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
12,581 Commits 15 Branches 93 Tags
8adf43640e6957d939bd76eaf9d2420c33083886
Commit Graph

55 Commits

Author SHA1 Message Date
Rasmus Munk Larsen
3f067c4850 Add exp2() as a packet op and array method. 2024-10-22 22:09:34 +00:00
Rasmus Munk Larsen
bbdabebf44 Vectorize atanh<double>. Make atanh(x) standard compliant for |x| >= 1. 2024-08-30 17:27:55 +00:00
Rasmus Munk Larsen
f91f8e9ab9 Consolidate float and double implementations of patan(). 2024-08-21 20:44:18 +00:00
Rasmus Munk Larsen
32d95bb097 Add vectorized implementation of tanh<double> 2024-08-21 02:29:45 +00:00
Charles Schlosser
fcaf03ef7c fix pendantic compiler warnings 2024-04-17 16:55:45 +00:00
Rasmus Munk Larsen
b5feca5d03 Fix build for pblend and psin_double, pcos_double when AVX but not AVX2 is supported. 2024-04-16 16:12:41 +00:00
Tobias Wood
f38e16c193 Apply clang-format 2023-11-29 11:12:48 +00:00
Antonio Sánchez
6e4d5d4832 Add IWYU private pragmas to internal headers. 2023-08-21 16:25:22 +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
Rasmus Munk Larsen
ce62177b5b Vectorize atanh & add a missing definition and unit test for atan. 2023-02-21 03:14:05 +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
Sean McBride
f1b9692d63 Removed EIGEN_UNUSED decorations from many functions that are in fact used 2022-03-03 20:19:33 +00:00
Antonio Sánchez
9c07e201ff Modified sqrt/rsqrt for denormal handling. 2022-03-02 17:20:47 +00:00
Rasmus Munk Larsen
8b875dbef1 Changes to fast SQRT/RSQRT 2022-02-23 17:32:21 +00:00
Rasmus Munk Larsen
979fdd58a4 Add generic fast psqrt and prsqrt impls and make them correct for 0, +Inf, NaN, and negative arguments. 2022-02-05 00:20:13 +00:00
Rasmus Munk Larsen
ea2c02060c Add reciprocal packet op and fast specializations for float with SSE, AVX, and AVX512. 2022-01-21 23:49:18 +00:00
Kolja Brix
8d81a2339c Reduce usage of reserved names 2022-01-10 20:53:29 +00:00
Rasmus Munk Larsen
96dc37a03b Some fixes/cleanups for numeric_limits & fix for related bug in psqrt 2022-01-07 01:10:17 +00:00
Rasmus Munk Larsen
d7d0bf832d Issue an error in case of direct inclusion of internal headers. 2021-09-10 19:12:26 +00:00
Antonio Sanchez
b2126fd6b5 Fix pfrexp/pldexp for half. 
The recent addition of vectorized pow (!330) relies on `pfrexp` and
`pldexp`.  This was missing for `Eigen::half` and `Eigen::bfloat16`.
Adding tests for these packet ops also exposed an issue with handling
negative values in `pfrexp`, returning an incorrect exponent.

Added the missing implementations, corrected the exponent in `pfrexp1`,
and added `packetmath` tests.
 2021-01-21 19:32:28 +00:00
Rasmus Munk Larsen
05754100fe * Add iterative psqrt<double> for AVX and SSE when FMA is available. This provides a ~10% speedup. 
* Write iterative sqrt explicitly in terms of pmadd. This gives up to 7% speedup for psqrt<float> with AVX & SSE with FMA.
* Remove iterative psqrt<double> for NEON, because the initial rsqrt apprimation is not accurate enough for convergence in 2 Newton-Raphson steps and with 3 steps, just calling the builtin sqrt insn is faster.

The following benchmarks were compiled with clang "-O2 -fast-math -mfma" and with and without -mavx.

AVX+FMA (float)

name                      old cpu/op  new cpu/op  delta
BM_eigen_sqrt_float/1     1.08ns ± 0%  1.09ns ± 1%    ~
BM_eigen_sqrt_float/8     2.07ns ± 0%  2.08ns ± 1%    ~
BM_eigen_sqrt_float/64    12.4ns ± 0%  12.4ns ± 1%    ~
BM_eigen_sqrt_float/512   95.7ns ± 0%  95.5ns ± 0%    ~
BM_eigen_sqrt_float/4k     776ns ± 0%   763ns ± 0%  -1.67%
BM_eigen_sqrt_float/32k   6.57µs ± 1%  6.13µs ± 0%  -6.69%
BM_eigen_sqrt_float/256k  83.7µs ± 3%  83.3µs ± 2%    ~
BM_eigen_sqrt_float/1M     335µs ± 2%   332µs ± 2%    ~

