Commit Graph

383 Commits

Author SHA1 Message Date
Benoit Jacob
19bf13aa62 Automatically serialize partial results to disk, reboot, and resume, when timings are getting bad 2015-03-06 19:11:50 -05:00
Gael Guennebaud
4c8eeeaed6 update gemm changeset list 2015-03-06 15:08:20 +01:00
Gael Guennebaud
eedd5063fd Update gemm performance monitoring tool:
- permit to recompute a subset of changesets
 - update changeset list
 - add a few more cases
2015-03-06 11:47:13 +01:00
Benoit Jacob
4ab01f7c21 slightly increase tolerance to clock speed variation 2015-03-05 14:41:16 -05:00
Benoit Jacob
5db2baa573 Make benchmark-blocking-sizes detect changes to clock speed and be resilient to that. 2015-03-05 13:44:20 -05:00
Benoit Jacob
2231b3dece output to cout, not cerr, the actual results 2015-03-04 09:45:12 -05:00
Benoit Jacob
00ea121881 Complete the tool to analyze the efficiency of default sizes. 2015-03-04 09:30:56 -05:00
Benoit Jacob
f64b4480af Add missing copyright notices 2015-03-03 11:43:56 -05:00
Benoit Jacob
eae8e27b7d Add a benchmark-default-sizes action to benchmark-blocking-sizes.cpp 2015-03-03 11:41:21 -05:00
Benoit Jacob
9930e9583b Improve analyze-blocking-sizes, and in particular give it a evaluate-defaults tool
that shows the efficiency of Eigen's default blocking sizes choices, using a
previously computed table from benchmark-blocking-sizes.
2015-03-02 18:08:38 -05:00
Gael Guennebaud
2e9cb06a87 Update changeset list to be checked by perf_monitoring/gemm. 2015-02-26 16:13:33 +01:00
Gael Guennebaud
a46061ab7b Make perf_monitoring/gemm script more flexible:
- skip existing dataset
  - add a "-up" option to recompute the dataset (see script header)
  - allow to specify a filename prefix
2015-02-26 16:12:58 +01:00
Benoit Jacob
488874781b Add analyze-blocking-sizes program under bench/ to analyze multiple logs
generated by benchmark-blocking-sizes.
2015-02-23 14:02:29 -05:00
Benoit Jacob
458cf91cd9 Add benchmark-blocking-sizes.cpp to bench/ per mailing list discussion. 2015-02-20 17:08:04 -05:00
Gael Guennebaud
03ec601ff7 Initial version of a small script to help tracking performance regressions 2015-02-20 19:20:34 +01:00
Gael Guennebaud
333b497383 update bench_gemm 2015-02-20 11:59:49 +01:00
Benoit Steiner
c739102ef9 Pulled the latest changes from the trunk 2015-02-06 05:25:03 -08:00
Benoit Steiner
46fc881e4a Added a few benchmarks for the tensor code 2015-01-26 17:46:40 -08:00
Benoit Steiner
7acd38d19e Created some benchmarks for the tensor code 2014-10-17 09:49:03 -07:00
Konstantinos Margaritis
60e093a9dc Merged eigen/eigen into default 2014-09-21 14:02:51 +03:00
Konstantinos Margaritis
7ff266e3ce Initial VSX commit 2014-08-29 20:03:49 +00:00
Gael Guennebaud
57f71a5552 Update bench_norm utility 2014-09-11 10:27:46 +02:00
Chen-Pang He
