gaspersic/triton
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
546 Commits 14 Branches 9 Tags
3293c45e60332eea57de81e3303acec864c689eb
Commit Graph

99 Commits

Author SHA1 Message Date
Philippe Tillet
3293c45e60 GEMM: Enabled use of cuBLAS when predicted beneficial 2016-10-04 21:17:17 -04:00
Philippe Tillet
0158ce9978 README: Added AMD Fury benchmark 2016-09-30 02:24:15 -04:00
Philippe Tillet
b514638d86 Bench: re-order GEMM-bench order (Putting DeepBench on top as it's most relevant) 2016-09-30 01:21:41 -04:00
Philippe Tillet
5d0e29db1f Bench: Fixed CUDA synchronization issue 2016-09-30 01:21:24 -04:00
Philippe Tillet
f474fc03b9 Bench: style improvement 2016-09-29 22:27:38 -04:00
Philippe Tillet
ac2f0cab86 Bench: Major cleaning. Now displaying best result in bold. 2016-09-29 17:31:50 -04:00
Philippe Tillet
fa4cb6866d Bench: Now displaying results in a table 2016-09-29 14:50:42 -04:00
Philippe Tillet
97a0d65a4d Code quality: reorganized files structure 2016-04-10 13:13:16 -04:00
Philippe Tillet
7f77fba4d4 General: Internal code generator overhaul 2016-04-02 18:19:33 -04:00
Philippe Tillet
05af3ccc10 Bench: clearer benchmark names 2015-12-21 18:42:56 -05:00
Philippe Tillet
db4e286b9e Bench: Now using host timer for CUDA benchmarks 2015-12-18 14:51:11 -05:00
Philippe Tillet
373771c796 Tuner: more polishing of intermediate BLAS3 sizes 2015-12-18 03:38:01 -05:00
Philippe Tillet
f99293816c Bench: now using host timer rather than events 2015-12-17 01:37:42 -05:00
Philippe Tillet
f975ea7621 GEMM: reverted AMD optimizations 2015-12-12 01:29:08 -05:00
Philippe Tillet
b3c5251f91 CMake: Fixed clBLAS handling 2015-12-12 01:29:08 -05:00
Philippe Tillet
7140f065c2 Benchmark: rectified sizes for benchmarks to fit convnet-benchmarks 2015-11-28 21:49:55 -05:00
Philippe Tillet
386963a6cc Core: added queue-wise temporary workspace. WARNING: breaks the fused computation of multiple DOT/GEMV operations 2015-11-27 18:43:46 -05:00
Philippe Tillet
c6333c993a API: adding cuBLAS interface 2015-11-20 12:46:42 -05:00
Philippe Tillet
da1b0a9571 GEMM: performance regression fix 2015-11-19 20:49:38 -05:00
Philippe Tillet
1e076c131b API: clearer interface for transposition 2015-10-01 21:58:59 -04:00
Philippe Tillet
feeb1e9862 Feature: Merged kernel-fusion branch 
* Fuses multiple AXPY kernel
* Possibility to add thread-wise for loops in AXPY-like kernels
 2015-09-30 15:31:41 -04:00
Philippe Tillet
149441b9e2 Bench: improved output formatting 2015-08-31 13:35:29 -04:00
Philippe Tillet
836a955663 GEMV: bugfix with CUDA 2015-08-30 02:35:55 -04:00
Philippe Tillet
8dcf062342 Benchmarks: added consistency between CUDA and the rest 2015-08-27 22:55:20 -04:00
Philippe Tillet
6676b94d00 Bench: no longer reallocating memory for CUDA. 2015-08-27 19:09:22 -04:00
Philippe Tillet
f06a3bdf53 Bugfix: fixed bug in dynamic kernel selection 2015-08-26 19:11:09 -04:00
Philippe Tillet
ffb3c01b77 Code quality: fixed typo 2015-08-26 14:24:12 -04:00
Philippe Tillet
69c11d16cc Code quality: bugfix in bench/test to note call clBLAS on CUDA backend 2015-08-26 14:12:50 -04:00
Philippe Tillet
10524ebdee CUDA: various improvements 2015-08-24 17:03:31 -04:00
Philippe Tillet
33dac6b05a Code quality: fixed compilation errors with CUDA 2015-08-20 21:24:41 -04:00
Philippe Tillet
efdbf5f4a6 Bench: Added LeNet sizes 2015-08-18 16:44:35 -07:00
Philippe Tillet
470ae7f4be Benchmarks: More human-readable labels 2015-08-17 16:32:10 -07:00
Philippe Tillet
99e2592c34 Benchmarks: Fixed MKL setup 2015-08-13 17:20:41 -07:00
Philippe Tillet
71224a1507 Code quality: renamed isaac shortcut from isc to sc 2015-08-12 19:38:53 -07:00
Philippe Tillet
ff4cf94df7 Code quality: significant cleaning of namespaces, etc.. 2015-08-12 00:47:58 -07:00
Philippe Tillet
f60b82af25 Kernels: more generic temporary workspace checks 2015-08-10 10:19:50 -07:00
Philippe Tillet
2dc351f932 Benchmarks: removed useless linebreaks after some benchmarks 2015-08-10 09:38:13 -07:00
Philippe Tillet
89b0873c24 Benchmarks: now benchmarking the GEMV sizes of AlexNet 2015-08-10 09:37:44 -07:00
Philippe Tillet
0f0946b7a7 CMake: Now converting JSon to source with EOG / Warnings fix on GCC 2015-08-06 17:54:04 -07:00
Philippe Tillet
db090d7942 Code quality: Large clean-up of the codebase and especially of the include/ folder 2015-08-06 12:05:12 -07:00
Philippe Tillet
ebab87af61 Code quality: everything now compiles without warnings on MSVC 2015-08-05 13:01:20 -07:00
Philippe Tillet
ef14e3aa17 Code quality: more MSVD warnings removal 2015-08-05 12:47:20 -07:00
Philippe Tillet
54c4a31af0 Code quality: more warning fixes 2015-08-05 12:07:51 -07:00
Philippe Tillet
f4c597b294 Code quality: fixed compilation errors/warnings with Clang 2015-08-05 09:26:50 -07:00
Philippe Tillet
5b58769fbc Code quality: some more cleaning 2015-08-03 22:50:49 -07:00
Philippe Tillet
1565363722 Driver: reverted lookup map key (for queues/program) to objects rather than pointers 2015-08-03 17:37:19 -07:00
Philippe Tillet
92d8afc888 Code quality: cleaned backend terminology 2015-08-03 17:11:24 -07:00
Philippe Tillet
dd5f592fd8 Code quality: destructors of static global objects will no longer call external dependencies. Manual cleaning possible through backend::release() 
This should solve errors when libOpenCL's ICD was unloaded prior to libisaac.
 2015-07-31 15:47:41 -07:00
Philippe Tillet
dfe3d26f5e Misc: Fixed compilation issues on tests/bench 2015-07-31 01:02:59 -07:00
Philippe Tillet
29bdf7f546 Code quality: made the backend static 2015-07-30 21:15:37 -07:00