docs: document interrupt & HIL #1218
Triggered via pull request
December 11, 2024 02:53
eyurtsev
Status
Success
Total duration
46m 43s
Artifacts
–
Annotations
1 warning and 2 notices
|
benchmark
ubuntu-latest pipelines will use ubuntu-24.04 soon. For more details, see https://github.com/actions/runner-images/issues/10636
|
|
Benchmark results:
docs/docs/concepts/breakpoints.md#L1
.........................................
fanout_to_subgraph_10x: Mean +- std dev: 61.0 ms +- 1.5 ms
.........................................
fanout_to_subgraph_10x_sync: Mean +- std dev: 51.7 ms +- 0.7 ms
.........................................
fanout_to_subgraph_10x_checkpoint: Mean +- std dev: 74.5 ms +- 1.4 ms
.........................................
fanout_to_subgraph_10x_checkpoint_sync: Mean +- std dev: 94.0 ms +- 2.3 ms
.........................................
fanout_to_subgraph_100x: Mean +- std dev: 594 ms +- 29 ms
.........................................
WARNING: the benchmark result may be unstable
* the standard deviation (70.8 ms) is 13% of the mean (560 ms)
Try to rerun the benchmark with more runs, values and/or loops.
Run 'python -m pyperf system tune' command to reduce the system jitter.
Use pyperf stats, pyperf dump and pyperf hist to analyze results.
Use --quiet option to hide these warnings.
fanout_to_subgraph_100x_sync: Mean +- std dev: 560 ms +- 71 ms
.........................................
WARNING: the benchmark result may be unstable
* the standard deviation (167 ms) is 17% of the mean (990 ms)
Try to rerun the benchmark with more runs, values and/or loops.
Run 'python -m pyperf system tune' command to reduce the system jitter.
Use pyperf stats, pyperf dump and pyperf hist to analyze results.
Use --quiet option to hide these warnings.
fanout_to_subgraph_100x_checkpoint: Mean +- std dev: 990 ms +- 167 ms
.........................................
fanout_to_subgraph_100x_checkpoint_sync: Mean +- std dev: 939 ms +- 18 ms
.........................................
react_agent_10x: Mean +- std dev: 30.6 ms +- 0.6 ms
.........................................
react_agent_10x_sync: Mean +- std dev: 22.5 ms +- 0.4 ms
.........................................
WARNING: the benchmark result may be unstable
* the maximum (73.6 ms) is 93% greater than the mean (38.2 ms)
Try to rerun the benchmark with more runs, values and/or loops.
Run 'python -m pyperf system tune' command to reduce the system jitter.
Use pyperf stats, pyperf dump and pyperf hist to analyze results.
Use --quiet option to hide these warnings.
react_agent_10x_checkpoint: Mean +- std dev: 38.2 ms +- 3.5 ms
.........................................
WARNING: the benchmark result may be unstable
* the maximum (56.1 ms) is 53% greater than the mean (36.8 ms)
Try to rerun the benchmark with more runs, values and/or loops.
Run 'python -m pyperf system tune' command to reduce the system jitter.
Use pyperf stats, pyperf dump and pyperf hist to analyze results.
Use --quiet option to hide these warnings.
react_agent_10x_checkpoint_sync: Mean +- std dev: 36.8 ms +- 1.8 ms
.........................................
react_agent_100x: Mean +- std dev: 339 ms +- 6 ms
.........................................
react_agent_100x_sync: Mean +- std dev: 275 ms +- 15 ms
.........................................
react_agent_100x_checkpoint: Mean +- std dev: 846 ms +- 19 ms
.........................................
react_agent_100x_checkpoint_sync: Mean +- std dev: 838 ms +- 8 ms
.........................................
wide_state_25x300: Mean +- std dev: 22.9 ms +- 0.4 ms
.........................................
wide_state_25x300_sync: Mean +- std dev: 14.7 ms +- 0.1 ms
.........................................
wide_state_25x300_checkpoint: Mean +- std dev: 283 ms +- 12 ms
.........................................
wide_state_25x300_checkpoint_sync: Mean +- std dev: 275 ms +- 14 ms
.........................................
wide_state_15x600: Mean +- std dev: 27.0 ms +- 0.7 ms
.........................................
