Gael Guennebaud
|
d06a48959a
|
bug #1383: Fix regression from 3.2 with LinSpaced(n,0,n-1) with n==0.
|
2017-01-25 15:27:13 +01:00 |
|
Rasmus Munk Larsen
|
ae3e43a125
|
Remove extra space.
|
2017-01-24 16:16:39 -08:00 |
|
Benoit Steiner
|
e96c77668d
|
Merged in rmlarsen/eigen2 (pull request PR-292)
Adds a fast memcpy function to Eigen.
|
2017-01-25 00:14:04 +00:00 |
|
Rasmus Munk Larsen
|
3be5ee2352
|
Update copy helper to use fast_memcpy.
|
2017-01-24 14:22:49 -08:00 |
|
Rasmus Munk Larsen
|
e6b1020221
|
Adds a fast memcpy function to Eigen. This takes advantage of the following:
1. For small fixed sizes, the compiler generates inline code for memcpy, which is much faster.
2. My colleague eriche at googl dot com discovered that for large sizes, memmove is significantly faster than memcpy (at least on Linux with GCC or Clang). See benchmark numbers measured on a Haswell (HP Z440) workstation here: https://docs.google.com/a/google.com/spreadsheets/d/1jLs5bKzXwhpTySw65MhG1pZpsIwkszZqQTjwrd_n0ic/pubhtml This is of course surprising since memcpy is a less constrained version of memmove. This stackoverflow thread contains some speculation as to the causes: http://stackoverflow.com/questions/22793669/poor-memcpy-performance-on-linux
Below are numbers for copying and slicing tensors using the multithreaded TensorDevice. The numbers show significant improvements for memcpy of very small blocks and for memcpy of large blocks single threaded (we were already able to saturate memory bandwidth for >1 threads before on large blocks). The "slicingSmallPieces" benchmark also shows small consistent improvements, since memcpy cost is a fair portion of that particular computation.
The benchmarks operate on NxN matrices, and the names are of the form BM_$OP_${NUMTHREADS}T/${N}.
Measured improvements in wall clock time:
Run on rmlarsen3.mtv (12 X 3501 MHz CPUs); 2017-01-20T11:26:31.493023454-08:00
CPU: Intel Haswell with HyperThreading (6 cores) dL1:32KB dL2:256KB dL3:15MB
Benchmark Base (ns) New (ns) Improvement
------------------------------------------------------------------
BM_memcpy_1T/2 3.48 2.39 +31.3%
BM_memcpy_1T/8 12.3 6.51 +47.0%
BM_memcpy_1T/64 371 383 -3.2%
BM_memcpy_1T/512 66922 66720 +0.3%
BM_memcpy_1T/4k 9892867 6849682 +30.8%
BM_memcpy_1T/5k 14951099 10332856 +30.9%
BM_memcpy_2T/2 3.50 2.46 +29.7%
BM_memcpy_2T/8 12.3 7.66 +37.7%
BM_memcpy_2T/64 371 376 -1.3%
BM_memcpy_2T/512 66652 66788 -0.2%
BM_memcpy_2T/4k 6145012 6117776 +0.4%
BM_memcpy_2T/5k 9181478 9010942 +1.9%
BM_memcpy_4T/2 3.47 2.47 +31.0%
BM_memcpy_4T/8 12.3 6.67 +45.8
BM_memcpy_4T/64 374 376 -0.5%
BM_memcpy_4T/512 67833 68019 -0.3%
BM_memcpy_4T/4k 5057425 5188253 -2.6%
