Jeffrey Walton
e0a7597199
Update Doxygen comments
2019-11-30 19:16:18 -05:00
Jeffrey Walton
f045993723
Add s390 testing for Travis (PR #912 )
...
Add s390 testing for Travis. Move ppc64 to include jobs. Travis now officially supports both. Also see https://blog.travis-ci.com/2019-11-12-multi-cpu-architecture-ibm-power-ibm-z .
2019-11-16 18:26:04 -05:00
Jeffrey Walton
3de53891f2
Fix Power9 test program for GCC
...
GCC defines _ARCH_PWR9, but cannot consume vec_xl_be. GCC is doing that shit Clang does...
2019-11-16 05:54:46 -05:00
Jeffrey Walton
3427ca90b8
Update documentation
2019-11-16 00:46:36 -05:00
Jeffrey Walton
b50abd7554
Remove unneeded undef's
2019-11-16 00:36:50 -05:00
Jeffrey Walton
043a294dd1
Use const int off=0 in VecLoadBE
2019-11-16 00:34:59 -05:00
Jeffrey Walton
eacd39c4a2
Remove CONST_V32_CAST and NCONST_V32_CAST
...
We need to use byte*, not word32*
2019-11-16 00:28:37 -05:00
Jeffrey Walton
9e824c4369
Use byte pointer for CONST_V32_CAST and NCONST_V32_CAST
...
The underlying byte arrays may not be aligned for a word
2019-11-16 00:08:56 -05:00
Jeffrey Walton
759e586bf0
Sidestep failed self tests on GCC135
2019-11-15 22:55:21 -05:00
Jeffrey Walton
8c9f3821e4
Update documentation
2019-10-29 10:33:39 -04:00
Jeffrey Walton
97cba9c01e
Use simdBlockSize for Altivec. The VSX unit is Power7
2019-10-29 10:18:36 -04:00
Jeffrey Walton
e858473dbb
Update documentation
2019-10-29 10:16:11 -04:00
Jeffrey Walton
74e0c668fc
Update documentation
2019-10-29 09:21:49 -04:00
Jeffrey Walton
90ebe6a362
Add documentation headings
2019-10-29 08:43:24 -04:00
Jeffrey Walton
0968bb53c2
Update documentation
2019-10-28 15:33:36 -04:00
Jeffrey Walton
c831d6ffee
Pre-splat SIMON and SPECK keys when appropriate for Altivec (PR #910 )
...
SIMON and SPECK keys can be pre-splatted in the forward direction when Altivec instructions will be used. Pre-splatting does not work for the reverse transformation. It breaks modes like CBC, so the speed-up is only applied to the forward transformation.
2019-10-28 01:12:58 -04:00
Jeffrey Walton
6da6b7f5ac
Enable Power7 for Simon and Speck (PR #909 )
2019-10-27 14:02:41 -04:00
Jeffrey Walton
e09bdbe5c4
Add C++17 test program
2019-10-27 00:29:18 -04:00
Jeffrey Walton
183fba44bf
Rename VecPolyMultiplyLE to VecIntelMultiply (PR #908 )
...
The LE multiplies are compatible with Intel's _mm_clmulepi64_si128
2019-10-26 22:57:58 -04:00
Jeffrey Walton
fa39314b7a
Add XLC 12 loads and stores for AIX (PR #907 )
...
Add XLC 12 loads and stores for AIX
2019-10-26 22:11:49 -04:00
Jeffrey Walton
1bfb8760bb
Update documentation
2019-10-26 06:26:06 -04:00
Jeffrey Walton
f29bdbb61e
Update documentation
2019-10-25 10:15:16 -04:00
Jeffrey Walton
bff3ed5e27
Improve PPC64 feature tests for XLC on AIX
2019-10-25 09:35:27 -04:00
Jeffrey Walton
1e5beb32d4
Remove unneeded PPC code
...
Update comments
2019-10-25 06:28:47 -04:00
Jeffrey Walton
27f9f948a9
Use builtin datatype pointer
2019-10-24 23:32:06 -04:00
Jeffrey Walton
44cecf22cb
Rework CPU_ProbePower7 (PR #905 )
...
