viskores: new package (#50078)
* viskores: add package * Update var/spack/repos/builtin/packages/viskores/package.py Co-authored-by: Kenneth Moreland <morelandkd@ornl.gov> * Apply suggestions from code review Co-authored-by: Kenneth Moreland <morelandkd@ornl.gov> * Update var/spack/repos/builtin/packages/viskores/package.py --------- Co-authored-by: Kenneth Moreland <morelandkd@ornl.gov>
This commit is contained in:
parent
76168292c3
commit
94719a55b4
215
var/spack/repos/builtin/packages/viskores/package.py
Normal file
215
var/spack/repos/builtin/packages/viskores/package.py
Normal file
@ -0,0 +1,215 @@
|
||||
# Copyright Spack Project Developers. See COPYRIGHT file for details.
|
||||
#
|
||||
# SPDX-License-Identifier: (Apache-2.0 OR MIT)
|
||||
|
||||
|
||||
import os
|
||||
import sys
|
||||
|
||||
from spack.build_systems.cmake import CMakeBuilder
|
||||
from spack.package import *
|
||||
|
||||
|
||||
class Viskores(CMakePackage, CudaPackage, ROCmPackage):
|
||||
"""Viskores is a toolkit of scientific visualization algorithms for emerging
|
||||
processor architectures. Viskores 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://github.com/Viskores/viskores"
|
||||
maintainers("kmorel", "vicentebolea")
|
||||
|
||||
url = "https://github.com/Viskores/Viskores/archive/refs/tags/v1.0.0.tar.gz"
|
||||
git = "https://github.com/Viskores/Viskores.git"
|
||||
tags = ["e4s"]
|
||||
|
||||
test_requires_compiler = True
|
||||
|
||||
version("master", branch="master")
|
||||
version("release", branch="release")
|
||||
version(
|
||||
"1.0.0",
|
||||
sha256="5bff5bbd747b7662bb4630889960371d06fcc5e5a962d974a898d1883f196eba",
|
||||
preferred=True,
|
||||
)
|
||||
|
||||
variant("shared", default=True, description="build shared libs")
|
||||
variant("doubleprecision", default=True, description="enable double precision")
|
||||
variant("logging", default=True, description="build logging support")
|
||||
variant("mpi", default=True, 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")
|
||||
variant("fpic", default=False, description="build fpic support")
|
||||
variant("examples", default=False, description="Install builtin examples")
|
||||
|
||||
# Device variants
|
||||
# CudaPackage provides cuda variant
|
||||
# ROCmPackage provides rocm variant
|
||||
variant("kokkos", default=False, 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")
|
||||
variant("sycl", default=False, description="Build with SYCL backend")
|
||||
|
||||
depends_on("c", type="build")
|
||||
depends_on("cxx", type="build")
|
||||
|
||||
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="viskores 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")
|
||||
depends_on("llvm-openmp", when="+openmp %apple-clang")
|
||||
|
||||
# Viskores uses the default Kokkos backend
|
||||
depends_on("kokkos", when="+kokkos")
|
||||
# Viskores 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,
|
||||
)
|
||||
# Viskores uses the Kokkos HIP backend.
|
||||
# If Kokkos provides multiple backends, the HIP backend may or
|
||||
# may not be used for Viskores 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")
|
||||
# CUDA thrust is already include in the CUDA pkg
|
||||
depends_on("rocthrust", when="+kokkos+rocm ^cmake@3.24:")
|
||||
|
||||
# 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="+cuda")
|
||||
conflicts("+rocm", when="~kokkos", msg="Viskores does not support HIP without Kokkos")
