device,
FillArrayKernel<decltype(portal)>{portal, 1}, size);
}
{
std::cout << "-------------------------------------------\n";
std::cout << "Testing Kernel + ArrayHandle PrepareForInPlace" << std::endl;
vtkm::cont::Token token;
auto portal = handle.PrepareForInPlace(desired_device, token);
vtkm::cont::Algorithm::Schedule(desired_device,
FillArrayKernel<decltype(portal)>{portal, 2}, size);
}
std::cout << "-------------------------------------------\n";
std::cout << "Ran tests on: " << desired_device.GetName() << std::endl;
return 0;
}
"""
cmakelists = r"""
##============================================================================
## Copyright (c) Kitware, Inc.
## All rights reserved.
## See LICENSE.txt for details.
##
## This software is distributed WITHOUT ANY WARRANTY; without even
## the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
## PURPOSE. See the above copyright notice for more information.
##============================================================================
cmake_minimum_required(VERSION 3.12...3.18 FATAL_ERROR)
project(VTKmSmokeTest CXX)
#Find the VTK-m package
find_package(VTKm REQUIRED QUIET)
add_executable(VTKmSmokeTest main.cxx)
target_link_libraries(VTKmSmokeTest PRIVATE vtkm_cont)
vtkm_add_target_information(VTKmSmokeTest
DROP_UNUSED_SYMBOLS MODIFY_CUDA_FLAGS
DEVICE_SOURCES main.cxx)
"""
with open("main.cxx", 'w') as f:
f.write(source)
with open("CMakeLists.txt", 'w') as f:
f.write(cmakelists)
builddir = "build"
with working_dir(builddir, create=True):
cmake = Executable(spec['cmake'].prefix.bin + "/cmake")
cmakefiledir = spec['vtk-m'].prefix.lib + "/cmake"
cmakefiledir = cmakefiledir + "/" + os.listdir(cmakefiledir)[0]
cmake(*(["..", "-DVTKm_DIR=" + cmakefiledir]))
cmake(*(["--build", "."]))
try:
test = Executable('./VTKmSmokeTest')
output = test(output=str)
except ProcessError:
output = ""
print(output)
if "+cuda_native" in spec:
expected_device = 'Cuda'
elif "+kokkos" in spec:
expected_device = 'Kokkos'
elif "+tbb" in spec:
expected_device = 'TBB'
elif "+openmp" in spec:
expected_device = 'OpenMP'
expected = """\
-------------------------------------------
Testing No Argument Kernel
-------------------------------------------
Testing Kernel + ArrayHandle PrepareForOutput
-------------------------------------------
Testing Kernel + ArrayHandle PrepareForInPlace
-------------------------------------------
Ran tests on: """ + expected_device + "\n"
success = output == expected
if success:
print("Test success")
if not success:
print("Produced output does not match expected output.")
print("Expected output:")
print('-' * 80)
print(expected)
print('-' * 80)
print("Produced output:")
print('-' * 80)
print(output)
print('-' * 80)
raise RuntimeError("VTK-m install check failed")
shutil.rmtree(checkdir)
@run_after('install')
@on_package_attributes(run_tests=True)
def check_install(self):
spec = self.spec
if "@master" in spec:
self.smoke_test()