BM_memcpy_4T/5k 7555638 7779468 -3.0%
BM_memcpy_6T/2 3.51 2.50 +28.8%
BM_memcpy_6T/8 12.3 7.61 +38.1%
BM_memcpy_6T/64 373 378 -1.3%
BM_memcpy_6T/512 66871 66774 +0.1%
BM_memcpy_6T/4k 5112975 5233502 -2.4%
BM_memcpy_6T/5k 7614180 7772246 -2.1%
BM_memcpy_8T/2 3.47 2.41 +30.5%
BM_memcpy_8T/8 12.4 10.5 +15.3%
BM_memcpy_8T/64 372 388 -4.3%
BM_memcpy_8T/512 67373 66588 +1.2%
BM_memcpy_8T/4k 5148462 5254897 -2.1%
BM_memcpy_8T/5k 7660989 7799058 -1.8%
BM_memcpy_12T/2 3.50 2.40 +31.4%
BM_memcpy_12T/8 12.4 7.55 +39.1
BM_memcpy_12T/64 374 378 -1.1%
BM_memcpy_12T/512 67132 66683 +0.7%
BM_memcpy_12T/4k 5185125 5292920 -2.1%
BM_memcpy_12T/5k 7717284 7942684 -2.9%
BM_slicingSmallPieces_1T/2 47.3 47.5 +0.4%
BM_slicingSmallPieces_1T/8 53.6 52.3 +2.4%
BM_slicingSmallPieces_1T/64 491 476 +3.1%
BM_slicingSmallPieces_1T/512 21734 18814 +13.4%
BM_slicingSmallPieces_1T/4k 394660 396760 -0.5%
BM_slicingSmallPieces_1T/5k 218722 209244 +4.3%
BM_slicingSmallPieces_2T/2 80.7 79.9 +1.0%
BM_slicingSmallPieces_2T/8 54.2 53.1 +2.0
BM_slicingSmallPieces_2T/64 497 477 +4.0%
BM_slicingSmallPieces_2T/512 21732 18822 +13.4%
BM_slicingSmallPieces_2T/4k 392885 390490 +0.6%
BM_slicingSmallPieces_2T/5k 221988 208678 +6.0%
BM_slicingSmallPieces_4T/2 80.8 80.1 +0.9%
BM_slicingSmallPieces_4T/8 54.1 53.2 +1.7%
BM_slicingSmallPieces_4T/64 493 476 +3.4%
BM_slicingSmallPieces_4T/512 21702 18758 +13.6%
BM_slicingSmallPieces_4T/4k 393962 404023 -2.6%
BM_slicingSmallPieces_4T/5k 249667 211732 +15.2%
BM_slicingSmallPieces_6T/2 80.5 80.1 +0.5%
BM_slicingSmallPieces_6T/8 54.4 53.4 +1.8%
BM_slicingSmallPieces_6T/64 488 478 +2.0%
BM_slicingSmallPieces_6T/512 21719 18841 +13.3%
BM_slicingSmallPieces_6T/4k 394950 397583 -0.7%
BM_slicingSmallPieces_6T/5k 223080 210148 +5.8%
BM_slicingSmallPieces_8T/2 81.2 80.4 +1.0%
BM_slicingSmallPieces_8T/8 58.1 53.5 +7.9%
BM_slicingSmallPieces_8T/64 489 480 +1.8%
BM_slicingSmallPieces_8T/512 21586 18798 +12.9%
BM_slicingSmallPieces_8T/4k 394592 400165 -1.4%
BM_slicingSmallPieces_8T/5k 219688 208301 +5.2%
BM_slicingSmallPieces_12T/2 80.2 79.8 +0.7%
BM_slicingSmallPieces_12T/8 54.4 53.4 +1.8
BM_slicingSmallPieces_12T/64 488 476 +2.5%
BM_slicingSmallPieces_12T/512 21931 18831 +14.1%
BM_slicingSmallPieces_12T/4k 393962 396541 -0.7%
BM_slicingSmallPieces_12T/5k 218803 207965 +5.0%
|
2017-01-24 13:55:18 -08:00 |
|
Rasmus Munk Larsen
|
7b6aaa3440
|
Fix NaN propagation for AVX512.
|
2017-01-24 13:37:08 -08:00 |
|
Rasmus Munk Larsen
|
5e144bbaa4
|
Make NaN propagatation consistent between the pmax/pmin and std::max/std::min. This makes the NaN propagation consistent between the scalar and vectorized code paths of Eigen's scalar_max_op and scalar_min_op.
