zhangyiss/spack
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

hydrogen: update to use oneapi packages (#49293)

Browse Source 
Signed-off-by: Massimiliano Culpo <massimiliano.culpo@gmail.com>
This commit is contained in:
Massimiliano Culpo 2025-03-05 09:06:32 +01:00 committed by GitHub
parent aa9e610fa6
commit f015b18230
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 27 additions and 33 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

60
var/spack/repos/builtin/packages/hydrogen/package.py
View File

@ -15,7 +15,10 @@
class Hydrogen(CachedCMakePackage, CudaPackage, ROCmPackage):
"""Hydrogen: Distributed-memory dense and sparse-direct linear algebra
and optimization library. Based on the Elemental library."""
and optimization library.
Based on the Elemental library.
"""
homepage = "https://libelemental.org"
url = "https://github.com/LLNL/Elemental/archive/v1.5.1.tar.gz"
@ -31,8 +34,8 @@ class Hydrogen(CachedCMakePackage, CudaPackage, ROCmPackage):
version("1.5.2", sha256="a902cad3962471216cfa278ba0561c18751d415cd4d6b2417c02a43b0ab2ea33")
version("1.5.1", sha256="447da564278f98366906d561d9c8bc4d31678c56d761679c2ff3e59ee7a2895c")
depends_on("c", type="build") # generated
depends_on("cxx", type="build") # generated
depends_on("c", type="build")
depends_on("cxx", type="build")
# Older versions are no longer supported.
variant("shared", default=True, description="Enables the build of shared libraries.")
@ -71,13 +74,6 @@ class Hydrogen(CachedCMakePackage, CudaPackage, ROCmPackage):
description="Use OpenMP taskloops instead of parallel for loops",
)
# Users should spec this on their own on the command line, no?
# This doesn't affect Hydrogen itself at all. Not one bit.
# variant(
# "openmp_blas",
# default=False,
# description="Use OpenMP for threading in the BLAS library")
variant("test", default=False, description="Builds test suite")
conflicts("+cuda", when="+rocm", msg="CUDA and ROCm support are mutually exclusive")
@ -90,22 +86,25 @@ class Hydrogen(CachedCMakePackage, CudaPackage, ROCmPackage):
depends_on("blas")
depends_on("lapack")
# Note that #1712 forces us to enumerate the different blas variants
# Note that this forces us to use OpenBLAS until #1712 is fixed
depends_on("openblas", when="blas=openblas")
depends_on("openblas +ilp64", when="blas=openblas +int64_blas")
depends_on("openblas@0.3.21:0.3.23", when="blas=openblas arch=ppc64le:")
with when("blas=openblas"):
requires("^[virtuals=blas,lapack] openblas")
requires("^[virtuals=blas,lapack] openblas~ilp64", when="~int64_blas")
requires("^[virtuals=blas,lapack] openblas+ilp64", when="+int64_blas")
requires("^[virtuals=blas,lapack] openblas@0.3.21:0.3.23", when="arch=ppc64le:")
depends_on("intel-mkl", when="blas=mkl")
depends_on("intel-mkl +ilp64", when="blas=mkl +int64_blas")
with when("blas=mkl"):
requires("^[virtuals=blas,lapack] intel-oneapi-mkl")
requires("^[virtuals=blas,lapack] intel-oneapi-mkl ~ilp64", when="~int64_blas")
requires("^[virtuals=blas,lapack] intel-oneapi-mkl +ilp64", when="+int64_blas")
# I don't think this is true...
depends_on("veclibfort", when="blas=accelerate")
depends_on("essl", when="blas=essl")
depends_on("essl +ilp64", when="blas=essl +int64_blas")
depends_on("netlib-lapack +external-blas", when="blas=essl")
with when("blas=essl"):
requires("^[virtuals=blas] essl")
requires("^[virtuals=blas] essl ~ilp64", when="~int64_blas")
requires("^[virtuals=blas] essl +ilp64", when="+int64_blas")
requires("^[virtuals=lapack] netlib-lapack +external-blas")
depends_on("cray-libsci", when="blas=libsci")
@ -117,14 +116,12 @@ class Hydrogen(CachedCMakePackage, CudaPackage, ROCmPackage):
depends_on("aluminum +rocm +ht", when="+al +rocm")
for arch in CudaPackage.cuda_arch_values:
depends_on("aluminum +cuda cuda_arch=%s" % arch, when="+al +cuda cuda_arch=%s" % arch)
depends_on(f"aluminum +cuda cuda_arch={arch}", when=f"+al +cuda cuda_arch={arch}")
# variants +rocm and amdgpu_targets are not automatically passed to
# dependencies, so do it manually.
for val in ROCmPackage.amdgpu_targets:
depends_on(
"aluminum +rocm amdgpu_target=%s" % val, when="+al +rocm amdgpu_target=%s" % val
)
depends_on(f"aluminum +rocm amdgpu_target={val}", when=f"+al +rocm amdgpu_target={val}")
depends_on("cuda@11.0.0:", when="+cuda")
depends_on("hipcub +rocm", when="+rocm +cub")
@ -142,8 +139,7 @@ def libs(self):
return find_libraries("libHydrogen", root=self.prefix, shared=shared, recursive=True)
def cmake_args(self):
args = []
return args
return []
def get_cuda_flags(self):
spec = self.spec
@ -188,9 +184,7 @@ def initconfig_compiler_entries(self):
entries.append(cmake_cache_string("OpenMP_CXX_FLAGS", "-fopenmp=libomp"))
entries.append(cmake_cache_string("OpenMP_CXX_LIB_NAMES", "libomp"))
entries.append(
cmake_cache_string(
"OpenMP_libomp_LIBRARY", "{0}/lib/libomp.dylib".format(clang_root)
)
cmake_cache_string("OpenMP_libomp_LIBRARY", f"{clang_root}/lib/libomp.dylib")
)
return entries
@ -259,9 +253,9 @@ def initconfig_package_entries(self):
)
# CMAKE_PREFIX_PATH should handle this
entries.append(cmake_cache_string("OpenBLAS_DIR", spec["openblas"].prefix))
elif spec.satisfies("blas=mkl") or spec.satisfies("^intel-mkl"):
elif spec.satisfies("blas=mkl"):
entries.append(cmake_cache_option("Hydrogen_USE_MKL", True))
elif spec.satisfies("blas=essl") or spec.satisfies("^essl"):
elif spec.satisfies("blas=essl"):
entries.append(cmake_cache_string("BLA_VENDOR", "IBMESSL"))
# IF IBM ESSL is used it needs help finding the proper LAPACK libraries
entries.append(
@ -280,7 +274,7 @@ def initconfig_package_entries(self):
)
elif spec.satisfies("blas=accelerate"):
entries.append(cmake_cache_option("Hydrogen_USE_ACCELERATE", True))
elif spec.satisfies("^netlib-lapack"):
elif spec.satisfies("^[virtuals=blas,lapack] netlib-lapack"):
entries.append(cmake_cache_string("BLA_VENDOR", "Generic"))
return entries