Commit Graph

166 Commits

Author SHA1 Message Date
chendongsheng ce4d86171f support https 2021-04-06 15:37:59 +08:00
lvchangquan 0a7df321fe add op atomic clean to clear input addr in launch allreduce 2021-03-25 11:19:23 +08:00
yepei6 0cb6adce9e add ENABLE_GLIBCXX param 2021-03-22 20:40:01 +08:00
lvchangquan 31f9e6a42c add op_mul fusion based on allreduce fusion in pynative mode 2021-03-15 21:12:14 +08:00
caifubi 171b468bb3 PyNative AllReduce Bucket 2021-03-03 15:47:18 +08:00
chendongsheng db0a6f1e19 replace ps-lite 2021-03-02 11:19:11 +08:00
chendongsheng febbbf45c8 added server 2021-02-27 15:49:18 +08:00
mindspore-ci-bot 7c32bbe972 !12516 added worker
From: @anancds
Reviewed-by: 
Signed-off-by:
2021-02-27 14:22:39 +08:00
chendongsheng 6c22dc0d55 added worker 2021-02-26 21:10:16 +08:00
LianLiguang 2fa4997674 remove all of cpp infer reg 2021-02-26 15:19:26 +08:00
Zhenglong Li e2d3495925 Dvpp refactor and 6 Dvpp operators 2021-02-10 10:34:38 +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
jinyaohui d9be0c102d add some ops 2021-02-03 19:47:53 +08:00
yanghaitao1 8d147deb07 profiler memory 2021-01-19 10:21:43 +08:00
mindspore-ci-bot ca3ac703c9 !10627 handler signal
From: @zhupuxu
Reviewed-by: 
Signed-off-by:
2020-12-29 15:47:30 +08:00
zhupuxu e36175b0dc handler signal
Signed-off-by: zhupuxu <zhupuxu@huawei.com>
2020-12-29 09:39:13 +08:00
LianLiguang cb909c7242 add ut for conv2d 2020-12-26 17:04:37 +08:00
yanghaoran b8345d03b6 Synchronize latest Ascend software 18 Dec 2020, with profiler fixes 2020-12-24 15:11:01 +08:00
xuyongfei c9b072b94d remove serving in mindspore repo 2020-12-10 18:50:30 +08:00
limingqi107 eca951c7bc add ps cache manager 2020-12-09 19:48:00 +08:00
limingqi107 a5f57ce8a0 add ps cache 2020-12-08 20:34:51 +08:00
kswang cbece74a23 add cpu adam delta 2020-11-06 13:52:54 +08:00
zhoufeng 183742009f mindspore cxx api for 310 inference
Signed-off-by: zhoufeng <zhoufeng54@huawei.com>
2020-11-04 13:59:33 +08:00
anancds 28c9e9d9a5 Added http server based on libevent 2020-10-16 14:26:42 +08:00
ZPaC 0b49f0fb57 change PS dir 2020-09-18 10:48:09 +08:00
caifubi 372c2e7951 Combine Async Dump and E2E Dump 2020-09-14 10:57:46 +08:00
xulei2020 b356211413 add init code 2020-09-10 15:35:08 +08:00
dayschan 37a48f6aac GraphKernel supports GPU
1. Update akg submodule
2. Refactor akg_kernel_build, akg_ascend_kernel_build, akg_gpu_kernel_build
3. Add akg_kernel_json_decoder to support converting kernel_json to AnfNode.
4. Add GraphKernel Cost Model. (mindspore/_extends/graph_kernel)
5. Add some GraphKernel passes to GpuSession, move these passes to backend/optimizer/graph_kernel.
6. Add global id for ir files.
7. Fix bug in ConstInputToAttr.
2020-09-09 17:46:50 +08:00
lizhenyu c3d6918649 add kernel select after optimize pass 2020-09-04 10:52:51 +08:00
zhoufeng 6d500223b2 utest compiling performance optimization
Signed-off-by: zhoufeng <zhoufeng54@huawei.com>
2020-09-03 16:57:19 +08:00
zhousiyi 301aeaaf1d remove gvar source file from compile as ut_tests will use gvar so 2020-09-03 06:49:30 +00:00
kswang 756bb6d53f async run graph 2020-08-28 18:18:47 +08:00
mindspore-ci-bot 2db4bda219 !4622 serving restful init
Merge pull request !4622 from hexia/http_init
2020-08-19 16:15:58 +08:00
hexia 75d116b5db serving http init 2020-08-18 20:15:47 +08:00
zhoufeng 663278112f optimize code compile performance
Signed-off-by: zhoufeng <zhoufeng54@huawei.com>
2020-08-18 14:33:24 +08:00
huangdongrun 2a6d346d2f support if by if grad parameter
add join for ref

