Charles Schlosser
|
59b3ef5409
|
Partially Vectorize Cast
|
2023-06-09 16:54:31 +00:00 |
|
Rasmus Munk Larsen
|
7d7576f326
|
Avoid underflow in prsqrt.
|
2023-06-06 14:06:19 -07:00 |
|
Charles Schlosser
|
b7151ffaab
|
Fix unary pow error handling and test
|
2023-06-06 18:46:55 +00:00 |
|
Rasmus Munk Larsen
|
7ac8897431
|
Reduce max relative error of prsqrt from 3 to 2 ulps.
|
2023-06-04 22:25:33 +00:00 |
|
Charles Schlosser
|
1d80e23186
|
Optimize scalar_unary_pow_op error handling
|
2023-06-02 18:53:06 +00:00 |
|
Alexander Shaposhnikov
|
316eab8deb
|
Do not set EIGEN_HAS_ARM64_FP16_SCALAR_ARITHMETIC for cuda compilation
|
2023-05-31 15:15:06 +00:00 |
|
Alejandro Acosta
|
07e4604b19
|
Replace usage of CudaStreamDevice with GpuStreamDevice in tensor benchmarks GPU
|
2023-05-30 15:44:07 +00:00 |
|
Rasmus Munk Larsen
|
8c43bf2b5b
|
Clean up Redux.h and fix vectorization_logic test after changes to traversal order in Redux.
|
2023-05-24 20:26:52 +00:00 |
|
Charles Schlosser
|
da6a71faf0
|
Add linear redux evaluators
|
2023-05-24 17:07:25 +00:00 |
|
Charles Schlosser
|
67a1e881d9
|
Sparse matrix column/row removal
|
2023-05-24 17:04:45 +00:00 |
|
Rasmus Munk Larsen
|
de1c884687
|
Add reference to writeup of approach used in canonicalEulerAngles.
|
2023-05-24 15:52:26 +00:00 |
|
Charles Schlosser
|
307a417e1c
|
Fix unrolled assignment evaluator
|
2023-05-22 16:39:24 +00:00 |
|
Juraj Oršulić
|
c18f94e3b0
|
Geometry/EulerAngles: introduce canonicalEulerAngles
|
2023-05-19 15:42:22 +00:00 |
|
Charles Schlosser
|
7d9bb90f15
|
SVD: fix numerous compiler warnings / failures
|
2023-05-15 16:56:47 +00:00 |
|
Rasmus Munk Larsen
|
2709f4c8fb
|
Use relative path to include EmulateArray.h in CXX11Meta.h, and get rid of redundant meta-programming code, which was moved to Core.
|
2023-05-09 23:21:35 +00:00 |
|
Rasmus Munk Larsen
|
9a02c977ec
|
Use relative paths to include Meta.h and MaxSizeVector.h in Tensor
|
2023-05-09 22:07:55 +00:00 |
|
Rasmus Munk Larsen
|
96c42771d6
|
Make it possible to override the synchonization primitives used by the threadpool using macros.
|
2023-05-09 19:36:17 +00:00 |
|
Rasmus Munk Larsen
|
1321821e86
|
Add missing braces in Umeyama.h
|
2023-05-09 19:10:50 +00:00 |
|
Rasmus Munk Larsen
|
524c329ab2
|
Work around compiler bug in Umeyama.h.
|
2023-05-09 18:53:56 +00:00 |
|
Charles Schlosser
|
fbf7189bd5
|
Fix cuda compilation
|
2023-05-08 16:15:47 +00:00 |
|
Mehdi Goli
|
0623791930
|
[SYCL-2020] Enabling USM support for SYCL. SYCL-1.2.1 did not have support for USM.
|
2023-05-05 17:30:36 +00:00 |
|
Andrzej Ciarkowski
|
1698c367a0
|
Use std::shared_ptr for FFTW/IMKL FFT plan implementation; Fixes #2651
|
2023-05-05 16:58:23 +00:00 |
|
Antonio Sánchez
|
1f79a6078f
|
Return NaN in ndtri for values outside valid input range.
