devtools/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2026-04-10 11:34:33 +08:00
Code Issues Packages Projects Releases Wiki Activity
12,261 Commits 15 Branches 93 Tags
4e598ad259bf9561cd5882326e7cafc585d14f47
Commit Graph

243 Commits

Author SHA1 Message Date
Chip Kerchner
4e598ad259 New panel modes for GEMM MMA (real & complex). 2023-09-06 20:03:45 +00:00
Antonio Sánchez
6e4d5d4832 Add IWYU private pragmas to internal headers. 2023-08-21 16:25:22 +00:00
Chip Kerchner
7769eb1b2e Fix problems with recent changes and Tensorflow in Power 2023-07-26 16:24:58 +00:00
Marcus Comstedt
8f927fb52e Altivec: fix compilation with C++20 and higher 2023-07-05 13:14:02 +00:00
Chip Kerchner
3791ac8a1a Fix supportsMMA to obey EIGEN_ALTIVEC_MMA_DYNAMIC_DISPATCH compilation flag and compiler support. 2023-06-28 17:57:21 +00:00
Chip Kerchner
211c5dfc67 Add optional offset parameter to ploadu_partial and pstoreu_partial 2023-06-23 19:53:05 +00:00
Chip Kerchner
b8208b363c Specialized loadColData correctly - fix previous BF16 GEMV MR 2023-05-04 16:38:17 +00:00
Chip Kerchner
fda1373a15 Fix ColMajor BF16 GEMV for when vector is RowMajor 2023-05-03 20:12:50 +00:00
Chip Kerchner
6418ac0285 Unroll F32 to BF16 loop - 1.8X faster conversions for LLVM. Use vector pairs for GCC. 2023-05-01 16:54:16 +00:00
Chip Kerchner
03f646b7e3 New VSX version of BF16 GEMV (Power) - up to 6.7X faster 2023-04-21 17:06:59 +00:00
Chip Kerchner
3f3ce214e6 New BF16 pcast functions and move type casting to TypeCasting.h 2023-04-18 02:38:38 +00:00
Chip Kerchner
1148f0a9ec Add dynamic dispatch to BF16 GEMM (Power) and new VSX version 2023-04-14 22:20:42 +00:00
Rasmus Munk Larsen
df1049ddf4 Small packet math cleanup. 2023-04-04 16:14:32 +00:00
Chip Kerchner
d71ac6a755 Fix recent PowerPC warnings and clang warning 2023-03-15 16:50:46 +00:00
Chip Kerchner
23e1541863 Put deadcode checks back in from previous change. 2023-03-14 00:57:16 +00:00
Chip Kerchner
6c58f0fe1f Revert changes that made BF16 GEMM to cause bad register spillage for LLVM (Power) 2023-03-13 23:36:06 +00:00
Chip Kerchner
9d72412385 Add MMA to BF16 GEMV - 5.0-6.3X faster (for Power) 2023-03-13 19:37:13 +00:00
Rasmus Munk Larsen
ee0ff0ab3a Fix typo in MathFunctions.h 2023-03-13 15:50:40 +00:00
Rasmus Munk Larsen
d6235d76db Clean up generic packetmath specializations for various backends with the help of a macro. 2023-03-10 22:02:23 +00:00
Chip Kerchner
2b513ca2a0 Added partial linear access for LHS & Output - 30% faster for bfloat16 GEMM MMA (Power) 2023-03-02 19:22:43 +00:00
Chip Kerchner
e4598fedbe Fix compiler versions for certain instructions on Power. 2023-02-23 23:24:41 +00:00
Rasmus Munk Larsen
ce62177b5b Vectorize atanh & add a missing definition and unit test for atan. 2023-02-21 03:14:05 +00:00
Chip Kerchner
e797974689 Add and enable Packet int divide for Power10. 2023-02-17 19:04:18 +00:00
Chip Kerchner
4a03409569 Fix problem with array conversions BF16->F32 in Power. 2023-02-13 21:30:45 +00:00
Chip Kerchner
0ecae61568 Disable array BF16 to F32 conversions in Power 2023-02-10 20:06:58 +00:00