1eefa5a841 Find benchmark opponents also in /usr/lib64 2014-07-07 22:55:28 +08:00
Chen-Pang He
e4b6979334 Find OpenBLAS more aggressively. This made a difference on Fedora 20 2014-07-07 21:32:33 +08:00
Florian George
f56d452c7e Enable atv in Blaze Benchmark 2014-05-04 17:07:17 +02:00
Florian George
af79b158a1 Use trans(X) instead of X.transpose() in Blaze Benchmark 2014-05-04 17:06:34 +02:00
Gael Guennebaud
2fb64578aa Add a small benchmark to compare dense solvers for small to large problems. 2014-04-28 16:16:29 +02:00
Gael Guennebaud
c354bd47f7 Make our gemm bench a little more powerful. 2014-04-17 21:03:26 +02:00
Gael Guennebaud
9777a5ca60 Various minor fixes in BTL 2014-04-17 21:01:45 +02:00
Benoit Steiner
aecc78325a Pulled the latest updates from the eigen trunk. 2014-04-01 22:07:05 -07:00
Florian George
56c4851323 Fixed typo: symmretric -> symmetric 2014-04-01 15:52:25 +02:00
Gael Guennebaud
1221dd90aa Fix no newline at end of file warning 2014-04-01 11:21:14 +02:00
Gael Guennebaud
93870d95b7 BTL: add blaze 2014-03-31 10:59:55 +02:00
Gael Guennebaud
f603823ef3 BTL: fix warnings and extend to 5k matrices, update GotoBlas to OpenBlas, etc. 2014-03-31 10:58:30 +02:00
Gael Guennebaud
33ca9b4ee6 Add support for OSX in BTL and fix a few warnings 2014-03-07 23:11:38 +01:00
Gael Guennebaud
c0e08e9e4b fix stable norm benchmark 2014-02-13 15:53:51 +01:00
Gael Guennebaud
05c9be65ce Fix bug #595: typo 2013-06-10 13:10:36 +02:00
Christoph Hertzberg
6300e8ca02 replaced compiler specific __attribute__((noinline)) by EIGEN_DONT_INLINE 2012-12-17 16:55:14 +01:00
Jakob Schwendner
22e6741da9 updated geometry benchmark to handle additional cases 2012-12-17 09:33:22 +01:00
Jakob Schwendner
98798e904b added benchmark for test vectorization in geometry package 2012-12-16 23:30:56 +01:00
Desire NUENTSA
15a9f6b9c1 Doc for sparseLU 2012-09-25 11:48:18 +02:00
Desire NUENTSA
45672e724e Incomplete Cholesky preconditioner... not yet stable 2012-09-11 12:12:19 +02:00
Desire NUENTSA
2c99d84133 add SparseLU in sparse bench 2012-09-10 12:41:26 +02:00
Desire NUENTSA
fdd0f0c5fc merge Sparse LU branch 2012-09-07 13:18:16 +02:00
Desire NUENTSA
063705b5be Add tutorial for sparse solvers 2012-09-07 13:14:57 +02:00
Desire NUENTSA
2280f2490e Init perf values 2012-09-04 12:21:07 +02:00
Desire NUENTSA
288e6aab14 Insert XSL styles into output XML files 2012-09-03 10:33:39 +02:00
Desire NUENTSA W.
fe9956defe Read real and complex bench matrices from a unique folder
Output and display bench results using XML and XSLT
2012-08-27 22:52:43 +02:00
Desire NUENTSA
4d3b7e2a13 Add support for Metis fill-reducing ordering ; it is generally more efficient than COLAMD ordering 2012-08-06 14:55:02 +02:00
Desire NUENTSA
7dc39b7037 Add unit tests 2012-08-03 13:05:45 +02:00
Desire NUENTSA W.
