Rasmus Munk Larsen
|
edaa0fc5d1
|
Revert PR-292. After further investigation, the memcpy->memmove change was only good for Haswell on older versions of glibc. Adding a switch for small sizes is perhaps useful for string copies, but also has an overhead for larger sizes, making it a poor trade-off for general memcpy.
This PR also removes a couple of unnecessary semi-colons in Eigen/src/Core/AssignEvaluator.h that caused compiler warning everywhere.
|
2017-01-26 12:46:06 -08:00 |
|
Gael Guennebaud
|
25a1703579
|
Merged in ggael/eigen-flexidexing (pull request PR-294)
generalized operator() for indexed access and slicing
|
2017-01-26 08:04:23 +00:00 |
|
Gael Guennebaud
|
607be65a03
|
Fix duplicates of array_size bewteen unsupported and Core
|
2017-01-25 22:53:58 +01:00 |
|
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
|
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
|
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 |
|
NeroBurner
|
c4fc2611ba
|
add cmake-option to enable/disable creation of tests
* * *
disable unsupportet/test when test are disabled
* * *
rename EIGEN_ENABLE_TESTS to BUILD_TESTING
* * *
consider BUILD_TESTING in blas
|
2017-01-02 09:09:21 +01:00 |
|
Benoit Steiner
|
660da83e18
|
Pulled latest update from trunk
|
2016-12-21 16:43:27 -08:00 |
|
Benoit Steiner
|
4236aebe10
|
Simplified the contraction code`
|
2016-12-21 16:42:56 -08:00 |
|
Benoit Steiner
|
3cfa16f41d
|
Merged in benoitsteiner/opencl (pull request PR-279)
Fix for auto appearing in functor template argument.
|
2016-12-21 15:08:54 -08:00 |
|
Benoit Steiner
|
519d63d350
|
Added support for libxsmm kernel in multithreaded contractions
|
2016-12-21 15:06:06 -08:00 |
|
Benoit Steiner
|
0657228569
|
Simplified the way we link libxsmm
|
2016-12-21 14:40:08 -08:00 |
|
Benoit Steiner
|
f9eff17e91
|
Leverage libxsmm kernels within signle threaded contractions
|
2016-12-21 12:32:06 -08:00 |
|
Benoit Steiner
|
c19fe5e9ed
|
Added support for libxsmm in the eigen makefiles
|
2016-12-21 10:43:40 -08:00 |
|
Luke Iwanski
|
c55ecfd820
|
Fix for auto appearing in functor template argument.
|
2016-12-21 15:42:51 +00:00 |
|
Benoit Steiner
|
0f577d4744
|
Merged eigen/eigen into default
|
2016-12-20 17:02:06 -08:00 |
|
Luke Iwanski
|
29186f766f
|
Fixed order of initialisation in ExecExprFunctorKernel functor.
|
2016-12-20 21:32:42 +00:00 |
|
Gael Guennebaud
|
e8d6862f14
|
Properly adjust precision when saving to Market format.
|
2016-12-20 22:10:33 +01:00 |
|
Gael Guennebaud
|
e2f4ee1c2b
|
Speed up parsing of sparse Market file.
|
2016-12-20 21:56:21 +01:00 |
|
Luke Iwanski
|
8245851d1b
|
Matching parameters order between lambda and the functor.
|
2016-12-20 16:18:15 +00:00 |
|
Benoit Steiner
|
548ed30a1c
|
Added an OpenCL regression test
|
2016-12-19 18:56:26 -08:00 |
|
Benoit Steiner
|
27ceb43bf6
|
Fixed race condition in the tensor_shuffling_sycl test
|
2016-12-19 15:34:42 -08:00 |
|
Benoit Steiner
|
70d0172f0c
|
Merged eigen/eigen into default
|
2016-12-16 17:37:04 -08:00 |
|
Benoit Steiner
|
8910442e19
|
Fixed memcpy, memcpyHostToDevice and memcpyDeviceToHost for Sycl.
|
2016-12-16 15:45:04 -08:00 |
|
Luke Iwanski
|
54db66c5df
|
struct -> class in order to silence compilation warning.
|
2016-12-16 20:25:20 +00:00 |
|
Mehdi Goli
|
35bae513a0
|
Converting all parallel for lambda to functor in order to prevent kernel duplication name error; adding tensorConcatinationOp backend for sycl.
|
2016-12-16 19:46:45 +00:00 |
|
Mehdi Goli
|
c5e8546306
|
Adding asynchandler to sycl queue as lack of it can cause undefined behaviour.
