Commit Graph

138 Commits

Author SHA1 Message Date
zhoufeng ecae690a19 Revert "fix same node is used by two comm op"
This reverts commit b09d411dc4.

add AlltoAll and NeighborExchange as interface
2021-08-30 16:54:38 +08:00
Zhang Qinghua a137fa1d0b Optimize the Executors routines.
- Fix the key generating.
- Distinguish the executors.
2021-08-28 17:36:12 +08:00
i-robot 32281f84e7 !19000 update LayerNormGrad split pass to V2
Merge pull request !19000 from yuchaojie/ir_fusion2
2021-08-26 08:03:33 +00:00
yuchaojie 1d1490df0b xupdate LayerNormGrad split pass to V2 2021-08-24 14:43:09 +08:00
He Wei c9ecb27db8 Change monad.py as internal usage only 2021-08-24 10:45:40 +08:00
ms_yan 36a8886ca2 Revert "[feat] [assistant] [I3T96T] add new Dataset operator CMUARCTICDataset"
This reverts commit b077aa1cab.

Revert "[feat] [assistant] [I3T96X] add new Dataset operator LibriSpeechDataset"

This reverts commit 4e6f7dc97d.

delete pass_registry_test.cc

comment  hiai_nlu_model_multi.pb related  line
2021-08-23 01:46:38 +08:00
djc 4e6f7dc97d [feat] [assistant] [I3T96X] add new Dataset operator LibriSpeechDataset 2021-08-22 13:39:37 +08:00
zhoufeng 03a56f2bb0 alltoall exception handle
Signed-off-by: zhoufeng <zhoufeng54@huawei.com>
2021-08-11 17:30:49 +08:00
hwjiaorui e498c96a20 tbe json creator UT 2021-08-02 18:51:33 +08:00
zhoufeng 2143241092 change neighbor exchange to all to all
Signed-off-by: zhoufeng <zhoufeng54@huawei.com>
2021-08-01 18:21:28 +08:00
laiyongqiang 1533435015 replace memcpy_async with tensor move 2021-04-15 16:12:01 +08:00
buxue d830a2b55a optimize exception mode when use undefined name in if for and while 2021-03-27 17:28:52 +08:00
dingpeifei 87e41aaeee IR operators of GPU and CPU are unified as batchnorm 2021-03-18 19:02:28 +08:00
chenfei fc335daa30 get all j node and then expand them 2021-03-12 20:06:43 +08:00
l00591931 bbdb050fc7 Change switch to Switch 2021-03-11 17:06:55 +08:00
l00591931 680324f225 Change make tuple in core.ops 2021-03-05 10:54:33 +08:00
yuchaojie 6d195f340c add SyncBatchNorm 2021-02-27 14:32:54 +08:00
He Wei 7d9a783993 [auto-monad] Support side-effects by auto-monad
The basic idea is: exploits data dependency to control the execution order
of side-effect operations, and keep the semantics of ANF unchanged.

The ControlDepend primitive is removed and there are two primitives added:

1. UpdateState:
```
  a = Assign(para, value)
```
became:
```
  a = Assign(para, value, u)
  u = UpdateState(u, a)
```

2. Load:
```
  x = Add(para, value)
```
became:
```
  p = Load(para, u)
  x = Add(p, value)
  u = UpdateState(u, p)
```
2021-02-08 09:01:15 +08:00
mindspore-ci-bot aebe263dce !11895 unify mindir for different backend: the output num of optimizer ops, the backward of concat
From: @wangnan39
Reviewed-by: 
Signed-off-by:
2021-02-05 16:27:19 +08:00
jinyaohui 8022f9a6ed modify pack to stack 2021-02-04 18:54:31 +08:00
wangnan39@huawei.com cd9173fdfd unify the output num of optimizer ops 2021-02-04 17:46:00 +08:00
mindspore-ci-bot b189f177bb Change tuple_getitem to TupleGetItem and some other ops, merge from r1.1 to master 2021-02-02 15:54:15 +08:00
l00591931 9ec100d069 Change TensorAdd to Add, from r1.1 to master 2021-02-01 17:53:52 +08:00
yuchaojie b51b3a6764 update Pool's attr kernel_size, pad_mode 2021-01-18 12:10:20 +08:00
lilei 9a45c4419c modify batch_normal 2021-01-13 20:59:46 +08:00
chenhaozhe b3add83bf0 support const input in graph_ir convertor, add value inference in Concat 2020-11-19 20:34:49 +08:00
Hoai Linh Tran 46f07efc31 Fix AdjustAllReduceMulAdd pass
Revive tests in lib_test

Code cleaning
2020-09-19 01:07:01 -04:00
yujianfeng 4b77f6b53c Add AdamApplyOneWithDecayAssign fusion pass 2020-08-29 17:54:15 +08:00
huanghui b8d7f6d77f add UnsortedSegmentSum fission pass 2020-08-12 19:17:19 +08:00
yujianfeng 8a77751988 Add AdamApplyOneAssign and AdamApplyOneWithDecayAssign fusion pass 2020-08-10 14:46:37 +08:00
panyifeng 34e50e5d6e fix cell bprop 2020-08-07 10:57:01 +08:00
huanghui 30000fdb52 add ReduceMin fission pass 2020-08-05 10:46:01 +08:00
laiyongqiang 2458431750 not eliminate memcpy when nexe node is graph output 2020-07-30 19:51:57 +08:00
huanghui f1563d2d37 insert memcpy async if hccl op cascade 2020-07-22 10:43:38 +08:00
mindspore-ci-bot 6f8863b65d !3198 synchronize latest Ascend software suite 18 Jul 2020, and merging branches
Merge pull request !3198 from yanghaoran/code_sync_0718
2020-07-19 14:14:45 +08:00
yanghaoran 859acc6d2a synchronize latest Ascend software suite 18 Jul 2020, and merging branches 2020-07-18 21:09:55 +08:00
yujianfeng fa0684d12d Add pack and concat fission pass 2020-07-18 15:39:06 +08:00
mindspore-ci-bot 4a19e6b8cb !3114 add coo_tensor
Merge pull request !3114 from riemann_penn/coo_tensor
2020-07-18 10:24:58 +08:00
panyifeng 5a10383cc3 add coo_tensor 2020-07-17 18:53:03 +08:00
yujianfeng 188d74f15e Remove transdata and cast for internal outputs 2020-07-17 16:55:47 +08:00
changzherui f4cb445ea8 syn code for 0715 2020-07-16 17:18:21 +08:00
mindspore-ci-bot a661545d49 !265 Synchronize Ascend software suite 07 Jul 2020
Merge pull request !265 from yanghaoran/incubator
2020-07-15 21:04:59 +08:00
jiangjinsheng 7f9bbfd338 add Conv1d ops 2020-07-14 18:59:45 +08:00
mindspore-ci-bot 69ce4bf2b2 !2970 Del mindspore/model_zoo
Merge pull request !2970 from panbingao/master
2020-07-14 08:57:31 +08:00
panbingao 45dbc8bf04 Move model_zoo.resnet.py 2020-07-10 17:36:34 +08:00
huanghui 3eaf663545 add tensor scatter update fission pass 2020-07-10 15:41:22 +08:00
yujianfeng 24f6b9d77e Add input2output pass 2020-07-09 21:44:43 +08:00
zhousiyi b29d260be3 reuse AddN primitive in opt as AddN will replicated by program_specialize 2020-07-03 03:18:25 +00:00
panyifeng d6635bbbe2 Add IndexedSlices 2020-06-30 17:38:24 +08:00
yujianfeng 7ad877a948 Add Split fission pass 2020-06-20 14:28:25 +08:00