blob: eccf87258ace3061b9e86437a0dd59894abbc6b0 (
plain) (
blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
|
{ stdenv
, lib
, fetchFromGitHub
, cmake
, mpi
, blas
, gfortran
, llvmPackages
, cudaPackages
, rocmPackages
, config
, gpuBackend ? (
if config.cudaSupport
then "cuda"
else if config.rocmSupport
then "rocm"
else "none"
)
}:
assert builtins.elem gpuBackend [ "none" "cuda" "rocm" ];
stdenv.mkDerivation rec {
pname = "spla";
version = "1.6.1";
src = fetchFromGitHub {
owner = "eth-cscs";
repo = pname;
rev = "v${version}";
hash = "sha256-fNH1IOKV1Re8G7GH9Xfn3itR80eonTbEGKQRRD16/2k=";
};
outputs = [ "out" "dev" ];
postPatch = ''
substituteInPlace src/gpu_util/gpu_blas_api.hpp \
--replace '#include <rocblas.h>' '#include <rocblas/rocblas.h>'
'';
nativeBuildInputs = [
cmake
gfortran
];
buildInputs = [
blas
mpi
]
++ lib.optional (gpuBackend == "cuda") cudaPackages.cudatoolkit
++ lib.optionals (gpuBackend == "rocm") [
rocmPackages.clr
rocmPackages.rocblas
] ++ lib.optional stdenv.isDarwin llvmPackages.openmp
;
cmakeFlags = [
"-DSPLA_OMP=ON"
"-DSPLA_FORTRAN=ON"
"-DSPLA_INSTALL=ON"
# Required due to broken CMake files
"-DCMAKE_INSTALL_LIBDIR=lib"
"-DCMAKE_INSTALL_INCLUDEDIR=include"
]
++ lib.optional (gpuBackend == "cuda") "-DSPLA_GPU_BACKEND=CUDA"
++ lib.optional (gpuBackend == "rocm") [ "-DSPLA_GPU_BACKEND=ROCM" ]
;
preFixup = ''
substituteInPlace $out/lib/cmake/SPLA/SPLASharedTargets-release.cmake \
--replace-fail "\''${_IMPORT_PREFIX}" "$out"
'';
meta = with lib; {
description = "Specialized Parallel Linear Algebra, providing distributed GEMM functionality for specific matrix distributions with optional GPU acceleration";
homepage = "https://github.com/eth-cscs/spla";
license = licenses.bsd3;
maintainers = [ maintainers.sheepforce ];
};
}
|