|
||||
|
||||
# Viskores uses the Kokkos SYCL backend.
|
||||
# If Kokkos provides multiple backends, the SYCL backend may or
|
||||
# may not be used for Viskores depending on the default selected by Kokkos
|
||||
depends_on("kokkos +sycl", when="+kokkos +sycl")
|
||||
conflicts("+sycl", when="~kokkos", msg="Viskores does not support SYCL without Kokkos")
|
||||
conflicts("+cuda~cuda_native~kokkos", msg="Cannot have +cuda without a cuda device")
|
||||
conflicts("+cuda", when="cuda_arch=none", msg="viskores +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):
|
||||
is_release = spec.variants["build_type"].value == "Release"
|
||||
options = [
|
||||
self.define("Viskores_ENABLE_TESTING", False),
|
||||
self.define("Viskores_NO_ASSERT", is_release),
|
||||
self.define_from_variant("BUILD_SHARED_LIBS", "shared"),
|
||||
self.define_from_variant("Viskores_ENABLE_KOKKOS", "kokkos"),
|
||||
self.define_from_variant("Viskores_ENABLE_LOGGING", "logging"),
|
||||
self.define_from_variant("Viskores_ENABLE_MPI", "mpi"),
|
||||
self.define_from_variant("Viskores_ENABLE_OPENMP", "openmp"),
|
||||
self.define_from_variant("Viskores_ENABLE_RENDERING", "rendering"),
|
||||
self.define_from_variant("Viskores_ENABLE_TBB", "tbb"),
|
||||
self.define_from_variant("Viskores_ENABLE_TESTING_LIBRARY", "testlib"),
|
||||
self.define_from_variant("Viskores_INSTALL_EXAMPLES", "examples"),
|
||||
self.define_from_variant("Viskores_USE_64BIT_IDS", "64bitids"),
|
||||
self.define_from_variant("Viskores_USE_DOUBLE_PRECISION", "doubleprecision"),
|
||||
self.define(
|
||||
"Viskores_USE_DEFAULT_TYPES_FOR_ASCENT", "~64bitids +doubleprecision" in spec
|
||||
),
|
||||
]
|
||||
|
||||
if "+tbb" in spec:
|
||||
# viskores detectes tbb via TBB_ROOT env var
|
||||
os.environ["TBB_ROOT"] = spec["tbb"].prefix
|
||||
|
||||
if "+kokkos" in spec and "+rocm" in spec and spec.satisfies("^kokkos@4:"):
|
||||
options.append(f"-DCMAKE_CXX_COMPILER:FILEPATH={spec['hip'].prefix.bin.hipcc}")
|
||||
|
||||
# Support for relocatable code
|
||||
if "~shared" in spec and "+fpic" in spec:
|
||||
options.append("-DCMAKE_POSITION_INDEPENDENT_CODE:BOOL=ON")
|
||||
|
||||
# cuda support
|
||||
if "+cuda_native" in spec:
|
||||
options.append("-DViskores_ENABLE_CUDA:BOOL=ON")
|
||||
options.append("-DCMAKE_CUDA_HOST_COMPILER={0}".format(env["SPACK_CXX"]))
|
||||
|
||||
if spec.satisfies("^cmake@3.18:"):
|
||||
options.append(CMakeBuilder.define_cuda_architectures(self))
|
||||
|
||||
else:
|
||||
# Viskores_CUDA_Architecture only accepts a single CUDA arch
|
||||
num_cuda_arch = spec.variants["cuda_arch"].value[0]
|
||||
str_cuda_arch = str()
|
||||
|
||||
try:
|
||||
str_cuda_arch = gpu_name_table[num_cuda_arch]
|
||||
except KeyError:
|
||||
raise InstallError(f"cuda_arch={num_cuda_arch} needs cmake>=3.18")
|
||||
options.append(f"-DViskores_CUDA_Architecture={str_cuda_arch}")
|
||||
|
||||
else:
|
||||
options.append("-DViskores_ENABLE_CUDA:BOOL=OFF")
|
||||
|
||||
# hip support
|
||||
if "+rocm" in spec:
|
||||
options.append(CMakeBuilder.define_hip_architectures(self))
|
||||
|
||||
return options
|
||||
|
||||
def test_smoke_test(self):
|
||||
"""Build and run ctests"""
|
||||
if "+examples" not in self.spec:
|
||||
raise SkipTest("Package must be installed with +examples")
|
||||
|
||||
testdir = "smoke_test_build"
|
||||
with working_dir(testdir, create=True):
|
||||
cmake = Executable(self.spec["cmake"].prefix.bin.cmake)
|
||||
ctest = Executable(self.spec["cmake"].prefix.bin.ctest)
|
||||
|
||||
mpi_home = str()
|
||||
if "+mpi" in self.spec:
|
||||
mpi_home = self.spec["mpi"].prefix
|
||||
cmake(
|
||||
self.prefix.share.doc.Viskores.examples.smoke_test,
|
||||
f"-DCMAKE_C_COMPILER={self.compiler.cc}",
|
||||
f"-DCMAKE_CXX_COMPILER={self.compiler.cxx}",
|
||||
f"-DMPI_HOME={mpi_home}",
|
||||
f"-DViskores_ROOT={self.prefix}",
|
||||
)
|
||||
cmake("--build", ".")
|
||||
ctest("--verbose")
|
||||
|
||||
@run_after("install")
|
||||
@on_package_attributes(run_tests=True)
|
||||
def build_test(self):
|
||||
self.test_smoke_test()
|
Loading…
Reference in New Issue
Block a user