wassname/ray
1
0
Fork 0
mirror of https://github.com/wassname/ray.git synced 2026-07-20 12:40:20 +08:00
Code Issues Packages Projects Releases Wiki Activity

[rllib][asv] Support ASV for RLlib (#2304)

Browse Source
This commit is contained in:
Richard Liaw
2018-06-28 17:20:09 -07:00
committed by GitHub
parent 92ab7e56ec
commit 3cc27d2840
5 changed files with 283 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/README-benchmarks.rst
+12 -3
View File
@@ -8,22 +8,31 @@ You can run the benchmark suite by doing the following:
To run ASV inside docker, you can use the following command:
``docker run --rm --shm-size=10G --memory=10G $DOCKER_SHA bash -c '/ray/test/jenkins_tests/run_asv.sh'``
``docker run --rm --shm-size=10G --memory=10G $DOCKER_SHA bash -c '/ray/test/jenkins_tests/run_rllib_asv.sh'``
Visualizing Benchmarks
======================
To visualize benchmarks, you must copy the S3 bucket down to `$RAY_DIR/python`. Assuming asv is installed,
For visualizing regular Ray benchmarks, you must copy the S3 bucket down to `$RAY_DIR/python`.
.. code-block::
cd $RAY_DIR/python
aws s3 sync s3://$BUCKET/ASV/ .
Then, you can run:
For rllib, you must sync a _particular_ folder down to `$RLLIB_DIR (ray/python/ray/rllib)`.
.. code-block::
cd $RAY_DIR/python/ray/rllib
aws s3 sync s3://$BUCKET/RLLIB_RESULTS/ ./RLLIB_RESULTS
Then, in the directory, you can run:
.. code-block::
asv publish --no-pull
asv preview
This creates the directory and then launches a server.
This creates the directory and then launches a server at which you can visualize results.