Skip to content

Commit e72b5d3

Browse files
author
Guang Yang
committed
Update doc to include Benchmark Dashboard
1 parent aa852cc commit e72b5d3

File tree

1 file changed

+20
-30
lines changed

1 file changed

+20
-30
lines changed

extension/benchmark/README.md

Lines changed: 20 additions & 30 deletions
Original file line numberDiff line numberDiff line change
@@ -8,7 +8,7 @@ The ExecuTorch project introduces an advanced benchmarking infrastructure design
88

99
- **Device Support**: Includes popular phones like latest Apple iPhone, Google Pixel, and Samsung Galaxy, etc.
1010

11-
- **Backend Delegates**: Supports XNNPACK, Apple CoreML, Qualcomm QNN, and more in the near future.
11+
- **Backend Delegates**: Supports XNNPACK, Apple CoreML and MPS, Qualcomm QNN, and more in the near future.
1212

1313
- **Benchmark Apps:** Generic apps that support both GenAI and non-GenAI models, capable of measuring performance offline. [Android App](android/benchmark/) | [iOS App](apple/Benchmark/). Popular Android and iOS profilers with in-depth performance analysis will be integrated with these apps in the future.
1414

@@ -17,35 +17,25 @@ The ExecuTorch project introduces an advanced benchmarking infrastructure design
1717
> **Disclaimer:** The infrastructure is new and experimental. We're working on improving its accessibility and stability over time.
1818
1919