See #1373 for details.
|
2017-01-24 13:32:50 -08:00 |
|
Gael Guennebaud
|
d83db761a2
|
Add support for std::integral_constant
|
2017-01-24 16:28:12 +01:00 |
|
Gael Guennebaud
|
bc10201854
|
Add test for multiple symbols
|
2017-01-24 16:27:51 +01:00 |
|
Gael Guennebaud
|
c43d254d13
|
Fix seq().reverse() in c++98
|
2017-01-24 11:36:43 +01:00 |
|
Gael Guennebaud
|
5783158e8f
|
Add unit test for FixedInt and Symbolic
|
2017-01-24 10:55:12 +01:00 |
|
Gael Guennebaud
|
ddd83f82d8
|
Add support for "SymbolicExpr op fix<N>" in C++98/11 mode.
|
2017-01-24 10:54:42 +01:00 |
|
Gael Guennebaud
|
228fef1b3a
|
Extended the set of arithmetic operators supported by FixedInt (-,+,*,/,%,&,|)
|
2017-01-24 10:53:51 +01:00 |
|
Gael Guennebaud
|
bb52f74e62
|
Add internal doc
|
2017-01-24 10:13:35 +01:00 |
|
Gael Guennebaud
|
41c523a0ab
|
Rename fix_t to FixedInt
|
2017-01-24 09:39:49 +01:00 |
|
Gael Guennebaud
|
156e6234f1
|
bug #1375: fix cmake installation with cmake 2.8
|
2017-01-24 09:16:40 +01:00 |
|
Gael Guennebaud
|
ba3f977946
|
bug #1376: add missing assertion on size mismatch with compound assignment operators (e.g., mat += mat.col(j))
|
2017-01-23 22:06:08 +01:00 |
|
Gael Guennebaud
|
b0db4eff36
|
bug #1382: move using std::size_t/ptrdiff_t to Eigen's namespace (still better than the global namespace!)
|
2017-01-23 22:03:57 +01:00 |
|
Gael Guennebaud
|
ca79c1545a
|
Add std:: namespace prefix to all (hopefully) instances if size_t/ptrdfiff_t
|
2017-01-23 22:02:53 +01:00 |
|
Gael Guennebaud
|
4b607b5692
|
Use Index instead of size_t
|
2017-01-23 22:00:33 +01:00 |
|
Luke Iwanski
|
bf44fed9b7
|
Allows AMD APU
|
2017-01-23 15:56:45 +00:00 |
|
Gael Guennebaud
|
0fe278f7be
|
bug #1379: fix compilation in sparse*diagonal*dense with openmp
|
2017-01-21 23:27:01 +01:00 |
|
Gael Guennebaud
|
22a172751e
|
bug #1378: fix doc (DiagonalIndex vs Diagonal)
|
2017-01-21 22:09:59 +01:00 |
|
Mehdi Goli
|
602f8c27f5
|
Reverting back to the previous TensorDeviceSycl.h as the total number of buffer is not enough for tensorflow.
|
2017-01-20 18:23:20 +00:00 |
|
Gael Guennebaud
|
4d302a080c
|
Recover compile-time size from seq(A,B) when A and B are fixed values. (c++11 only)
|
2017-01-19 20:34:18 +01:00 |
|
Gael Guennebaud
|
54f3fbee24
|
Exploit fixed values in seq and reverse with C++98 compatibility
|
2017-01-19 19:57:32 +01:00 |
|
Gael Guennebaud
|
7691723e34
|
Add support for fixed-value in symbolic expression, c++11 only for now.
|
2017-01-19 19:25:29 +01:00 |
|
Benoit Steiner
|
924600a0e8
|
Made sure that enabling avx2 instructions enables avx and sse instructions as well.
