yankf21
/
mindspore2022
forked from huawei/mindspore2022
0
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
42,203 Commits 18 Branches 22 Tags 3.6 GiB
Commit Graph

60 Commits

Author SHA1 Message Date
zhoufeng b7bb53ff73 build libmindspore_backend.so 
Signed-off-by: zhoufeng <zhoufeng54@huawei.com>
 2022-03-17 19:14:45 +08:00
chenfei 3872e29c3b add ut for tuple trans 2022-03-16 21:33:54 +08:00
zhoufeng f49b195c39 extract common as an independent shared library 
Signed-off-by: zhoufeng <zhoufeng54@huawei.com>
 2022-02-26 10:55:51 +08:00
chenfei a96f0b11e0 open closure 2022-01-28 22:04:31 +08:00
zhousiyi d9aa48bc64 Dont liftfv before grad in higher order grad if closure is kept as bprop of Partial primitive is not supported 2021-12-28 07:54:22 +00:00
zhangzhaoju 722086435d 1. fix:unexpected eiminate of tile 
2. fix:delay memory allocat in NewTensorFilledWithData
 2021-12-06 14:49:48 +08:00
lianliguang 0dc5d7f4f9 modify some eroor log and remove some useless code 2021-11-29 10:23:03 +08:00
He Wei 41dcac9c49 Replace std::unordered_map/set with robin-hood-hashing 
Robin-hood-hashing (https://github.com/martinus/robin-hood-hashing)
is considered faster then std::unordered_map/set,
so we use it to improve mindspore performance.

1. robin_hood head file in `third_party/robin_hood/include`;
2. In `utils/hash_map.h` and `utils/hash_set.h`, we define:
 - mindspore::HashMap as an alias of robin_hood::unordered_map;
 - mindspore::HashSet as an alias of robin_hood::unordered_set;
3. Replace:
 - `#include <unordered_map>` --> `#include "utils/hash_map.h"`;
 - `#include <unordered_set>` --> `#include "utils/hash_set.h"`;
 - `std::unordered_map` --> `mindspore::HashMap`;
 - `std::unordered_set` --> `mindspore::HashSet`;
 - `map.insert(std::pair(key, value))` --> `map.emplace(key, value)`;
 - `[] (const std::pair<K, V> &p) {..} ` --> `[] (const auto &p) {..} `;
4. Fix issues found by switch to robin_hood:
 - AnfNodeConfig hash and equal;
 - Fix a bug in `Slice::operator==()`;
 - Fix a bug in `CNode::HasPrimalAttr()`;
 - Fix map.erase() usage bugs: `map.erase(iter++)` --> `iter = map.erase(iter)`;
 - Fix some iterator invalidated problem;
5. Some std::unordered_map/set can not replace by robin_hood:
 - As parameter of functions that exposed to python by pybind11;
 - Use bad hash that cause robin_hood::map over_flow, such as AbstractBasePtrListHasher;
6. Update cpp unit tests;
7. Add build option '-F' to enable robin_hood, default on.
 2021-11-24 10:47:40 +08:00
zhang__sss 0dd68aa7eb grad 2021-11-11 23:18:35 +08:00
zhang__sss 8bae4d9d50 grad 2021-11-08 14:12:58 +08:00
huanghui ba66c0d491 add security isolate for save_graphs 2021-09-14 10:24:12 +08:00
zhousiyi a5b4e7bbf8 lift fv before grad except weight, then convert 
switch(cond, partial(g1, xs), partial(g2, ys))(Zs)
 to
 switch(cond, g1, g2)(Xs, Ys, Zs)

 switch_layer(index, make_tuple(partial(g1, xs), partial(g2, ys)))(Zs)
 to
 switch_layer(index, make_tuple(g1, g2))(Xs, Ys, Zs)

put Zs at last when unifyparameter as it may have u-monad or io-monad

use joined args other than broadened one as some extra parameter which is not a parameter of while_header can be add to while_body

inline fprop_switch forcely

reorder the parameter if one of the parameter is Monad when incorporate call

incorporate switch tuple_getitem if item 0 of tuple is EnvInstance or
item 1 of tuple is bprop function

addn with shape() and shape(1)

remove context_ from FuncGraphEvaluator to make it re-entry able to resolve evaluator stuck issue because of re-entry of the same FuncGraphEvaluator
 2021-07-03 00:03:37 +08:00
l00591931 0424b2d8d6 Fix warning in master 2021-06-04 17:40:01 +08:00
lvliang 116781041a fix ut compile failed 2021-05-19 19:59:10 +08:00
zhousiyi 4ff19b7082 add API to build bprop funcgraph which can be used for higher grad 2021-05-19 19:59:06 +08:00
lvliang c759effa94 fix_compile_ut_failed 2021-05-19 19:59:06 +08:00
zhousiyi 1843740e0d 1. set abstract for node. 2. Adjust the position of sens parameter from last to position just after Inputs. 3. Set debug name for Parameter 2021-05-19 19:58:51 +08:00
zhousiyi b40816e66a add support for stop_gradient primitive. And defer BackPropagation to End API 2021-05-19 19:58:49 +08:00
zhousiyi 0d6a5ccfe4 grad op for pynative 2021-05-19 19:58:46 +08:00
huangbingjian 134d5dfe4b separate irpass EnvGetItemEliminater and ItemTupleOrListEliminator 2021-04-02 17:38:40 +08:00
chenfei fc335daa30 get all j node and then expand them 2021-03-12 20:06:43 +08:00
yujianfeng f0b7c3dd4d Add a switch to control grad for scalar 2021-02-27 18:12:37 +08:00
yujianfeng 41189781f3 support scalar input for cell 2021-02-25 09:56:15 +08:00
buxue 6ccc4379b4 do not broaden scalar 2021-02-03 11:30:25 +08:00
zhouyuanshen d1e6878011 remove dot 2021-02-01 09:39:27 +08:00
buxue 2530943a7f optimize list getitem in bprop 2020-12-15 15:00:49 +08:00
yujianfeng 3176d377e6 Remove redundant phi nodes 2020-11-16 09:36:55 +08:00
Yi Huaijie d7faa77b5e support int64 shape 2020-11-05 21:15:49 +08:00
Hoai Linh Tran 46f07efc31 Fix AdjustAllReduceMulAdd pass 
Revive tests in lib_test

Code cleaning
 2020-09-19 01:07:01 -04:00
fary86 fcbb3e0edc Refactor ms_context implementation 2020-08-30 22:51:40 +08:00
zhousiyi c25e37e7bf make backend/optimizer pybind free 2020-08-28 08:56:58 +00:00
panyifeng 34e50e5d6e fix cell bprop 2020-08-07 10:57:01 +08:00
kingfo 28dabf0332 fix grad flag update issue in pynative 2020-08-06 15:12:31 +08:00
liubuyu a499d4e47c decoupling core and debug 2020-07-31 10:39:31 +08:00
liubuyu f4bc0bc9fe move the dependency of utils to core 2020-07-27 10:28:46 +08:00
He Wei 60d168adec Decouple ir from optimizer 2020-07-21 14:48:26 +08:00
mindspore-ci-bot db80643ee2 !3177 Update Arithmetic Simplify to use Pattern Matcher (2) 
Merge pull request !3177 from Giancarlo/pm_arithmetic_simplify
 2020-07-20 10:02:06 +08:00
Giancarlo Colmenares cfbfaddfe1 Updated arithmetic simplify to use Pattern Matcher 2020-07-17 10:05:09 -04:00
panyifeng 5a10383cc3 add coo_tensor 2020-07-17 18:53:03 +08:00
liubuyu 43c79eb853 mindspore path adjust 2020-07-14 18:07:28 +08:00
BowenK af36bd7698 Revert PR 2923 2020-07-14 10:01:08 +08:00
Giancarlo Colmenares 7f759c2a4c Updated arithmetic simplify to use Pattern Matcher 2020-07-08 16:01:04 -04:00
panyifeng d6635bbbe2 Add IndexedSlices 2020-06-30 17:38:24 +08:00
Giancarlo Colmenares aabec55c79 Removing TransformFuncType 2020-06-24 13:48:57 -04:00
zhousiyi b2ec296f99 add opt pass for tuple_getitem with constant input 2020-06-23 07:41:21 +00:00
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
BowenK 8f29e7242f add pass to eliminate depend value 2020-06-19 14:52:52 +08:00
Kang 2974b906d3 Optimization for opt 2020-06-17 16:04:54 +08:00
mindspore-ci-bot 9e1dea1ff9 !1928 delete redundant reshape op 
Merge pull request !1928 from yoonlee666/deletereshape
 2020-06-10 20:11:22 +08:00
yoonlee666 881af56ec3 optimize redundant reshape op 2020-06-10 18:21:00 +08:00