c0fa5811ec Refactoring codes for numeric updates 2012-07-27 11:36:58 +02:00
Desire NUENTSA
925ace196c correct bug in the complex version 2012-07-19 18:15:23 +02:00
Desire NUENTSA
59642da88b Add exception handler to memory allocation 2012-07-19 18:03:44 +02:00
Desire NUENTSA
b0cba2d988 Add a draft (not clean ) version of the COLAMD ordering implementation 2012-07-18 16:59:00 +02:00
Benoit Jacob
b596f6c10c remove outdated "Eigen itself is part of the KDE project" outside of eigen2 files 2012-07-15 10:33:40 -04:00
Benoit Jacob
69124cfca2 Automatic relicensing to MPL2 using Keirs script. Manual fixup follows. 2012-07-13 14:42:47 -04:00
Gael Guennebaud
ff12a6cd43 fix include path 2012-07-08 16:18:51 +02:00
Desire NUENTSA
b5a83867ca Update Ordering interface 2012-07-06 20:18:16 +02:00
Desire NUENTSA
15f1563533 Before moving to the new building 2012-06-29 17:45:10 +02:00
Gael Guennebaud
eb626877d7 fix sparse benchmark help 2012-06-21 10:53:36 +02:00
Desire NUENTSA
0c9b08e46e build complete... almost 2012-06-14 18:45:04 +02:00
Desire NUENTSA
f8a0745cb0 Build process... 2012-06-13 18:26:05 +02:00
Gael Guennebaud
5cec86cb1e BTL: add missing TRMM plots, update Eigen's interface 2012-06-07 18:35:38 +02:00
Gael Guennebaud
83c932ed15 fix a warning 2012-06-07 18:22:13 +02:00
Desire NUENTSA
1b9097644d Add common options to the benchmark interface 2012-05-25 17:58:43 +02:00
Desire NUENTSA
2d35f88bcf Cholmod does not compute a determinant 2012-03-29 19:07:13 +02:00
Desire NUENTSA
ada9e79145 add a benchmark routine for all sparse linear solvers in Eigen 2012-03-29 14:29:55 +02:00
Marc Glisse
8107b3da75 bug #383 - EIGEN_ASM_COMMENT broken in C++11
this is due to the new user-defined literals syntax.
2011-11-26 17:55:18 -05:00
Bram de Jong
961a825b97 Add method which returns worst time (and make some methods const). 2011-09-26 14:39:23 +01:00
Gael Guennebaud
edf4c4b217 add support for macosx 2011-09-17 10:57:27 +02:00
Gael Guennebaud
3e7aaadb1d fix bench_gemm 2011-09-09 10:36:20 +02:00
Eamon Nerbonne
e48bc0dfe3 WIN32 isn't defined ?? but _WIN32 is. 2011-04-19 14:37:04 +02:00
Gael Guennebaud
4ada45bc76 BTL: add eigen2 backend 2011-03-23 16:59:12 +01:00
Gael Guennebaud
7bb4f6ae2f BTL: do not enable GOTO1 if GOTO2 was found 2011-03-23 16:28:43 +01:00
Gael Guennebaud
3ef0da6efb fix tridiagonalization action 2011-03-23 16:28:09 +01:00
Gael Guennebaud
611fc17894 add support for ublas 2011-03-23 11:39:35 +01:00
Gael Guennebaud
ec32d2c807 BTL: by default use current Eigen headers, and disable the novec version 2011-03-23 11:08:10 +01:00
Gael Guennebaud
b3e43246bc BTL: add a Eigen-blas backend 2011-03-23 11:00:31 +01:00
Gael Guennebaud
f9da1ccc3b BTL: clean the BLAS implementation 2011-03-23 10:35:54 +01:00
Gael Guennebaud
e35b1ef3f3 BTL: rm stupid backends 2011-03-23 10:07:24 +01:00
Gael Guennebaud
fe595e91ae update plot settings 2011-03-23 10:03:01 +01:00
Gael Guennebaud
9cca79f5ca update aat action to do a syrk operation, and remove (comment) ata action 2011-03-23 10:02:00 +01:00
Gael Guennebaud
da3f3586e0 BTl: GMM++ LU is not a full pivoting LU 2011-03-22 15:39:23 +01:00
Gael Guennebaud
f7cd63b964 fix bug #189 (issue with fortran concentions to return COMPLEX values) 2011-02-18 15:11:31 +01:00
Gael Guennebaud
1eae6d0fb9 an even more stable procedure 2011-02-03 11:25:34 +01:00
Gael Guennebaud
5beb2f4f0d slightly more stable eigen vector computation 2011-02-03 10:31:45 +01:00