|
2016-12-15 16:59:57 +00:00 |
|
Benoit Steiner
|
2c2e218471
|
Avoid using #define since they can conflict with user code
|
2016-12-14 19:49:15 -08:00 |
|
Benoit Steiner
|
3beb180ee5
|
Don't call EnvThread::OnCancel by default since it doesn't do anything.
|
2016-12-14 18:33:39 -08:00 |
|
Benoit Steiner
|
9ff5d0f821
|
Merged eigen/eigen into default
|
2016-12-14 17:32:16 -08:00 |
|
Mehdi Goli
|
730eb9fe1c
|
Adding asynchronous execution as it improves the performance.
|
2016-12-14 17:38:53 +00:00 |
|
Mehdi Goli
|
2d4a091beb
|
Adding tensor contraction operation backend for Sycl; adding test for contractionOp sycl backend; adding temporary solution to prevent memory leak in buffer; cleaning up cxx11_tensor_buildins_sycl.h
|
2016-12-14 15:30:37 +00:00 |
|
Benoit Steiner
|
a432fc102d
|
Moved the choice of ThreadPool to unsupported/Eigen/CXX11/ThreadPool
|
2016-12-12 15:24:16 -08:00 |
|
Benoit Steiner
|
8ae68924ed
|
Made ThreadPoolInterface::Cancel() an optional functionality
|
2016-12-12 11:58:38 -08:00 |
|
Benoit Steiner
|
76fca22134
|
Use a more accurate timer to sleep on Linux systems.
|
2016-12-09 15:12:24 -08:00 |
|
Benoit Steiner
|
4deafd35b7
|
Introduce a portable EIGEN_SLEEP macro.
|
2016-12-09 14:52:15 -08:00 |
|
Benoit Steiner
|
aafa97f4d2
|
Fixed build error with MSVC
|
2016-12-09 14:42:32 -08:00 |
|
Benoit Steiner
|
2f5b7a199b
|
Reworked the threadpool cancellation mechanism to not depend on pthread_cancel since it turns out that pthread_cancel doesn't work properly on numerous platforms.
|
2016-12-09 13:05:14 -08:00 |
|
Benoit Steiner
|
3d59a47720
|
Added a message to ease the detection of platforms on which thread cancellation isn't supported.
|
2016-12-08 14:51:46 -08:00 |
|
Benoit Steiner
|
28ee8f42b2
|
Added a Flush method to the RunQueue
|
2016-12-08 14:07:56 -08:00 |
|
Benoit Steiner
|
69ef267a77
|
Added the new threadpool cancel method to the threadpool interface based class.
|
2016-12-08 14:03:25 -08:00 |
|
Benoit Steiner
|
7bfff85355
|
Added support for thread cancellation on Linux
|
2016-12-08 08:12:49 -08:00 |
|
Benoit Steiner
|
462c28e77a
|
Merged in srvasude/eigen (pull request PR-265)
Add Expm1 support to Eigen.
|
2016-12-05 02:31:11 +00:00 |
|
Gael Guennebaud
|
4465d20403
|
Add missing generic load methods.
|
2016-12-03 21:25:04 +01:00 |
|
Srinivas Vasudevan
|
218764ee1f
|
Added support for expm1 in Eigen.
|
2016-12-02 14:13:01 -08:00 |
|
Mehdi Goli
|
592acc5bfa
|
Makingt default numeric_list works with sycl.
|
2016-12-02 17:58:30 +00:00 |
|
Mehdi Goli
|
79aa2b784e
|
Adding sycl backend for TensorPadding.h; disbaling __unit128 for sycl in TensorIntDiv.h; disabling cashsize for sycl in tensorDeviceDefault.h; adding sycl backend for StrideSliceOP ; removing sycl compiler warning for creating an array of size 0 in CXX11Meta.h; cleaning up the sycl backend code.
|
2016-12-01 13:02:27 +00:00 |
|
Benoit Steiner
|
a70393fd02
|
Cleaned up forward declarations
|
2016-11-30 21:59:07 -08:00 |
|
Benoit Steiner
|
e073de96dc
|
Moved the MemCopyFunctor back to TensorSyclDevice since it's the only caller and it makes TensorFlow compile again
|
2016-11-30 21:36:52 -08:00 |
|
Benoit Steiner
|
fca27350eb
|
Added the deallocate_all() method back
|
2016-11-30 20:45:20 -08:00 |
|