SSE+FMA (float)
name                      old cpu/op  new cpu/op  delta
BM_eigen_sqrt_float/1     1.08ns ± 0%  1.09ns ± 0%    ~
BM_eigen_sqrt_float/8     2.07ns ± 0%  2.06ns ± 0%    ~
BM_eigen_sqrt_float/64    12.4ns ± 0%  12.4ns ± 1%    ~
BM_eigen_sqrt_float/512   95.7ns ± 0%  96.3ns ± 4%    ~
BM_eigen_sqrt_float/4k     774ns ± 0%   763ns ± 0%  -1.50%
BM_eigen_sqrt_float/32k   6.58µs ± 2%  6.11µs ± 0%  -7.06%
BM_eigen_sqrt_float/256k  82.7µs ± 1%  82.6µs ± 1%    ~
BM_eigen_sqrt_float/1M     330µs ± 1%   329µs ± 2%    ~

SSE+FMA (double)
BM_eigen_sqrt_double/1      1.63ns ± 0%  1.63ns ± 0%     ~
BM_eigen_sqrt_double/8      6.51ns ± 0%  6.08ns ± 0%   -6.68%
BM_eigen_sqrt_double/64     52.1ns ± 0%  46.5ns ± 1%  -10.65%
BM_eigen_sqrt_double/512     417ns ± 0%   374ns ± 1%  -10.29%
BM_eigen_sqrt_double/4k     3.33µs ± 0%  2.97µs ± 1%  -11.00%
BM_eigen_sqrt_double/32k    26.7µs ± 0%  23.7µs ± 0%  -11.07%
BM_eigen_sqrt_double/256k    213µs ± 0%   206µs ± 1%   -3.31%
BM_eigen_sqrt_double/1M      862µs ± 0%   870µs ± 2%   +0.96%

AVX+FMA (double)
name                        old cpu/op  new cpu/op  delta
BM_eigen_sqrt_double/1      1.63ns ± 0%  1.63ns ± 0%     ~
BM_eigen_sqrt_double/8      6.51ns ± 0%  6.06ns ± 0%   -6.95%
BM_eigen_sqrt_double/64     52.1ns ± 0%  46.5ns ± 1%  -10.80%
BM_eigen_sqrt_double/512     417ns ± 0%   373ns ± 1%  -10.59%
BM_eigen_sqrt_double/4k     3.33µs ± 0%  2.97µs ± 1%  -10.79%
BM_eigen_sqrt_double/32k    26.7µs ± 0%  23.8µs ± 0%  -10.94%
BM_eigen_sqrt_double/256k    214µs ± 0%   208µs ± 2%   -2.76%
BM_eigen_sqrt_double/1M      866µs ± 3%   923µs ± 7%     ~
 2020-12-16 18:16:11 +00:00
Rasmus Munk Larsen
f9fac1d5b0 Add log2() to Eigen. 2020-12-04 21:45:09 +00:00
Rasmus Munk Larsen
f23dc5b971 Revert "Add log2() operator to Eigen" 
This reverts commit 4d91519a9b.
 2020-12-03 14:32:45 -08:00
Rasmus Munk Larsen
4d91519a9b Add log2() operator to Eigen 2020-12-03 22:31:44 +00:00
Antonio Sanchez
a3b300f1af Implement missing AVX half ops. 
Minimal implementation of AVX `Eigen::half` ops to bring in line
with `bfloat16`.  Allows `packetmath_13` to pass.

Also adjusted `bfloat16` packet traits to match the supported set
of ops (e.g. Bessel is not actually implemented).
 2020-11-24 16:46:41 +00:00
Guoqiang QI
4700713faf Add AVX plog<Packet4d> and AVX512 plog<Packet8d> ops,also unified AVX512 plog<Packet16f> op with generic api 2020-10-15 00:54:45 +00:00
Sheng Yang
56b3e3f3f8 AVX path for BF16 2020-07-14 01:34:03 +00:00
Rasmus Munk Larsen
f1e8307308 1. Fix a bug in psqrt and make it return 0 for +inf arguments. 
2. Simplify handling of special cases by taking advantage of the fact that the
   builtin vrsqrt approximation handles negative, zero and +inf arguments correctly.
   This speeds up the SSE and AVX implementations by ~20%.
3. Make the Newton-Raphson formula used for rsqrt more numerically robust:

Before: y = y * (1.5 - x/2 * y^2)
After: y = y * (1.5 - y * (x/2) * y)

Forming y^2 can overflow for very large or very small (denormalized) values of x, while x*y ~= 1. For AVX512, this makes it possible to compute accurate results for denormal inputs down to ~1e-42 in single precision.

4. Add a faster double precision implementation for Knights Landing using the vrsqrt28 instruction and a single Newton-Raphson iteration.