Gael Guennebaud
e2642ed620 clean the script to generate the plots 2011-01-31 12:45:18 +01:00
Gael Guennebaud
3874e6a72b include cblas.h header file to ease configuration 2011-01-31 11:02:59 +01:00
Gael Guennebaud
476cb4c65c fix name collision 2011-01-31 10:54:21 +01:00
Gael Guennebaud
9a73bfeb85 add GOTO2 and clean a bit the cmake macros 2011-01-31 10:45:03 +01:00
Gael Guennebaud
1731a432e7 fix BTL cholesky action and output errors if the factorization failed 2011-01-28 11:24:18 +01:00
Gael Guennebaud
c90d0c363b improve automatic handling of gotoblas and atlas 2011-01-26 19:39:10 +01:00
Gael Guennebaud
0e8a532f87 always link to gfortran for gotoblas, it seems to be harmless for 1.x but needed for 2.x 2011-01-26 19:16:06 +01:00
Gael Guennebaud
240bfdd142 finish the move to Eigen3 in BTL, and let's use our own FindEigen3.cmake script 2011-01-26 19:12:35 +01:00
Gael Guennebaud
86acb46518 pass to eigen3 ;) 2011-01-26 18:41:06 +01:00
Gael Guennebaud
faeae169dd fix compilation 2011-01-26 17:58:17 +01:00
Gael Guennebaud
210a280daf update FindMKL to match the default installation behavior of MKL 11 2011-01-26 17:58:01 +01:00
Gael Guennebaud
147a63c4b5 compilation fix 2010-12-09 19:46:26 +01:00
Benoit Jacob
4716040703 bug #86 : use internal:: namespace instead of ei_ prefix 2010-10-25 10:15:22 -04:00
Benoit Jacob
e259f71477 rename PlanarRotation -> JacobiRotation 2010-10-19 21:56:26 -04:00
Gael Guennebaud
f9123df772 fix unitialized quaternion 2010-09-08 12:57:33 +02:00
Gael Guennebaud
d591b0466d add a bench to compare various transformation methods 2010-09-07 18:21:36 +02:00
Gael Guennebaud
e0ea25fc21 add missing copyrights 2010-09-01 12:59:38 +02:00
Benoit Jacob
e17d17cea3 didn't want to commit that bench change. 2010-08-24 10:57:22 -04:00
Benoit Jacob
bd8d06033d make a couple of typedefs public so stuff compiles 2010-08-24 10:53:33 -04:00
Benoit Jacob
6924d4eec5 update this test to build against current eigen.
remove the 'normal' path as it was not compiling anymore and I couldn't see the point of it (?)
2010-08-23 23:21:25 -04:00
Gael Guennebaud
7020f30da3 sync with default branch 2010-07-22 16:29:35 +02:00
Gael Guennebaud
06250a154c add matlab-like mixed product 2010-07-22 13:19:09 +02:00
Gael Guennebaud
4161b8be67 sync 2010-07-10 22:58:51 +02:00
Thomas Capricelli
551cb9b7b4 bench: use of Eigen/Array is deprecated + fix includes for iostream 2010-07-09 03:59:36 +02:00
Gael Guennebaud
02fd3acd81 update to support mixin types 2010-07-07 19:49:48 +02:00
Gael Guennebaud
861962c55f sync 2010-07-07 16:44:05 +02:00
Gael Guennebaud
0f2d480af0 add support for complex 2010-07-07 16:41:29 +02:00
Gael Guennebaud
e1eccfad3f add intitial support for the vectorization of complex<float> 2010-07-05 16:18:09 +02:00
Konstantinos Margaritis
1daf9b11ba check for !x86 platforms, otherwise the BTL benchmark doesn't compile on arm/powerpc 2010-07-05 16:42:11 +03:00
Gael Guennebaud
f096452dfd Fix cache computation on old Intel CPUs which do not
support the cpuid function 0x4
2010-06-27 00:17:38 +02:00
Gael Guennebaud
5e7bd967cc add the manual Intel's way to query cache info 2010-06-26 23:37:42 +02:00
Gael Guennebaud
78d3c54631 add a small bench demoing the possibilities of a direct 3x3 eigen decomposition 2010-07-18 17:26:06 +02:00
Gael Guennebaud
2a820d41df finish/fix level1 blas, all test pass 2010-07-17 13:49:43 +02:00