* Rework CPU_ProbePower7
* Remove XLC VSX workaround
* Update documentation
2019-10-24 23:30:05 -04:00
Jeffrey Walton
c363e43968
More Clang and XLC workarounds (PR #904 )
...
I think this is the final set. Clang 3.8 is OK on Ubuntu Xenial. XLC 11.0 is OK on Power 7.
Clang 3.4 crashes on gcc110.fsffrance.org, but it looks like another compiler bug. I'm not spending time trying to work around it. Clang sucks days of your life at times...
2019-10-24 16:48:40 -04:00
Jeffrey Walton
670ceb19a2
Some more workarounds for Clang and XLC on PowerPC (PR #903 )
2019-10-24 15:36:39 -04:00
Jeffrey Walton
a4da9bf6be
Guard PowerPC feature tests
...
Clang is producing spurious noise
2019-10-24 05:33:15 -04:00
Jeffrey Walton
20d3db813b
Update Power7 tests for Clang
...
Clang 3.4 and 3.8 are resiting the builtins
2019-10-24 05:18:12 -04:00
Jeffrey Walton
743cb6a0e6
Update Travis comments
2019-10-24 04:17:08 -04:00
Jeffrey Walton
f8a3b2c6f0
Fix Power7 feature test
2019-10-24 03:30:32 -04:00
Jeffrey Walton
7a7d97a5a5
Fix Power7 feature test
2019-10-24 03:29:46 -04:00
Jeffrey Walton
e7284f8b5d
Fix Power7 feature test
2019-10-24 03:20:12 -04:00
Jeffrey Walton
6b6a84e39a
Guard VecRotateLeft and VecRotateRight with Power8
2019-10-24 02:02:38 -04:00
Jeffrey Walton
4d61cd7310
Try to replace vec_rl for Clang 3.8
...
The PowerPC builds are failing at Travis. Travis provides a Power7 machine with Clang 3.8.
2019-10-23 19:55:13 -04:00
Jeffrey Walton
e59e3c0d2f
Remove unneeded source files from Makefile
2019-10-23 18:36:05 -04:00
Jeffrey Walton
06cee3cdf4
Fix AIX and Linux compiles for PowerPC (PR #902 )
2019-10-23 18:15:47 -04:00
Jeffrey Walton
aeadfcfc5b
Add another workaround for Clang on PowerPC
2019-10-23 13:27:56 -04:00
Jeffrey Walton
c904a7f846
Add another workaround for Clang on PowerPC
2019-10-23 13:10:04 -04:00
Jeffrey Walton
c5be67b64a
Add another workaround for Clang on PowerPC
2019-10-23 12:50:58 -04:00
Jeffrey Walton
1f3e65fc35
More Clang workarounds on PowerPC (PR #901 )
2019-10-23 12:40:45 -04:00
Jeffrey Walton
249eb807bf
Add another Clang workaround on PowerPC
2019-10-23 10:51:39 -04:00
Jeffrey Walton
e912c621e4
Add another Clang workaround on PowerPC
2019-10-23 10:35:55 -04:00
Jeffrey Walton
40f139074b
Port ppc_simd.h to Clang 3.8 (PR #900 )
2019-10-23 08:36:52 -04:00
Jeffrey Walton
8f6e156394
Use Bionic for ARM64 testing
2019-10-22 20:13:47 -04:00
Jeffrey Walton
66321bda10
Cleanup config_asm for PowerPC
2019-10-22 19:52:54 -04:00
Jeffrey Walton
1cb796038d
Add PPC64 builds
2019-10-22 16:59:39 -04:00
Jeffrey Walton
5f57476178
Revert to Commit 0281083 (PR #899 )
...
Travis CI is failing on arm64 testing. Commit 0281083https://travis-ci.community/t/segfaults-in-arm64-environment/5617 .
2019-10-22 09:21:38 -04:00
Jeffrey Walton
3a8d45b5e3
Fix PPC64 compile with Clang 3.8 on Little Endian machines
...
Lets see what this breaks. Clang is a bane in the PowerPC world. Its stupid preprocessor define tricks create more work than it is worth.
2019-10-21 20:09:09 -04:00