Benchmark results: https://bitbucket.org/snippets/rmlarsen/5LBq9o
 2019-11-15 17:09:46 -08:00
Rasmus Munk Larsen
13ef08e5ac Move implementation of vectorized error function erf() to SpecialFunctionsImpl.h. 2019-09-27 13:56:04 -07:00
Rasmus Munk Larsen
6de5ed08d8 Add generic PacketMath implementation of the Error Function (erf). 2019-09-19 12:48:30 -07:00
Rasmus Munk Larsen
a3298b22ec Implement vectorized versions of log1p and expm1 in Eigen using Kahan's formulas, and change the scalar implementations to properly handle infinite arguments. 
Depending on instruction set, significant speedups are observed for the vectorized path:
log1p wall time is reduced 60-93% (2.5x - 15x speedup)
expm1 wall time is reduced 0-85% (1x - 7x speedup)

The scalar path is slower by 20-30% due to the extra branch needed to handle +infinity correctly.

Full benchmarks measured on Intel(R) Xeon(R) Gold 6154 here: https://bitbucket.org/snippets/rmlarsen/MXBkpM
 2019-08-12 13:53:28 -07:00
Gael Guennebaud
c68bd2fa7a Cleanup 2018-11-30 14:32:31 +01:00
Gael Guennebaud
b477d60bc6 Extend the generic psin_float code to handle cosine and make SSE and AVX use it (-> this adds pcos for AVX) 2018-11-30 11:26:30 +01:00
Gael Guennebaud
fa7fd61eda Unify SSE/AVX psin functions. 
It is based on the SSE version which is much more accurate, though very slightly slower.
This changeset also includes the following required changes:
 - add packet-float to packet-int type traits
 - add packet float<->int reinterpret casts
 - add faster pselect for AVX based on blendv
 2018-11-27 22:41:51 +01:00
Gael Guennebaud
7655a8af6e cleanup 2018-11-26 23:21:29 +01:00
Gael Guennebaud
502f92fa10 Unify SSE and AVX pexp for double. 2018-11-26 23:12:44 +01:00
Gael Guennebaud
cf8b85d5c5 Unify SSE and AVX implementation of pexp 2018-11-26 16:36:19 +01:00
Gael Guennebaud
2c44c40114 First step toward a unification of packet log implementation, currently only SSE and AVX are unified. 
To this end, I added the following functions: pzero, pcmp_*, pfrexp, pset1frombits functions.
 2018-11-26 14:21:24 +01:00
Rasmus Munk Larsen
47150af1c8 Fix copy-paste error: Must use _mm256_cmp_ps for AVX. 2016-10-12 08:34:39 -07:00
Rasmus Munk Larsen
7f67e6dfdb Update comment for fast sqrt. 2016-10-04 15:09:11 -07:00
Rasmus Munk Larsen
3ed67cb0bb Fix a bug in the implementation of Carmack's fast sqrt algorithm in Eigen (enabled by EIGEN_FAST_MATH), which causes the vectorized parts of the computation to return -0.0 instead of NaN for negative arguments. 
Benchmark speed in Giga-sqrts/s
Intel(R) Xeon(R) CPU E5-1650 v3 @ 3.50GHz
-----------------------------------------
                    SSE        AVX
Fast=1              2.529G     4.380G
Fast=0              1.944G     1.898G
Fast=1 fixed        2.214G     3.739G

This table illustrates the worst case in terms speed impact: It was measured by repeatedly computing the sqrt of an n=4096 float vector that fits in L1 cache. For large vectors the operation becomes memory bound and the differences between the different versions almost negligible.
 2016-10-04 14:22:56 -07:00
Gael Guennebaud
a4c266f827 Factorize the 4 copies of tanh implementations, make numext::tanh consistent with array::tanh, enable fast tanh in fast-math mode only. 2016-08-23 14:23:08 +02:00
Benoit Steiner
8ce46f9d89 Improved implementation of ptanh for SSE and AVX 2016-02-18 13:24:34 -08:00
Benoit Steiner
6d8b1dce06 Avoid implicit cast from double to float. 2016-02-10 18:07:11 -08:00
Benoit Steiner
2d523332b3 Optimized implementation of the hyperbolic tangent function for AVX 2016-02-10 08:48:05 -08:00
Gael Guennebaud
d2e288ae50 Workaround compilers that do not even define _mm256_set_m128. 2015-12-24 16:53:43 +01:00
Benoit Steiner
6d777e1bc7 Fixed a typo. 2015-12-18 19:25:50 -08:00
Gael Guennebaud
3abd8470ca bug #1140: remove custom definition and use of _mm256_setr_m128 2015-12-18 14:18:59 +01:00
Gael Guennebaud
75861f6650 bug #1069: fix AVX support on MSVC (use of non portable C-style cast) 2015-09-28 10:08:26 +02:00