OSSForks/pytorch
1
0
Fork 0
mirror of https://github.com/zebrajr/pytorch.git synced 2025-12-06 12:20:52 +01:00
Code Issues Actions 38 Packages Projects Releases Wiki Activity
200 Commits 8,304 Branches 150 Tags 1.5 GiB
809d54ee50
Commit Graph

200 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Yangqing Jia
809d54ee50 convnet benchmark minor change 2016-01-05 09:55:22 -08:00
Yangqing Jia
8c1bbaa2ab some fill ops that are not tested. 2016-01-05 09:55:22 -08:00
Yangqing Jia
6cb2072422 cudnn conv op backward compatibility back to v2 2016-01-05 09:55:21 -08:00
Yangqing Jia
778a1f6956 speed benchmark 2016-01-05 09:55:21 -08:00
Yangqing Jia
05eda208a5 Last commit for the day. With all the previous changes this should give an exact reference speed that TensorFlow with CuDNN3 should achieve in the end. 2016-01-05 09:55:21 -08:00
Yangqing Jia
896e8e5274 pooling backward cudnn, and constant for kOne and kZero. 2016-01-05 09:55:21 -08:00
Yangqing Jia
f8585bbf62 cudnn pool op. 2016-01-05 09:55:21 -08:00
Yangqing Jia
664bdf83d7 Pooling refactor so we can do a proper cudnn benchmark. 2016-01-05 09:55:21 -08:00
Yangqing Jia
288f350899 math_gpu.cu bugfix 2016-01-05 09:55:21 -08:00
Yangqing Jia
ebd6c9fab8 muji bugfix with ngpu=4 2016-01-05 09:55:21 -08:00
Yangqing Jia
55cced894d Some untested half float stuff for benchmarking. 2016-01-05 09:49:55 -08:00
Yangqing Jia
8d4683434b convnet benchmark: make it consistent with TF's model. 2015-12-17 11:25:51 -08:00
Yangqing Jia
b7c3b48469 copy matrix can be done with cudamemcpy. 2015-12-17 10:22:02 -08:00
Yangqing Jia
b10ee24fc3 conv op: backward exhaustive mode too. This does not seem to help much, suggesting that cudaGetConvolution*Algo is already doing a very good job. Verified with googlenet. 2015-12-17 10:21:16 -08:00
Yangqing Jia
d79cfb4ae7 exhaustive search for cudnn 2015-12-15 22:21:11 -08:00
Yangqing Jia
61c114971b fast path for copymatrix 2015-12-15 22:21:11 -08:00
Yangqing Jia
05e3207e26 fast path for copymatrix 2015-12-15 21:25:53 -08:00
Yangqing Jia
cc9323793e add relu cudnn code 2015-12-15 20:43:34 -08:00
Yangqing Jia
4f2530d8ce expose benchmark code to python 2015-12-15 20:42:54 -08:00
Yangqing Jia
6b27cabf17 net benchmark code 2015-12-15 20:42:22 -08:00
Yangqing Jia
cf8ffe215f minor tuning 2015-12-15 20:41:58 -08:00
Yangqing Jia
20ccca5b67 RTTI to true in default for the main model. 2015-12-15 11:01:09 -08:00
Yangqing Jia
f714ad0a70 number of blocks now makes more sense. 2015-12-15 10:46:50 -08:00
Yangqing Jia
3b0cc79465 context gpu: better error catching 2015-12-14 13:59:28 -08:00
Yangqing Jia
73f3daf736 minor bugfix for workspace 2015-12-13 08:37:36 -08:00
Yangqing Jia
bfae070de1 minor bugfix for net 2015-12-13 08:37:01 -08:00
Yangqing Jia
359f7685f8 halfway into timing test. 2015-12-11 11:01:40 -08:00
Yangqing Jia
03c777db72 boolean for has_gpu_support 2015-12-10 15:06:57 -08:00
Yangqing Jia
7bdc8a6c19 Pycaffe2: removed the clunky gpu support hack. 
Now, when one builds pycaffe2, if cuda is present, we will always build
pycaffe2 with gpu support.
 2015-12-10 15:06:57 -08:00
Yangqing Jia
becf9e85c1 remove no longer needed build_env_android.py. 2015-12-10 15:06:57 -08:00
Yangqing Jia
82696ebc5d Merge pull request #9 from Yangqing/master 
a script to test zeromq db throughput.
 2015-12-09 15:36:39 -08:00
Yangqing Jia
ae1ebd0f19 a script to test zeromq db throughput. 2015-12-09 15:15:06 -08:00
Yangqing Jia
77541ffe14 flags relaxation, or tightening? 2015-12-07 20:48:57 -08:00
Yangqing Jia
ceb4cde74a average pooling format change to fit the cudnn interface 2015-12-06 15:56:29 -08:00
Yangqing Jia
6bfb30047e deprecate legacy pooling 2015-12-06 11:28:00 -08:00
Yangqing Jia
20dbbbbb28 android: use full proto in default 2015-12-06 11:26:30 -08:00
Yangqing Jia
9022e4f499 pull protobuf to master 2015-12-05 18:34:48 -08:00
Yangqing Jia
05465783c6 optionally use protobuf lite 2015-12-05 16:15:00 -08:00
Yangqing Jia
3d7cb201a3 misc changes to reduce binary size. 2015-12-04 21:31:23 -08:00
Yangqing Jia
4eb486bd34 misc update to reduce binary size. Removed zmq.hpp 2015-12-03 21:28:55 -08:00
Yangqing Jia
ff04fe8b1b merge 2015-12-02 21:41:56 -08:00
Yangqing Jia
1a4ea7c8fc misc updates 2015-12-02 21:01:55 -08:00
Yangqing Jia
b64429bbc6 Merge branch 'dev' of https://github.com/Yangqing/caffe2 into dev 
Conflicts:
	caffe2/operators/spatial_batch_norm_op_cudnn.cc
 2015-12-02 20:57:36 -08:00
Yangqing Jia
25647f8c47 more test for tf benchmark purposes. 2015-12-02 16:55:51 -08:00
Yangqing Jia
01b45fd052 backward support to cudnn R2 for TensorFlow benchmark references 2015-12-02 15:12:04 -08:00
Yangqing Jia
acc16645d3 temp hack. Will rewrite the build script later. 2015-12-02 10:06:15 -08:00
Yangqing Jia
3a4d4285f2 Added more benchmarks. 2015-12-02 10:04:00 -08:00
Yangqing Jia
7d87fe788f alexnet benchmark code using cudnn: this should give a reference speed that TensorFlow should achieve after tuning. With R4 currently we have 29.5ms fwd / 93.4ms bwd. 2015-12-01 17:17:22 -08:00
Yangqing Jia
1499b87e56 cudnn spatial bn: optional compilation instead of throwing error 2015-12-01 14:20:28 -08:00
Yangqing Jia
5ba54180f5 various updates 2015-11-28 13:12:43 -08:00