WARNING: the benchmark result may be unstable
* the maximum (33.7 ms) is 94% greater than the mean (17.4 ms)
Try to rerun the benchmark with more runs, values and/or loops.
Run 'python -m pyperf system tune' command to reduce the system jitter.
Use pyperf stats, pyperf dump and pyperf hist to analyze results.
Use --quiet option to hide these warnings.
wide_state_15x600_sync: Mean +- std dev: 17.4 ms +- 1.7 ms
.............................
|
|
Comparison against main:
docs/docs/concepts/breakpoints.md#L1
+-----------------------------------------+---------+-----------------------+
| Benchmark | main | changes |
+=========================================+=========+=======================+
| fanout_to_subgraph_100x | 689 ms | 594 ms: 1.16x faster |
+-----------------------------------------+---------+-----------------------+
| fanout_to_subgraph_10x_checkpoint_sync | 100 ms | 94.0 ms: 1.07x faster |
+-----------------------------------------+---------+-----------------------+
| react_agent_100x_checkpoint | 900 ms | 846 ms: 1.06x faster |
+-----------------------------------------+---------+-----------------------+
| react_agent_100x_checkpoint_sync | 890 ms | 838 ms: 1.06x faster |
+-----------------------------------------+---------+-----------------------+
| fanout_to_subgraph_100x_checkpoint_sync | 989 ms | 939 ms: 1.05x faster |
+-----------------------------------------+---------+-----------------------+
| wide_state_25x300_checkpoint | 297 ms | 283 ms: 1.05x faster |
+-----------------------------------------+---------+-----------------------+
| wide_state_9x1200_checkpoint | 332 ms | 316 ms: 1.05x faster |
+-----------------------------------------+---------+-----------------------+
| react_agent_10x_sync | 23.6 ms | 22.5 ms: 1.05x faster |
+-----------------------------------------+---------+-----------------------+
| react_agent_10x_checkpoint_sync | 38.5 ms | 36.8 ms: 1.05x faster |
+-----------------------------------------+---------+-----------------------+
| react_agent_10x | 31.9 ms | 30.6 ms: 1.05x faster |
+-----------------------------------------+---------+-----------------------+
| wide_state_15x600_checkpoint | 509 ms | 487 ms: 1.04x faster |
+-----------------------------------------+---------+-----------------------+
| wide_state_9x1200_checkpoint_sync | 319 ms | 306 ms: 1.04x faster |
+-----------------------------------------+---------+-----------------------+
| wide_state_15x600_checkpoint_sync | 490 ms | 472 ms: 1.04x faster |
+-----------------------------------------+---------+-----------------------+
| wide_state_25x300_checkpoint_sync | 285 ms | 275 ms: 1.03x faster |
+-----------------------------------------+---------+-----------------------+
| react_agent_100x | 349 ms | 339 ms: 1.03x faster |
+-----------------------------------------+---------+-----------------------+
| wide_state_9x1200 | 27.4 ms | 26.6 ms: 1.03x faster |
+-----------------------------------------+---------+-----------------------+
| wide_state_9x1200_sync | 17.6 ms | 17.1 ms: 1.03x faster |
+-----------------------------------------+---------+-----------------------+
| react_agent_100x_sync | 282 ms | 275 ms: 1.03x faster |
+-----------------------------------------+---------+-----------------------+
| fanout_to_subgraph_10x_sync | 52.9 ms | 51.7 ms: 1.02x faster |
+-----------------------------------------+---------+-----------------------+
| fanout_to_subgraph_10x_checkpoint | 76.3 ms | 74.5 ms: 1.02x faster |
+-----------------------------------------+---------+-----------------------+
| wide_state_25x300_sync | 15.1 ms | 14.7 ms: 1.02x faster |
+-----------------------------------------+---------+-----------------------+
| wide_state_15x600 | 27.6 ms | 27.0 ms: 1.02x faster |
+-----------------------------------------+---------+-----------------------+
| react_agent_10x_checkpoint | 39.0 ms | 38.2 ms: 1.02x faster |
+-----------------------------------------+---------+-----------------------+
| wide_state_25x300 | 23.4 ms | 22.9 ms: 1.02x faster |
+-----------------------------------------+---------+-----------------------+
| fanout_to_subgraph_10x | 62.0 ms | 61.0 ms: 1.02x faster |
+---------------------------------------
|