Chip Kerchner
fba12e02b3 Fold extra column calculations into an extra MMA accumulator and other bfloat16 MMA GEMM improvements 2023-02-10 17:32:06 +00:00
Chip Kerchner
79cfc74f4d Revert ODR changes and make gemm_extra_cols and gemm_complex_extra_cols EIGEN_ALWAYS_INLINE to avoid external functions. 2023-02-10 17:05:07 +00:00
Alexander Grund
f9659d91f1 Fix ODR violation with gemm_extra_cols on PPC 2023-02-09 22:16:06 +00:00
Chip Kerchner
e71f88abce Change in Power eigen_asserts to eigen_internal_asserts since it is putting unnecessary error checking and assertions without NDEBUG. 2023-02-08 00:57:30 +00:00
Chip Kerchner
4a58f30aa0 Fix pre-POWER8_VECTOR bugs in pcmp_lt and pnegate and reactivate psqrt. 2023-01-31 19:40:24 +00:00
Chip Kerchner
ab8725d947 Turn off vectorize version of rsqrt - doesn't match generic version 2023-01-27 18:28:54 +00:00
Chip Kerchner
6fc9de7d93 Fix slowdown in bfloat16 MMA when rows is not a multiple of 8 or columns is not a multiple of 4. 2023-01-25 18:22:20 +00:00
Sean McBride
d70b4864d9 issue #2581: review and cleanup of compiler version checks 2023-01-17 18:58:34 +00:00
Sergey Fedorov
4d05765345 Altivec fixes for Darwin: do not use unsupported VSX insns 2023-01-12 16:33:33 +00:00
Chip Kerchner
d20fe21ae4 Improve performance for Power10 MMA bfloat16 GEMM 2023-01-06 23:08:37 +00:00
Pedro Caldeira
31ab62d347 Add support for Power10 (AltiVec) MMA instructions for bfloat16. 2022-11-30 23:33:37 +00:00
Chip Kerchner
399ce1ed63 Fix duplicate execution code for Power 8 Altivec in pstore_partial. 2022-11-16 13:41:42 +00:00
Antonio Sanchez
5eacb9e117 Put brackets around unsigned type names. 2022-11-15 09:09:45 -08:00
Antonio Sánchez
37e40dca85 Fix ambiguity in PPC for vec_splats call. 2022-11-14 18:58:16 +00:00
Charles Schlosser
82b152dbe7 Add signbit function 2022-11-04 00:31:20 +00:00
Antonio Sánchez
886aad1361 Disable patan for double on PPC. 2022-10-27 17:56:08 +00:00
Rasmus Munk Larsen
1414a76fa9 Only vectorize atan<double> for Altivec if VSX is available. 2022-10-03 22:06:58 +00:00
Rasmus Munk Larsen
c475228b28 Vectorize atan() for double. 2022-10-01 01:49:30 +00:00
Rasmus Munk Larsen
bd393e15c3 Vectorize acos, asin, and atan for float. 2022-08-29 19:49:33 +00:00
Charles Schlosser
e5af9f87f2 Vectorize pow for integer base / exponent types 2022-08-29 19:23:54 +00:00
Chip Kerchner
9e0afe0f02 Fix non-VSX PowerPC build 2022-08-08 18:18:17 +00:00
Chip Kerchner
84a9d6fac9 Fix use of Packet2d type for non-VSX. 2022-08-03 20:48:13 +00:00
Chip Kerchner
ce60a7be83 Partial Packet support for GEMM real-only (PowerPC). Also fix compilation warnings & errors for some conditions in new API. 2022-08-03 18:15:19 +00:00
Chip Kerchner
84cf3ff18d Add pload_partial, pstore_partial (and unaligned versions), pgather_partial, pscatter_partial, loadPacketPartial and storePacketPartial. 2022-06-27 19:18:00 +00:00
Chip Kerchner
c603275dc9 Better performance for Power10 using more load and store vector pairs for GEMV 2022-06-27 18:11:55 +00:00