|
2023-05-05 16:27:26 +00:00 |
|
Tobias Wood
|
94f57867fe
|
Thread pool
|
2023-05-05 16:23:34 +00:00 |
|
Charles Schlosser
|
9eb8e2afba
|
Change array_cwise test name
|
2023-05-05 03:08:43 +00:00 |
|
Charles Schlosser
|
725c11719b
|
Visitor: fix modulo by zero compiler warning
|
2023-05-04 18:21:09 +00:00 |
|
Chip Kerchner
|
b8208b363c
|
Specialized loadColData correctly - fix previous BF16 GEMV MR
|
2023-05-04 16:38:17 +00:00 |
|
Charles Schlosser
|
2af03fb685
|
clean up array_cwise test
|
2023-05-04 16:02:08 +00:00 |
|
Chip Kerchner
|
fda1373a15
|
Fix ColMajor BF16 GEMV for when vector is RowMajor
|
2023-05-03 20:12:50 +00:00 |
|
Charles Schlosser
|
fdc749de2a
|
JacobiSVD: set m_nonzeroSingularValues to zero if not finite
|
2023-05-02 17:48:21 +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 |
|
Pedro Gonnet
|
874f5947f4
|
Add half-Packet operations to StridedLinearBufferCopy.
|
2023-05-01 16:09:31 +00:00 |
|
Charles Schlosser
|
c9a14f48d9
|
SSE Packet4ui has pcmp, pmin, pmax
|
2023-04-28 20:36:08 +00:00 |
|
Rasmus Munk Larsen
|
0b51f763cb
|
Revert "Geometry/EulerAngles: make sure that returned solution has canonical ranges"
This reverts commit 7f06bcae2c4aae657fded7c7b999d69ee68962d9
|
2023-04-27 00:06:23 +00:00 |
|
Antonio Sánchez
|
2d0c6ad873
|
Revert "Vectorize cast"
This reverts commit eb5ff1861a4783876564a1a79573c3b9ff566863
|
2023-04-26 18:03:36 +00:00 |
|
Charles Schlosser
|
8999525c29
|
AVX2: Packet4ul has pmul, abs2
|
2023-04-26 16:22:16 +00:00 |
|
Charles Schlosser
|
eb5ff1861a
|
Vectorize cast
|
2023-04-26 02:50:13 +00:00 |
|
Antonio Sánchez
|
3918768be1
|
Fix sparse iterator and tests.
|
2023-04-25 19:05:49 +00:00 |
|
Antonio Sanchez
|
70410310a4
|
Fix boolean bitwise and warning.
|
2023-04-25 15:24:49 +00:00 |
|
Charles Schlosser
|
f6cf5dca80
|
Packet4ul does not have Abs2
|
2023-04-21 19:48:01 +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 |
|
Charles Schlosser
|
29c8e3c754
|
fix pow for uint32_t, disable pmul<Packet4ul>
|
2023-04-21 05:47:56 +00:00 |
|
Juraj Oršulić
|
7f06bcae2c
|
Geometry/EulerAngles: make sure that returned solution has canonical ranges
|
2023-04-19 19:12:24 +00:00 |
|
Rasmus Munk Larsen
|
a347dbbab2
|
Delete last few occurences of HasHalfPacket.
|
2023-04-19 10:36:59 -07:00 |
|
Rasmus Munk Larsen
|
b378014fef
|
Make sure we return +/-1 above the clamping point for Erf().
|
2023-04-18 20:53:01 +00:00 |
|
Charles Schlosser
|
e2bbf496f6
|
Use select ternary op in tensor select evaulator
|
2023-04-18 20:52:16 +00:00 |
|
Charles Schlosser
|
2b954be663
|
fix typo in sse packetmath
|
2023-04-18 18:17:41 +00:00 |
|
Rasmus Munk Larsen
|
25685c90ad
|
Fix incorrect packet type for unsigned int version of pfirst() in MSVC workaround in PacketMath.h.
|
2023-04-18 17:46:23 +00:00 |
|
Rasmus Munk Larsen
|
1e223a956c
|
Add missing 'f' in float literal in SpecialFunctionsImpl.h that triggers implicit conversion warning.
|
2023-04-18 17:33:29 +00:00 |
|
Chip Kerchner
|
3f3ce214e6
|
New BF16 pcast functions and move type casting to TypeCasting.h
|
2023-04-18 02:38:38 +00:00 |
|