|
2017-01-19 09:54:48 -08:00 |
|
Mehdi Goli
|
77cc4d06c7
|
Removing unused variables
|
2017-01-19 17:06:21 +00:00 |
|
Mehdi Goli
|
837fdbdcb2
|
Merging with Benoit's upstream.
|
2017-01-19 11:34:34 +00:00 |
|
Mehdi Goli
|
6bdd15f572
|
Adding non-deferrenciable pointer track for ComputeCpp backend; Adding TensorConvolutionOp for ComputeCpp; fixing typos. modifying TensorDeviceSycl to use the LegacyPointer class.
|
2017-01-19 11:30:59 +00:00 |
|
Benoit Steiner
|
aa7fb88dfa
|
Merged in LaFeuille/eigen (pull request PR-289)
Fix a typo
|
2017-01-18 16:44:39 -08:00 |
|
Gael Guennebaud
|
e84ed7b6ef
|
Remove dead code
|
2017-01-18 23:18:28 +01:00 |
|
Gael Guennebaud
|
f3ccbe0419
|
Add a Symbolic::FixedExpr helper expression to make sure the compiler fully optimize the usage of last and end.
|
2017-01-18 23:16:32 +01:00 |
|
Mehdi Goli
|
c6f7b33834
|
Applying Benoit's comment. Embedding synchronisation inside device memcpy so there is no need to externally call synchronise() for device memcopy.
|
2017-01-18 10:45:28 +00:00 |
|
Gael Guennebaud
|
15471432fe
|
Add a .reverse() member to ArithmeticSequence.
|
2017-01-18 11:35:27 +01:00 |
|
Gael Guennebaud
|
e4f8dd860a
|
Add missing operator*
|
2017-01-18 10:49:01 +01:00 |
|
Gael Guennebaud
|
198507141b
|
Update all block expressions to accept compile-time sizes passed by fix<N> or fix<N>(n)
|
2017-01-18 09:43:58 +01:00 |
|
Gael Guennebaud
|
5484ddd353
|
Merge the generic and dynamic overloads of block()
|
2017-01-17 22:11:46 +01:00 |
|
Gael Guennebaud
|
655ba783f8
|
Defer set-to-zero in triangular = product so that no aliasing issue occur in the common:
A.triangularView() = B*A.sefladjointView()*B.adjoint()
case that used to work in 3.2.
|
2017-01-17 18:03:35 +01:00 |
|
Gael Guennebaud
|
5e36ec3b6f
|
Fix regression when passing enums to operator()
|
2017-01-17 17:10:16 +01:00 |
|
Gael Guennebaud
|
f7852c3d16
|
Fix -Wunnamed-type-template-args
|
2017-01-17 16:05:58 +01:00 |
|
Gael Guennebaud
|
4f36dcfda8
|
Add a generic block() method compatible with Eigen::fix
|
2017-01-17 11:34:28 +01:00 |
|
Gael Guennebaud
|
71e5b71356
|
Add a get_runtime_value helper to deal with pointer-to-function hack,
plus some refactoring to make the internals more consistent.
|
2017-01-17 11:33:57 +01:00 |
|
Gael Guennebaud
|
59801a3250
|
Add \newin{3.x} doxygen command
|
2017-01-17 10:31:28 +01:00 |
|
Gael Guennebaud
|
23bfcfc15f
|
Add missing overload of get_compile_time for c++98/11
|
2017-01-17 10:30:21 +01:00 |
|
Gael Guennebaud
|
edff32c2c2
|
Disambiguate the two versions of fix for doxygen
|
2017-01-17 10:29:33 +01:00 |
|
Gael Guennebaud
|
4989922be2
|
Add support for symbolic expressions as arguments of operator()
|
2017-01-16 22:21:23 +01:00 |
|
Gael Guennebaud
|
12e22a2844
|
typos in doc
|
2017-01-16 16:31:19 +01:00 |
|
Gael Guennebaud
|
e70c4c97fa
|
Typo
|
2017-01-16 16:20:16 +01:00 |
|