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
10,702 Commits 15 Branches 93 Tags
d38e6fbc27abe0c354ffe90928f6741c378e76e1
Commit Graph

56 Commits

Author SHA1 Message Date
Rasmus Munk Larsen
6de5ed08d8 Add generic PacketMath implementation of the Error Function (erf). 2019-09-19 12:48:30 -07:00
Gael Guennebaud
55b63d4ea3 Fix compilation without vector engine available (e.g., x86 with SSE disabled): 
-> ppolevl is required by ndtri even for the scalar path
 2019-09-05 18:16:46 +02:00
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
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
luz.paz
e3912f5e63 MIsc. source and comment typos 
Found using `codespell` and `grep` from downstream FreeCAD
 2018-03-11 10:01:44 -04:00
Rasmus Munk Larsen
765615609d Update comment for fast sqrt. 2016-10-04 15:08:41 -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
bfb3fcd94f Optimized implementation of the tanh function for SSE 2016-02-10 08:52:30 -08:00
Benoit Jacob
e6ee18d6b4 Make the GCC workaround for sqrt GCC-only; detect Emscripten as non-GCC 2016-02-10 11:11:49 -05:00
Benoit Jacob
964a95bf5e Work around Emscripten bug - https://github.com/kripken/emscripten/issues/4088 2016-02-10 10:37:22 -05:00
Gael Guennebaud
7cae8918c0 Fix compilation on old gcc+AVX 2016-01-21 20:30:32 +01:00
Gael Guennebaud
8dca9f97e3 Add numext::sqrt function to enable custom optimized implementation. 
This changeset add two specializations for float/double on SSE. Those
are mostly usefull with GCC for which std::sqrt add an extra and costly
check on the result of _mm_sqrt_*. Clang does not add this burden.

In this changeset, only DenseBase::norm() makes use of it.
 2016-01-21 20:18:51 +01:00
Benoit Steiner
4fd7f47692 Added an optimized version of rsqrt for SSE and AVX that is used when EIGEN_FAST_MATH is defined. 2015-03-02 09:38:47 -08:00
Benoit Steiner
f41b1f1666 Added support for fast reciprocal square root computation. 2015-02-26 09:42:41 -08:00
Gael Guennebaud
eb563049f7 Remove some dead stores. 2015-02-18 11:26:48 +01:00
Christoph Hertzberg
84aaa03182 Addendum to bug #859: pexp(NaN) for double did not return NaN, also, plog(NaN) did not return NaN. 
psqrt(NaN) and psqrt(-1) shall return NaN if EIGEN_FAST_MATH==0
 2014-10-20 13:13:43 +02:00
Gael Guennebaud
aa5f79206f Fix bug #859: pexp(NaN) returned Inf instead of NaN 2014-10-20 11:38:51 +02:00
Gael Guennebaud
5c5231ab71 Workaround gcc's default ABI not being able to distinghish between vector types of different sizes. 2014-04-22 16:03:19 +02:00
Gael Guennebaud
9f3f42d66a fix a few "dead stores" warnings 2013-10-26 13:59:02 +02:00
Gael Guennebaud
c47010e3d2 typo 2013-08-19 16:10:00 +02:00
Gael Guennebaud
d4dd6aaed2 Fix bug #642: add vectorization of sqrt for doubles, and make sqrt really safe if EIGEN_FAST_MATH is disabled 2013-08-19 16:02:27 +02:00
Gael Guennebaud
9f11f80db1 Make psqrt works with numeric_limits<float>::min 2013-06-14 10:55:05 +02:00
Jeff Dean
d5fa5001a7 Fix bug #613: psqrt was incorrect for small numbers 2013-06-13 18:17:27 +02:00
Gael Guennebaud
8745da14d8 Fix SSE plog<float> to return -INF on 0 2013-02-14 23:34:05 +01:00
Gael Guennebaud
7d98c864ff fix warning 2012-08-01 10:44:59 +02:00
Gael Guennebaud
22e0ebbc2c fix lower acceptable bound of SSE pexp for double 2012-07-31 23:11:04 +02:00
Gael Guennebaud
e8aa1f00c5 add SSE pexp function for double, make use of _mm_floor_p* for pexp with SSE4.1 2012-07-27 23:40:04 +02:00
Benoit Jacob
69124cfca2 Automatic relicensing to MPL2 using Keirs script. Manual fixup follows. 2012-07-13 14:42:47 -04:00
Gael Guennebaud
a3e700db72 fix bug #475: .exp() now returns +inf when overflow occurs (SSE) 2012-06-14 10:38:39 +02:00
Jitse Niesen
3c412183b2 Get rid of include directives inside namespace blocks (bug #339). 2012-04-15 11:06:28 +01:00
Benoit Jacob
4716040703 bug #86 : use internal:: namespace instead of ei_ prefix 2010-10-25 10:15:22 -04:00
Gael Guennebaud
aa2b46aa91 allow vectorization of mat44.col() by adding a InnerPanel boolean 
template parameter to Block
 2010-07-23 16:29:29 +02:00
Benoit Jacob
97ced33b33 Backed out changeset 40f6e26a24 
See thread on mailing list: "InnerPanel change mis-detects alignment?"
 2010-08-11 00:04:06 -04:00
Gael Guennebaud
40f6e26a24 allow vectorization of mat44.col() by adding a InnerPanel boolean 
template parameter to Block
 2010-07-23 16:29:29 +02:00
Gael Guennebaud
ff96c94043 mixing types in product step 2: 
* pload* and pset1 are now templated on the packet type
* gemv routines are now embeded into a structure with
  a consistent API with respect to gemm
* some configurations of vector * matrix and matrix * matrix works fine,
  some need more work...
 2010-07-11 15:48:30 +02:00
Gael Guennebaud
28e64b0da3 email change 2010-06-24 23:21:58 +02:00
Benoit Jacob
f0a6d56f07 fix linking errors with multiply defined functions 2010-06-18 09:01:34 -04:00
Benoit Jacob
134ca4acb3 packet math functions: 
- take const Packet& args like the other packet funcs
 - SSE specializations: make them be actual template specializations
 2010-06-15 08:29:21 -04:00
Hauke Heibel
9d6afdeb22 ei_psqrt fix for zero input 2010-04-01 15:10:52 +02:00
Hauke Heibel
3ea1f97f69 Suppressed the warning for missing assignment generators (forgot that in the last submission). 
Commented Quake3's fast inverser sqrt in SSE's MathFunction header.
 2009-12-15 08:09:14 +01:00
Benoit Jacob
6347b1db5b remove sentence "Eigen itself is part of the KDE project." 
it never made very precise sense. but now does it still make any?
 2009-05-22 20:25:33 +02:00
Gael Guennebaud
1e286464ab * compilation fixes for gcc 3.3 
* test Part::swap
 2009-05-06 08:43:38 +00:00