Philippe Tillet
|
d62e581ab3
|
basic split-k across warps working for GEMM
|
2019-08-05 19:33:28 -07:00 |
|
Philippe Tillet
|
d9945692a9
|
[dnn] better specification of recompilation key
|
2019-08-02 17:42:48 -07:00 |
|
Philippe Tillet
|
5af7e5adac
|
Made sure it works for FP16
|
2019-07-30 20:02:16 -07:00 |
|
Philippe Tillet
|
17cb2db356
|
[dnn/blocksparse/dot] prototype version seems to pass basic test
|
2019-07-27 21:21:36 -07:00 |
|
Philippe Tillet
|
b1d81a5802
|
more work on heuristics
|
2019-07-21 18:11:54 -07:00 |
|
Philippe Tillet
|
bfa39b8992
|
preparing the field for tensor cores transposes
|
2019-07-17 13:20:33 -07:00 |
|
Philippe Tillet
|
b7986baffa
|
[dnn]: Now implementing all existing DNN routines using common base template and auto-tuner
|
2019-07-09 19:52:55 -07:00 |
|
Philippe Tillet
|
6300ec5080
|
[examples] added conv2d op in tensorflow
|
2019-06-26 18:50:53 -07:00 |
|
Philippe Tillet
|
f1a8972267
|
[examples] added tensorflow dense convolution templates
|
2019-06-26 11:39:22 -07:00 |
|
Philippe Tillet
|
25e9a10917
|
changed auto-tuner parameter ranges
|
2019-06-25 19:27:49 -07:00 |
|
Philippe Tillet
|
d945ce5e1b
|
Now showing valid parameter for NN
|
2019-06-25 19:18:43 -07:00 |
|
Philippe Tillet
|
616f22c610
|
confirmed this is the fastest bounds checking
|
2019-06-25 16:35:43 -07:00 |
|
Philippe Tillet
|
64513fb407
|
[codegen] added fallback when tensor cores cannot be used
|
2019-06-25 15:49:58 -07:00 |
|
Philippe Tillet
|
d52abc9379
|
[codegen] bugfix in alignment inference
|
2019-06-25 15:06:15 -07:00 |
|
Philippe Tillet
|
edc31cabb0
|
[codegen] rough template for axis_info pass
|
2019-06-24 18:57:32 -07:00 |
|
Philippe Tillet
|
72867d17d4
|
more cleaning
|
2019-06-24 12:37:13 -07:00 |
|
Philippe Tillet
|
f7dcea1187
|
Now doing double-buffering
|
2019-06-13 19:48:02 -07:00 |
|
Philippe Tillet
|
36e3667a9a
|
removed shared conflicts for 8x32x4 and 32x8x4 configurations
|
2019-06-13 17:51:54 -07:00 |
|
Philippe Tillet
|
21a9b92c87
|
disabling interleaving
|
2019-06-13 17:16:00 -07:00 |
|
Philippe Tillet
|
d487cf31ce
|
trying 128 bits loads
|
2019-06-12 21:07:01 -07:00 |
|
Philippe Tillet
|
a6b580ec05
|
interleaving fails with B
|
2019-06-12 19:46:43 -07:00 |
|
Philippe Tillet
|
1b5a742a88
|
[triton/codegen] added shared memory padding for HMMA arguments and vectorized loads
|
2019-06-11 19:51:08 -07:00 |
|
Philippe Tillet
|
06b5992509
|
[feature] added basic tensor core support
|
2019-06-11 10:24:49 -07:00 |
|
Philippe Tillet
|
d074a166e2
|
[feature] basic tensor core utilization works
|
2019-06-08 14:39:45 -07:00 |
|
Philippe Tillet
|
ec4c6aaaaa
|
Added inline PTX for mma.sync
|
2019-06-07 19:39:33 -07:00 |
|
Philippe Tillet
|
781b6d377d
|
seleciton now segfault (expected
|
2019-06-06 20:34:56 -07:00 |
|
Philippe Tillet
|
6045209d5b
|
Now find correct tuning configuration
|
2019-06-06 20:13:26 -07:00 |
|
Philippe Tillet
|
cdf5a0d011
|
[codegen/tune]: added fragmentation types
|
2019-06-06 16:48:32 -07:00 |
|
Philippe Tillet
|
f58c9a4d2b
|
[general] hmma baseline setup
|
2019-06-05 14:43:38 -07:00 |
|
Philippe Tillet
|
49fcfd6fc7
|
[examples/tensorflow] fixed #include issue
|
2019-06-05 11:09:41 -07:00 |
|
Philippe Tillet
|
70f49a56c1
|
[examples/python/tensorflow] better skeleton for blocksparse
|
2019-05-01 17:09:01 -04:00 |
|