Gael Guennebaud
f59226e901 fix compilation of blas lib 2010-07-16 22:27:24 +02:00
Gael Guennebaud
6a370f50c7 MPRealSupport was missing 2010-07-15 20:45:45 +02:00
Gael Guennebaud
e4f3759c4d add a bench for quaternion multiplication 2010-07-13 13:29:35 +02:00
Gael Guennebaud
931027f31b add a utilility to debug cpuid, and makes sure we get 0 if we query an unsupported cpuid function 2010-06-26 23:15:06 +02:00
Gael Guennebaud
28e64b0da3 email change 2010-06-24 23:21:58 +02:00
Gael Guennebaud
002f7114d1 add support for oski 2010-06-24 23:21:45 +02:00
Gael Guennebaud
98fec45d3c btl: add a trmm action and update eigen interface 2010-06-23 22:10:49 +02:00
Gael Guennebaud
b284bb8bba add a spmv mini becnhmark for Eigen, GMM++, ublas, mtl4, and oski 2010-06-22 21:39:55 +02:00
Gael Guennebaud
fd9a9fa0ae slightly optimize computeProductBlockingSizes by explicitely precomputing what is known at compile time 2010-06-22 11:10:38 +02:00
Gael Guennebaud
98686ab86c fix in case we don't know how to query the L1/L2 cache sizes 2010-06-21 23:44:20 +02:00
Gael Guennebaud
0212eec23f simplify and optimize block sizes computation for matrix products. They
are now automatically computed from the L1 and L2 cache sizes which are
themselves automatically determined at runtime.
2010-06-21 23:28:50 +02:00
Gael Guennebaud
4cd38b333c make bench_gemm print out the queried cache sizes 2010-06-21 12:07:05 +02:00
Gael Guennebaud
6db6e358f5 add the possibility to set the cache size at runtime 2010-06-18 23:25:57 +02:00
Gael Guennebaud
5b192930b6 add runtime API to control multithreading 2010-06-10 23:30:15 +02:00
Gael Guennebaud
0116261407 make BenchTimer compatible with 2.0 branch 2010-06-01 13:57:38 +02:00
Benoit Jacob
abbe260905 remove USING_PART_OF_NAMESPACE_EIGEN, leaving it in Eigen2Support.
improve porting-Eigen2-to-3 docs
2010-04-22 18:27:13 -04:00
Hauke Heibel
51b0159c96 Fixed line endings. 2010-03-05 18:11:54 +01:00
Gael Guennebaud
f2a246c225 add a small program to bench all combinations of small products 2010-03-05 17:16:19 +01:00
Gael Guennebaud
c442208358 clean a bit the bench_gemm files 2010-03-05 11:35:43 +01:00
Gael Guennebaud
24ef5fedcd minor cleaning 2010-03-05 09:57:04 +01:00
Gael Guennebaud
cefd9b8888 merge with default branch 2010-03-04 18:47:52 +01:00
Gael Guennebaud
b0ffd9bf04 clean #defined tokens, and use clock_gettime for the real time 2010-03-03 09:41:29 +01:00
Eamon Nerbonne
ff6b94d6d0 BenchTimer: avoid warning about symbol redefinition on win32, and include <Eigen/Core> (required to compile) 2010-03-02 08:46:11 +01:00
Gael Guennebaud
1710c07f63 remove Qt's atomic dependency, I don't know what I was doing wrong... 2010-03-01 13:09:47 +01:00
Gael Guennebaud
aeff3ff391 make Aron's idea work using Qt's atomic implementation for the synchronisation 2010-03-01 10:57:32 +01:00
Gael Guennebaud
ac425090f3 BTL: allow to bench real time 2010-02-26 14:57:49 +01:00
Gael Guennebaud
c05047d28e fix some BTL issues 2010-02-26 12:51:20 +01:00
Gael Guennebaud
3ac2b96a2f implement a smarter parallelization strategy for gemm avoiding multiple
paking of the same data
2010-02-26 12:32:00 +01:00
Gael Guennebaud
68eaefa5d4 update BTL (better timer, eigen2 => eigen3, etc) 2010-02-23 18:23:12 +01:00
Gael Guennebaud
3beedba244 merge 2010-02-22 21:32:29 +01:00
Thomas Capricelli
d3b314569b provide default values for CXX, remove duplicate define 2010-02-22 15:39:17 +01:00