wangshuide2020
a542f004ef
1. add the limitation of the number of tag in tensor visualization; 2. update the max step per tensor tag to 20; 3. support query one train_job in the interface of train_jobs.
5 years ago
Li Hongzhang
3da4d71dff
add the resource monitor api
- collect_cpu
- collect_mem
- collect_npu
5 years ago
wangshuide2020
7877f33b70
Use multiple processes to calc events.
1. To accelerate summary file parsing, multiple processes are used. As the first step to mindinsight parsing performance optimization, we only made changes to _load_single_file function.
2. This PR will imporve summary parsing throughput dramatically (about cpu_count times)
3. Changes are mainly about _load_single_file function
In the future, a more global concurrent computing framework is needed for mindinsight. See the gitee wiki doc for details.
5 years ago
wangshuide2020
e8ffeb70ef
Support tensor visualization. 1.Tensor display in a table, it can support no more than two dimensions tensor visualization; 2.Tensor histogram visualization for all step in cache.
5 years ago
zhangyunshu
5871887c0b
profiler: fixed error msg in timeline rest api
5 years ago
mindspore-ci-bot
449929adac
!370 update the display name of aicore detail execution time
Merge pull request !370 from chenchao99/profiler_analyser
5 years ago
mindspore-ci-bot
e0f08cf394
!374 Profiler: fixed the issue of using unverified dir path
Merge pull request !374 from zhangyunshu/zys_timeline_fixed_fuzz
5 years ago
chenchao99
0d8808e5b3
update the display name of aicore detail execution time
5 years ago
zhangyunshu
a5c474c211
profiler: fixed timeline fuzz issue
5 years ago
askmiao
21eda46fa3
use the custom response
5 years ago
mindspore-ci-bot
6f953c0cfe
!282 add path_prefix argument for start command
Merge pull request !282 from liangyongxiong/path-prefix
5 years ago
liangyongxiong
f2f80b177a
add path_prefix argument for start command
5 years ago
yuximiao
70857c7b5f
adapt profiling source path change
5 years ago
yelihua
15065c1095
enable to show the fp and bp point
5 years ago
yuximiao
98b5030b3b
add minddata proposer
5 years ago
mindspore-ci-bot
3956a90a39
!287 Change the method to find the step trace files.
Merge pull request !287 from yelihua/dev_profiler
5 years ago
yelihua
82ed137d4f
change the method to find step trace files
5 years ago
zhangyunshu
50b4b0b97a
Add new feature for Profiler:
Collect timeline information after training finished.
Support timeline display on web UI.
5 years ago
root
449bec1fa2
featrue for proposer
5 years ago
chenchao99
ee3f241205
add feature of minddata pipeline stage analysing function
5 years ago
mindspore-ci-bot
d200233c48
!265 Add Dataset profiling.
Merge pull request !265 from yuximiao/master
5 years ago
yuximiao
3732f1a2a8
add minddata profiling
5 years ago
yelihua
7abccc0724
add summary info for step trace graph and change the name.
5 years ago
yelihua
c52d050193
update ut, st and validation.
5 years ago
mindspore-ci-bot
b0eab91bed
!252 Implement step trace in profiler module
Merge pull request !252 from yelihua/dev_profiler
5 years ago
yelihua
cce05461ec
implementation of step trace profiler
5 years ago
ougongchang
3ac9b93ddf
add unsafe-eval CSP to application. The d3-graphviz 3.0 version is required to open unsafe-eval CSP.
5 years ago
liangyongxiong
b191a04a0c
add paramater validation for train job caches api
5 years ago
liangyongxiong
80ffdf06ca
check if parameter is missing for train job caches api
5 years ago
yuximiao
e5b7223ba5
add ut
5 years ago
yuximiao
ea8d64c100
:add profiler restful api
5 years ago
liangyongxiong
8893236417
compare scalars within multiple train jobs
5 years ago
liangyongxiong
4417c5d027
1) print workspace path on startup; 2) update gunicorn configuration
5 years ago
luopengting
6963af9374
add lineage cache, add update method and put api, modify ut/st
5 years ago
wenkai
99b40573b4
fix load data thread stuck in LineageCacheItemUpdater
5 years ago
mindspore-ci-bot
02ecf155f7
!106 refactor data manager to support unified cache and data access/reload layer
Merge pull request !106 from wenkai/wk02_cache_3cp1
5 years ago
wenkai
7e17d6ffae
refactor data manager and unify cache and data access/reload
5 years ago
ougongchang
f59d1798a5
1. Complete the optimization of the graph code
2. Optimize the python api of graph and graph_processor
3. delete unconnected const and parameter nodes
4. fix subnode count bug and fix name scope node have not input and output bug
5. add cyclic graph and high degree graph json file
6. update parameter node to polymeric scope node
7. delele all deepcopy to improve performance
8. add data type and update the attribute scope of edge to independtend_layout
9. add add_parameter_nodes function
10. delete scope attribute in input and optimiza code
11. parse parameter data type and shape
12. fix ut and st for graph
13. fix pylint
14. fix pylint arguments-differ
15. add scope proxy node instead of node
5 years ago
liangyongxiong
e3327a4b73
fix pylint warnings
5 years ago
mindspore-ci-bot
1f2f85844b
!44 start/stop command should return correct exit-code based on result of process
Merge pull request !44 from liangyongxiong/master
5 years ago
zhangyunshu
7b9823b179
support for histogram API
5 years ago
liangyongxiong
1bc356ad11
1) fix return code for start/stop command; 2) replace print with standard console logger
5 years ago
luopengting
2ddb2b9ce6
merge model_lineage and dataset_lineage, modify/add st and ut for lineage api
5 years ago
luopengting
455f1de7f6
unquote train id for scalar metadata api
5 years ago
ougongchang
e3c0bbac89
fix unsafe functions and duplication files and redundant codes
5 years ago
gaocongli
e7a0496e87
initial version
5 years ago