20-
## Preliminary Benchmark Results
21-
22-
Below is a table summarizing some example data points obtained via the infra. These numbers represent model load time and average inference latency across different platforms and backends.
23-
24-
| Model | Backend | Model Load Time (ms) | Avg Inference Latency (ms) | Device |
25-
|-----------------------|--------------|----------------------|----------------------------|------------------------|
26-
| MobileNetV3 (mv3) | XNNPACK Q8 | [34.024](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218988461) | [252.628](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218988461) | Samsung S22 |
27-
| MobileNetV3 (mv3) | QNN FP16 | [168.414](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218987785) | [1.182](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218987785) | Samsung S22 |
28-
| MobileNetV3 (mv3) | COREML FP16 | [57.372](https://github.com/pytorch/executorch/actions/runs/11136241814/job/30999930691) | [0.429](https://github.com/pytorch/executorch/actions/runs/11136241814/job/30999930691) | Apple iPhone 15 Pro |
29-
| MobileNetV2 (mv2) | XNNPACK Q8 | [14.397](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218987379) | [10.796](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218987379) | Samsung S22 |
30-
| MobileNetV2 (mv2) | QNN FP16 | [136.862](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218987097) | [0.673](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218987097) | Samsung S22 |
31-
| MobileNetV2 (mv2) | COREML FP16 | [50.892](https://github.com/pytorch/executorch/actions/runs/11171117103/job/31056078594) | [0.631](https://github.com/pytorch/executorch/actions/runs/11171117103/job/31056078594) | Apple iPhone 15 Pro |
32-
| InceptionV4 (ic4) | XNNPACK Q8 | [87.617](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218986716) | [117.937](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218986716) | Samsung S22 |
33-
| InceptionV4 (ic4) | QNN FP16 | [163.943](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218986381) | [2.734](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218986381) | Samsung S22 |
34-
| InceptionV4 (ic4) | COREML FP16 | [118.686](https://github.com/pytorch/executorch/actions/runs/11136241814/job/30999930140) | [4.289](https://github.com/pytorch/executorch/actions/runs/11136241814/job/30999930140) | Apple iPhone 15 Pro |
35-
| InceptionV3 (ic3) | XNNPACK Q8 | [60.708](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218986023) | [98.390](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218986023) | Samsung S22 |
36-
| InceptionV3 (ic3) | QNN FP16 | [134.732](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218985425) | [1.351](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218985425) | Samsung S22 |
37-
| InceptionV3 (ic3) | COREML FP16 | [86.728](https://github.com/pytorch/executorch/actions/runs/11171117103/job/31056078753) | [1.391](https://github.com/pytorch/executorch/actions/runs/11171117103/job/31056078753) | Apple iPhone 15 Pro |
38-
| DeepLabV3 (dl3) | XNNPACK Q8 | [90.616](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218985758) | [666.219](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218985758) | Samsung S22 |
39-
| DeepLabV3 (dl3) | QNN FP16 | [182.207](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218985141) | [9.759](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218985141) | Samsung S22 |
40-
| ResNet50 (resnet50) | XNNPACK Q8 | [55.462](https://github.com/pytorch/executorch/actions/runs/10911725781/job/30285857102) | [37.021](https://github.com/pytorch/executorch/actions/runs/10911725781/job/30285857102) | Apple iPhone 15 Pro |
41-
| ResNet50 (resnet50) | COREML FP16 | [68.943](https://github.com/pytorch/executorch/actions/runs/11136241814/job/30999930818) | [1.979](https://github.com/pytorch/executorch/actions/runs/11136241814/job/30999930818) | Apple iPhone 15 Pro |
42-
| TorchVisionVit (vit) | QNN FP16 | [174.430](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218989581) | [199.279](https://github.com/pytorch/executorch/actions/runs/10875550238/job/30218989581) | Samsung S22 |
43-
| Wave2Letter (w2l) | XNNPACK Q8 | [33.913](https://github.com/pytorch/executorch/actions/runs/10857890364/job/30139445319) | [135.584](https://github.com/pytorch/executorch/actions/runs/10857890364/job/30139445319) | Apple iPhone 15 Pro |
44-
| Wave2Letter (w2l) | COREML FP16 | [109.254](https://github.com/pytorch/executorch/actions/runs/11136241814/job/30999931566) | [28.465](https://github.com/pytorch/executorch/actions/runs/11136241814/job/30999931566) | Apple iPhone 15 |
45-
| MobileBERT (mobilebert) | XNNPACK FP32 | [26.499](https://github.com/pytorch/executorch/actions/runs/11136241814/job/30999930558) | [33.978](https://github.com/pytorch/executorch/actions/runs/11136241814/job/30999930558) | Apple iPhone 15 Pro |
46-
| MobileBERT (mobilebert) | COREML FP16 | [206.202](https://github.com/pytorch/executorch/actions/runs/11136241814/job/30999930398) | [1.873](https://github.com/pytorch/executorch/actions/runs/11136241814/job/30999930398) | Apple iPhone 15 Pro |
47-
| EDSR (edsr) | XNNPACK Q8 | [3.190](https://github.com/pytorch/executorch/actions/runs/11136241814/job/30999929836) | [168.429](https://github.com/pytorch/executorch/actions/runs/11136241814/job/30999929836) | Apple iPhone 15 Pro |
48-
| EDSR (edsr) | COREML FP16 | [156.075](https://github.com/pytorch/executorch/actions/runs/11136241814/job/30999929690) | [77.346](https://github.com/pytorch/executorch/actions/runs/11136241814/job/30999929690) | Apple iPhone 15 Pro |
20+
## Dashboard
21+
22+
The ExecuTorch Benchmark Dashboard tracks performance metrics for various models across different backend delegates and devices. It enables users to compare metrics, monitor trends, and identify optimizations or regressions in Executorch. The dashboard is accessible at **[ExecuTorch Benchmark Dashboard](https://hud.pytorch.org/benchmark/llms?repoName=pytorch%2Fexecutorch)**.
23+
24+
**Comprehensive Comparisons**:
25+
- Analyze performance differences between backend delegates (e.g., XNNPACK, CoreML, QNN, MPS) for the same model.
26+
- Compare performance across different models.
27+
- Track performance changes over time and across different commits.
28+
29+
**Metric Tracking**:
30+
- Monitor essential metrics such as load time and inference time for different model implementations.
31+
- For LLMs, additional metrics like tokens/s are available.
32+
- Observe performance trends over time to identify improvements or regressions.
33+
34+
**Visualizations**:
35+
- View detailed performance data through charts and graphs.
36+
- Color-coded highlights for improvements (green) and regressions (red) exceeding 5% compared to the baseline.
37+
38+
The dashboard is refreshed nightly with new models, metrics, and features to provide the most current and relevant information for ExecuTorch performance benchmarking across various model types and configurations.
4939

5040

5141
## Supported Use Cases

0 commit comments

Comments
 (0)