Commit Graph

263 Commits

Author SHA1 Message Date
Martin Kroeker
2e99e2699b Add workaround for gcc 4.6 miscompiling assembly kernels with -mavx 2020-11-29 15:32:17 +01:00
Martin Kroeker
437702e0e1 Merge pull request #2965 from epsilon-0/develop
allow setting soname without suffix or prefix
2020-11-22 12:25:33 +01:00
Xianyi Zhang
fc35b72ae1 Refs #2899
Merge branch 'openblas-open-910' of git://github.com/damonyu1989/OpenBLAS into damonyu1989-openblas-open-910
2020-11-10 09:38:04 +08:00
Xianyi Zhang
913cc9a4ca Merge branch 'develop' into risc-v 2020-11-10 09:18:25 +08:00
Martin Kroeker
1c4cfdc139 Stay compatible with old gmake that did not support undefine 2020-11-08 00:12:55 +01:00
Martin Kroeker
f6a57d8f63 Update Makefile.system 2020-11-08 00:01:36 +01:00
Martin Kroeker
f4b7ba12b7 Update Makefile.system 2020-11-07 23:37:21 +01:00
Martin Kroeker
a04f532edf Reset cpu property flags between build cycles in DYNAMIC_ARCH mode 2020-11-07 20:37:03 +01:00
Martin Kroeker
8cc73fee98 Export NO_EXPRECISION after overriding for DYNAMIC_ARCH with GENERIC target 2020-11-03 23:47:04 +01:00
Aisha Tammy
60997ddd73 allow setting soname without suffix or prefix
Allows to create a library with a different
SONAME without the need to add suffixes to symbols
Backwards compatible and should have no effect
on the workflow and previous users.
Useful for allowing INTERFACE64 library alongside
the standard library without file conflicts
2020-11-02 13:04:53 +00:00
Martin Kroeker
40a93c232b Disable EXPRECISION for DYNAMIC_ARCH in combination with TARGET=GENERIC
NO_EXPRECISION is disabled for the GENERIC_TARGET already, so prevent mixing with code parts that use a different float size by default
2020-11-01 21:58:26 +01:00
Chen, Guobing
c5e62dad69 Fix cooperlake compile issue
Add a missing macro which is required in Makefile.x86_64 due to recent
clearnup, which causes cooperlake platform build failure.
2020-10-29 03:37:59 +08:00
Martin Kroeker
878b6d1f41 Remove spurious expr in flang version check 2020-10-26 21:35:40 +01:00
Martin Kroeker
1a0f57c8f0 Fix missing backquotes 2020-10-20 08:37:53 +02:00
Martin Kroeker
bb8c3f6861 Add ld/binutils version check for POWER10 support 2020-10-20 01:04:20 +02:00
Zhang Xianyi
d7ba7679b6 Merge branch 'develop' into risc-v 2020-10-16 23:27:38 +08:00
damonyu
ef8e7d0279 Add the support for RISC-V Vector.
Change-Id: Iae7800a32f5af3903c330882cdf6f292d885f266
2020-10-15 16:09:02 +08:00
Martin Kroeker
2c552f1074 Change "HALF" and "sh" to "BFLOAT16" and "sb" 2020-10-12 00:11:31 +02:00
Martin Kroeker
ae8b0d257a Set BUILD_ options to 1 instead of just defining them 2020-10-11 18:08:21 +02:00
Martin Kroeker
8c5e08076e If none of the BUILD_ options is set, enable them all 2020-10-11 17:33:51 +02:00
Marius Hillenbrand
75d440caa0 s390x/DYNAMIC_ARCH: fixup broken merge and reapply simplification
An unrelated commit and merge inadvertently reverted our recent two
changes for simplifying DYNAMIC_ARCH on s390x. Simply reapply the
changes.

Simplify detection of which kernels we can compile on s390x. Instead of
decoding the gcc version in a complicated manner, just check if CC
supports a given -march=archXY flag. Together with the next patch, we
thereby gain support for builds with LLVM/clang with DYNAMIC_ARCH=1.

To enable builds with DYNAMIC_ARCH with older compiler releases, the
Makefile and drivers/other/dynamic_arch.c need a common view of the
architecture support built into the library.

