Skip to content
Snippets Groups Projects
Commit 595a5d5f authored by Laurent Chardon's avatar Laurent Chardon Committed by Vladimir Druzenko
Browse files

net/openmpi: Fix shortfloat support

- Improve the test for half precision support. The old test gave
incorrect results, in particular for older compilers.
The test is now compiler independent since both gcc and clang
have supported __Float16 for a long while.
- Remove unused code for plist substitution of NO_MCA_PATCHER_OVERWRITE

PR:	280879
parent 10e6520a
No related branches found
No related tags found
No related merge requests found
PORTNAME= openmpi
PORTVERSION= 5.0.5
DISTVERSION= 5.0.5
PORTREVISION= 1
CATEGORIES= net parallel
MASTER_SITES= https://download.open-mpi.org/release/open-mpi/v${PORTVERSION:R}/
......@@ -88,13 +89,9 @@ SLURM_CONFIGURE_WITH= slurm
FCFLAGS+= -fallow-argument-mismatch
.endif
.if ${ARCH} != aarch64 && ${ARCH} != amd64 && ${ARCH} != i386 && !${ARCH:Mpowerpc*}
PLIST_SUB+= NO_MCA_PATCHER_OVERWRITE="@comment "
.else
PLIST_SUB+= NO_MCA_PATCHER_OVERWRITE=""
.endif
.if ${COMPILER_TYPE} == clang && ${COMPILER_VERSION} >= 170 && !${ARCH:Mpowerpc*}
# Both clang and gcc support half precision on these platforms
# Note: RISC-V can support half precision with Zhf extension
.if ${ARCH:Maarch64} || ${ARCH:Mamd64}
PLIST_SUB+= SHORTFLOAT=""
.else
PLIST_SUB+= SHORTFLOAT="@comment "
......
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment