i-robot
b6a6dbbe8f
!27003 [Auto parallel] [Sharding propagation] Dealing with a parameter being used by mutiple ops
Merge pull request !27003 from Xiaoda/114-adapting-sharding-propagation-for-common-parameter
4 years ago
i-robot
04381ab336
!26971 fault_recover_by_mirror_group_fix_opt_shard1
Merge pull request !26971 from yao_yf/fault_recover_by_mirror_group_fix_opt_shard1
4 years ago
i-robot
58fc9a6d81
!26985 fix the negative axis problem of reduce_eliminate
Merge pull request !26985 from huangbingjian/reduce_eliminate
4 years ago
Xiaoda Zhang
364858cbc9
In sharding propagation, to keep strategy consistent of parameter being used by multiple operators, we check the edge with one node of TmpIdentityInfo
4 years ago
i-robot
4b089a9661
!19466 Fix some scope error
Merge pull request !19466 from huanghui/fix-clone-same-scope
4 years ago
huangbingjian
66309632c1
fix the negative axis problem of reduce_eliminate
4 years ago
yao_yf
5dc756adc9
fix fault recover judgement in optimizer shard
4 years ago
i-robot
53804eec0a
!26776 recompute state slice optimize
Merge pull request !26776 from yao_yf/add_transformer_slice_activation_config_opt
4 years ago
i-robot
51dbbda8f2
!26935 [Auto parallel] Recover 'auto_parallel_search_mode' in parallel_context
Merge pull request !26935 from Xiaoda/115-adapting-interface-of-search-mode
4 years ago
i-robot
e195131342
!26891 move key world 'trace' to interface DumpSourceLines
Merge pull request !26891 from liubuyu/SBB
4 years ago
i-robot
b9d56b9cce
!26914 fix some error report and remove some useless code
Merge pull request !26914 from lianliguang/master
4 years ago
lby
618d4b7dfe
move key 'trace' to interface DumpSourceLines
4 years ago
i-robot
8bdcc68bb7
!26918 [AutoParallel]pipeline_adape_interleaved
Merge pull request !26918 from lichen/pipeline_adapt_interleaved
4 years ago
yao_yf
59934198aa
recompute output slice optimize
4 years ago
Xiaoda Zhang
04db51a528
In a previous PR ( https://gitee.com/mindspore/mindspore/pulls/26807/ ), we replaced 'auto_parallel_search_mode' by 'search_mode' directly.
However, to be forward compatitable, it is suitable to keep 'auto_parallel_search_mode' available. This PR recovers the 'auto_parallel_search_mode' interface and adds a warning when using this old interface.
This PR also deals with other codestyle things.
4 years ago
lichenever
8207242ebf
pipeline_adapt_interleaved
4 years ago
lianliguang
0dc5d7f4f9
modify some eroor log and remove some useless code
4 years ago
i-robot
42d0054bcd
!26756 [AutoPar] [D-Rec] Adding cost graph edges for Rec search in order to support the initialization for Reshape
Merge pull request !26756 from daniel_hg/Gather_version_update
4 years ago
huanghui
c4d61629b5
fix wrong scope and debug info
4 years ago
i-robot
4a82477626
!26794 transformer_slice_activation_config_fix
Merge pull request !26794 from yao_yf/add_transformer_slice_activation_config_fix
4 years ago
yao_yf
facb0995f4
slice recompute state fix
4 years ago
i-robot
04132e0c50
!26766 fix error log and move some function to inner
Merge pull request !26766 from lianliguang/master
4 years ago
i-robot
9f8ec2c5ab
!26807 [Auto parallel] [Sharding propagation] Interface change of sharding propagation
Merge pull request !26807 from Xiaoda/113-auto-parallel-search-mode-changes-to-search-mode
4 years ago
i-robot
6ecbc97fd6
!26804 virtual_dataset_avoid_auto_parallel
Merge pull request !26804 from yao_yf/virtual_dataset_avoid_auto_parallel
4 years ago
lianliguang
5864d6a006
modify some eroor log
4 years ago
i-robot
ed79ba11af
!26475 Support fused node corresponding code print
Merge pull request !26475 from huanghui/fused-debug-infos
4 years ago
Xiaoda Zhang
ad5ac77ae8
1) 'auto_parallel_search_mode' changes to 'search_mode';
2) 'sharding_propagation' moves to 'search_mode';
4 years ago
yao_yf
f29ce1fb60
virtual dataset avoid auto parallel
4 years ago
i-robot
1c1795f1cd
!26761 [Auto parallel] [Sharding propagation] Avoid <= in std::sort
Merge pull request !26761 from Xiaoda/110-fix-some-bugs-in-sharding-propagation
4 years ago
i-robot
7dcdc7d0fb
!26400 [Fallback] add fallback testcases
Merge pull request !26400 from huangbingjian/eval_numpy
4 years ago
huanghui
767caad833
Support fused node corresponding code print
4 years ago
i-robot
3d0f9d8aae
!26683 Enable compile cache feature to load hyper parameter data from python
Merge pull request !26683 from LiangZhibo/mindir
4 years ago
i-robot
dfd71f668b
!26556 modify parallel api note for master
Merge pull request !26556 from lilei/modify_parallel_API_note
4 years ago
Xiaoda Zhang
6d7eaea884
1) fix the int64_t and size_t mixup problem; 2) avoid the <= in std::sort
4 years ago
haoran.wang
1bc9cad97d
Add Cost Graph Edges for RecSearch
4 years ago
lilei
f8827a09fb
modify parallel API note for master
4 years ago
huangbingjian
00f3fa99e2
add fallback testcases
4 years ago
i-robot
b472850a75
!26594 Replace std::unordered_map/set with robin-hood-hashing
Merge pull request !26594 from hewei/use_robin_hood
4 years ago
i-robot
519f14a909
!26006 slice recompute activation
Merge pull request !26006 from yao_yf/add_transformer_slice_activation_config
4 years ago
l00591931
21df240f23
Enable mindir to load initialize weight from python
4 years ago
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.
4 years ago
i-robot
46e53a51c9
!26570 [Auto-par][d-rec] Change Onehot OP type to increase partitioning quality
Merge pull request !26570 from petitquentin/Gather_version_update
4 years ago
i-robot
20be757f18
!26430 replace short-circuit eval with deferred evaluation of backward prop function.
Merge pull request !26430 from xychow/replace-shortcurit-eval-with-lazy-eval
4 years ago
i-robot
0d69920358
!26600 [Auto parallel] Adjusting sharding propagation
Merge pull request !26600 from Xiaoda/105-adjusting-sharding-propagation
4 years ago
i-robot
edb540d6a4
!26624 [Auto parallel] Adjusting codestyle in step_auto_parallel.cc on master
Merge pull request !26624 from Xiaoda/108-adjusting-ApplyApproximationForParaNode-r1.5
4 years ago
i-robot
050925bc80
!26572 virtual_dataset_fix_repeat_right
Merge pull request !26572 from yao_yf/virtual_dataset_fix_repeat_right
4 years ago
i-robot
6d4837fe93
!26383 reshape_dont_replace_shape_when_no_other_redistribution
Merge pull request !26383 from yao_yf/reshape_dont_replace_shape_when_no_other_redistribution
4 years ago
haoran.wang
2127c6411e
Modify OneHot as eliminated op in D-Rec
4 years ago
Xiaoda Zhang
9a2c92bfea
adjusting ApplyApproximationForParaNode's input as const-reference
4 years ago
zhousiyi
3d8435ad45
Replace shortcircuit eval with deferred GetResult from AsyncAbstract when use
4 years ago