We follow the notation from x86 when used with DYNAMIC_LIST, where
defines DYN_<ARCH NAME> denote support for a given generation to be
built in. Since there are far fewer architecture generations in OpenBLAS
for s390x, that does not bloat command lines too much.

Closes: #2842
Fixes: ba644378dc ("Copy BUILD_ options available to the compiler flags"

Signed-off-by: Marius Hillenbrand <mhillen@linux.ibm.com>
2020-09-17 17:09:03 +02:00
Martin Kroeker
ba644378dc Copy BUILD_ options available to the compiler flags 2020-09-14 00:03:33 +02:00
Marius Hillenbrand
4f34bcfb5e s390x/DYNAMIC_ARCH: pass supported arch levels from Makefile to run-time code
... instead of duplicating the (old) mechanism from the Makefile that
aimed to derive supported architecture generations from the gcc
version.

To enable builds with DYNAMIC_ARCH with older compiler releases, the
Makefile and drivers/other/dynamic_arch.c need a common view of the
architecture support built into the library.

We follow the notation from x86 when used with DYNAMIC_LIST, where
defines DYN_<ARCH NAME> denote support for a given generation to be
built in. Since there are far fewer architecture generations in OpenBLAS
for s390x, that does not bloat command lines too much.

Signed-off-by: Marius Hillenbrand <mhillen@linux.ibm.com>
2020-09-08 19:34:18 +02:00
Marius Hillenbrand
0629d8ebdb s390x/DYNAMIC_ARCH: generalize detecting supported archs for clang
Simplify detection of which kernels we can compile on s390x. Instead of
decoding the gcc version in a complicated manner, just check if CC
supports a given -march=archXY flag. Together with the next patch, we
thereby gain support for builds with LLVM/clang with DYNAMIC_ARCH=1.

Signed-off-by: Marius Hillenbrand <mhillen@linux.ibm.com>
2020-09-08 19:34:18 +02:00
pkubaj
48a1364e10 Add aliases for armv6, armv7
FreeBSD uses those names for 32-bit ARM variants.
2020-08-23 18:50:19 +00:00
Chen, Guobing
e740c4873d Enable COOPERLAKE build target
Enable new build target platform -- COOPERLAKE. This target platform
supports all the SKYLAKEX supported ISAs + avx512bf16. So all the
SKYLAKEX specific kernels/drivers and related code are now extended
to be also active on COOPERLAKE. Besides, new BF16 related kernels
are active under this target.
2020-08-13 06:18:00 +08:00
Ashwin Sekhar T K
4e1be0e481 ARM64: Add THUNDERX3T110 Target 2020-07-26 23:32:24 -07:00
Rajalakshmi Srinivasaraghavan
9be2688c78 Fix to store results in correct order for POWER10 GEMM kernels
There is a recent compiler change in __builtin_mma_disassemble_acc() which
affects the order of storing result in POWER10. Also removing new LDFLAG
-mno-power10-stub as it is handled by linker automatically.
2020-07-24 23:08:11 -05:00
Martin Kroeker
9796e552ea Avoid undefining NAME,CNAME etc for pgcc as it makes it ignore the new defininitions 2020-07-23 17:03:28 +02:00
Wileam Phan
9ae154ba89 Patch for building on Summit 2020-07-20 23:30:28 -04:00
Rajalakshmi Srinivasaraghavan
417c4e8af8 Add new linker option for POWER10
While building with DYNAMIC_ARCH on POWER9 with POWER10
aware toolchain, new LDFLAG is needed to avoid POWER10
instructions on PLT calls .
2020-07-14 11:54:04 -05:00
Martin Kroeker
419b8686d1 Merge pull request #2682 from martin-frbg/aix
[WIP] fix compilation on AIX
2020-07-13 14:43:24 +02:00
Martin Kroeker
5865c7d4d6 Make 32bit POWER8 use POWER6 kernels for now 2020-07-12 18:59:01 +02:00
Rajalakshmi Srinivasaraghavan
af1e140e35 Change minimum gcc version for POWER10
As the MMA patches for POWER10 are backported to gcc10.2, changing
the minimum gcc version needed to build OpenBLAS for POWER10.
2020-07-09 21:46:06 -05:00
Martin Kroeker
5ff83a4261 Merge pull request #2670 from mhillenibm/dumpfullversion_on_gcc7
RFC: Use -dumpfullversion to get minor version on gcc-7 and newer
2020-07-07 00:12:28 +02:00
Martin Kroeker
6eaeb01263 Merge pull request #2658 from RajalakshmiSR/p10
powerpc: Add support for future processor
2020-06-23 00:02:37 +02:00
Marius Hillenbrand
cde4690721 RFC: Use gcc -dumpfullversion to get minor version with gcc-7.x
In gcc-7.1, the behavior of -dumpversion changed to be configured
at compile-time. On some distributions it only dumps the major version
(e.g., Ubuntu), so the current checks for the gcc minor version report
false negatives. As a replacement, gcc-7.1 introduced -dumpfullversion
which always prints the full version.