adjust env eliminate to eliminate all env ops

add partial app cache

resolve while endless

fix env eliminate

support for "for while" cases

fix join shape error
2020-08-15 09:47:10 +08:00
huanghui 8f70af7300 add unique_with_pad cpu kernel 2020-08-13 21:55:00 +08:00
huanghui fa081092c8 add Unique cpu kernel 2020-08-11 21:36:55 +08:00
mindspore-ci-bot c6acafcbc6 !4268 fix glog level dont work
Merge pull request !4268 from liubuyu/master
2020-08-11 20:05:03 +08:00
liubuyu 25a67602b7 bug fix for glog level don't work 2020-08-11 16:44:50 +08:00
WilliamLian 1f5441d73a remove inferfunction to core 2020-08-11 14:22:29 +08:00
liubuyu 195f04e50e compile the core directory into a separate static library 2020-08-08 21:31:41 +08:00
mindspore-ci-bot 98c415a1d5 !3807 serving support acl dvpp process
Merge pull request !3807 from 徐永飞/master
2020-08-05 11:00:21 +08:00
xuyongfei c0b8df4f7c serving add acl dvpp support, ut tests, return input status string message 2020-08-04 21:40:21 +08:00
Zhang Qinghua 960da5cbed Decouple the backend TBE from binding Python API. 2020-08-01 17:09:08 +08:00
nhussain b65125b0a1 add more ifdef and include statments for pybind 2020-07-27 13:51:11 -04:00
liubuyu f4bc0bc9fe move the dependency of utils to core 2020-07-27 10:28:46 +08:00
Zhang Qinghua 18efdc8fe2 Move abstract/base into core, and remove static_analysis including symbols in ir. 2020-07-15 14:04:38 +08:00
liubuyu 43c79eb853 mindspore path adjust 2020-07-14 18:07:28 +08:00
Zhang Qinghua 87715ed127 Refactoring for base and abstract types: Moving Base into base folder; Splitting the abstract type and infer&specialize routines; 2020-07-14 09:38:00 +08:00
mindspore-ci-bot 65c6e54e2e !2773 Adding cpp_api
Merge pull request !2773 from 章一智/cpp_api
2020-07-11 18:58:42 +08:00
ZPaC 8273c2a39c Add ps optimizer info builder. 2020-07-11 16:07:09 +08:00
ZPaC 272e397be7 Add ps optimizer info. 2020-07-11 11:31:58 +08:00
ervinzhang bd5a777f81 introducing new C++ API 2020-07-10 16:32:38 -04:00
ZPaC 4645a43e08 Add ps module in batches 2020-07-09 15:45:56 +08:00
yujianfeng 91306cbd34 Add ut for cpu kernel of sparse optimizer 2020-06-28 21:12:08 +08:00
Shida He 4c056855e0 Implementation for mindspore debugger 2020-06-22 17:15:30 -04:00
gong chen a6dfa281ea Init GraphKernel.
- It provides a unified style to express graph and kernel for user.
- It provides a unified IR to represent graph and kernel for developer.
- It breaks the boundary between graph and kernel.
- It provides more opportunities to do compile optimization.
2020-06-20 22:31:54 +08:00
hangangqiang dfbb232468 mindspore server inference 2020-06-20 14:12:56 +08:00
jiangjinsheng 709704d339 fixed CMakeLists.txt 2020-06-10 10:58:26 +08:00
zhoufeng af5019b94f link child graphs
Signed-off-by: zhoufeng <zhoufeng54@huawei.com>
2020-05-22 15:37:19 +08:00
dengwentao fa83fc3e02 rename kernel/mng to kernel/rts 2020-05-13 16:26:17 +08:00
kswang bef62db128 add ascend mem pool 2020-04-09 16:04:59 +08:00
kswang fb343bd607 add mem manager 2020-04-09 10:05:45 +08:00
Jonathan Yan 9d0fde29f4 remove ENABLE_MINDRECORD flag 2020-04-07 14:55:45 -04:00
zhunaipan 930a1fb0a8 initial version
Signed-off-by: leonwanghui <leon.wanghui@huawei.com>
2020-03-27 22:54:54 +08:00