p47068915
/
mindspore
forked from mindspore-Ecosystem/mindspore
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

!22321 Update save_graph_paths in context 

Merge pull request !22321 from huanghui/code_docs_save_graphs
This commit is contained in:
i-robot 2021-08-24 13:49:41 +00:00 committed by Gitee
parent 3d5d3defc9 d3ddca3b35
commit a27e714c41
1 changed files with 3 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
mindspore/context.py
View File

@ -566,11 +566,9 @@ def set_context(**kwargs):
save_graphs (bool): Whether to save graphs. Default: False.
save_graphs_path (str): Path to save graphs. Default: ".".
If the program is executed in the parallel mode, `save_graphs_path` should consist of the path and the
current device id, to ensure that writing file conflicts won't happen when the different processes try to
create the files in the same directory. For example, the `device_id` can be generated by
`device_id = os.getenv("DEVICE_ID")` and the `save_graphs_path` can be set by
`context.set_context(save_graphs_path="path/to/ir/files"+device_id)`.
Graphs will be saved to the directory of `save_graphs_path/rank_${rank_id}/ir_dump`.
When distributed training, `rank_id` is the ID of the current device in the cluster.
Otherwise, `rank_id` is `0`.
enable_graph_kernel (bool): Whether to enable graph kernel fusion to optimize network execution performance.
Default: False.
graph_kernel_flags (str): Optimization options of graph kernel fusion. Experienced user only.