Update the gcc version detection in Makefile.system to employ
-dumpfullversion with gcc-7 and newer.

Posting this patch for discussion, since it emerged from discussions
around issue #2668 and PR #2669. It is not solving a problem right now,
but may be useful in the future.

Signed-off-by: Marius Hillenbrand <mhillen@linux.ibm.com>
2020-06-16 15:45:59 +02:00
Marius Hillenbrand
2389291766 Makefile.system: remove duplicate variable GCCVERSIONGT5
... to bring unified gcc version detection with common variables to the
one remaining spot in Makefile.system.

Signed-off-by: Marius Hillenbrand <mhillen@linux.ibm.com>
2020-06-16 15:06:03 +02:00
Marius Hillenbrand
a2d13ea611 Fix gcc version detection for zarch
Employ common variables for gcc version detection and fix the broken
check for gcc >= 5.2.
Fixes #2668

Signed-off-by: Marius Hillenbrand <mhillen@linux.ibm.com>
2020-06-16 15:06:03 +02:00
Martin Kroeker
efc53b6e7e Merge pull request #2665 from martin-frbg/flang-fixes-2a
Fix spelling of flang option -Mrecursive, add -Kieee and workaround for AOCC optimizer bug
2020-06-14 21:56:08 +02:00
Martin Kroeker
79cdcde717 Re-enable higher optimization levels for flang while disabling loop unrolling for AOCC flang 2020-06-14 17:18:16 +02:00
Martin Kroeker
0ed2adf0b2 Fix spelling of flang option -Mrecursive and add -Kieee 2020-06-14 00:01:20 +02:00
Martin Kroeker
abf670757b Respect predefined defaults for AR, AS, LD and RANLIB 2020-06-13 23:21:13 +02:00
Rajalakshmi Srinivasaraghavan
9fe930f205 powerpc: Add support for future processor
This is the initial patch to support build infrastructure
for POWER10 architecture.
2020-06-11 15:47:20 -05:00
Martin Kroeker
f07a80354b Apply previously AOCC-specific workaround to all versions of flang 2020-06-09 16:07:03 +02:00
Martin Kroeker
4e28dc6353 Use only -O1 with AMD AOCC version of flang
to prevent miscompilation of LAPACK codes and tests on Ryzen
2020-06-07 00:05:02 +02:00
Martin Kroeker
4a4c50a7ce Merge pull request #2627 from pkubaj/patch-1
Add powerpc (32-bit)
2020-05-26 08:36:24 +02:00
pkubaj
33c8790603 Add powerpc (32-bit)
Only powerpc64 is present.
2020-05-25 13:14:09 +02:00
Guodong Xu
06387ac0e6 make GCC version detection OS-independent
Previous design put GCC version detection inside of OSNAME 'WINNT'.
However, such detections are required for 'Linux' and possibly other
OS'es as well. For example, there is usage of the GCC versions
in Makefile.arm64. When compiling on Linux machine, in the previous
design, Markfile.arm64 will not know the correct GCC version.

The fix is to move GCC version detection into common part, not
wrapped by anything.

Signed-off-by: Guodong Xu <guodong.xu@linaro.com>
2020-05-25 10:57:03 +00:00
Marius Hillenbrand
62cf391cbb s390x: only build kernels supported by gcc with dynamic arch support
When building with dynamic arch support, only build kernels for
architectures that are supported by the gcc we are building with.

Signed-off-by: Marius Hillenbrand <mhillen@linux.ibm.com>
2020-05-12 11:01:16 +02:00