[Yolo Benchmark] Core Dumped when AGX Orin set to 30W power mode

Hi all,

I hope I am in the right place to ask for some help.

I am doing some standard benchmark from Yolo offical site running in the Jetson AGX Orin 64GB.

My problem is where I set the power mode to 30W (nvpmodel -m 2) of the Jetson AGX Orin 64GB, I got a “CORE DUMPED” error although in power mode 50W and MAXN (60W), I do not have it and all is working fine.

I did not touch anything in the Yolo Benchmark neither modify something on the Jetson.

I am executing from docker.

Here are the logs:

root@1febb2b472ab:/ultralytics# yolo benchmark model=yolov8m.pt data='coco8.yaml' imgsz=640 format=onnx
Creating new Ultralytics Settings v0.0.6 file ✅
View Ultralytics Settings with 'yolo settings' or at '/root/.config/Ultralytics/settings.json'
Update Settings with 'yolo settings key=value', i.e. 'yolo settings runs_dir=path/to/dir'. For help see https://docs.ultralytics.com/quickstart/#ultralytics-settings.
Downloading https://github.com/ultralytics/assets/releases/download/v8.3.0/yolov8m.pt to 'yolov8m.pt': 100% ━━━━━━━━━━━━ 49.7MB 2.8MB/s 17.7s
YOLOv8m summary (fused): 92 layers, 25,886,080 parameters, 0 gradients, 78.9 GFLOPs

PyTorch: starting from 'yolov8m.pt' with input shape (1, 3, 640, 640) BCHW and output shape(s) (1, 84, 8400) (49.7 MB)
requirements: Ultralytics requirement ['onnxslim>=0.1.71'] not found, attempting AutoUpdate...
Collecting onnxslim>=0.1.71
  Downloading onnxslim-0.1.78-py3-none-any.whl.metadata (10 kB)
Requirement already satisfied: onnx in /usr/local/lib/python3.10/dist-packages (from onnxslim>=0.1.71) (1.19.1)
Requirement already satisfied: sympy>=1.13.1 in /usr/local/lib/python3.10/dist-packages (from onnxslim>=0.1.71) (1.13.1)
Requirement already satisfied: packaging in /usr/local/lib/python3.10/dist-packages (from onnxslim>=0.1.71) (20.9)
Collecting colorama (from onnxslim>=0.1.71)
  Downloading colorama-0.4.6-py2.py3-none-any.whl.metadata (17 kB)
Requirement already satisfied: ml_dtypes in /usr/local/lib/python3.10/dist-packages (from onnxslim>=0.1.71) (0.5.4)
Requirement already satisfied: mpmath<1.4,>=1.1.0 in /usr/local/lib/python3.10/dist-packages (from sympy>=1.13.1->onnxslim>=0.1.71) (1.3.0)
Requirement already satisfied: numpy>=1.21 in /usr/local/lib/python3.10/dist-packages (from ml_dtypes->onnxslim>=0.1.71) (1.26.4)
Requirement already satisfied: protobuf>=4.25.1 in /usr/local/lib/python3.10/dist-packages (from onnx->onnxslim>=0.1.71) (5.29.5)
Requirement already satisfied: typing_extensions>=4.7.1 in /usr/local/lib/python3.10/dist-packages (from onnx->onnxslim>=0.1.71) (4.15.0)
Requirement already satisfied: pyparsing>=2.0.2 in /usr/local/lib/python3.10/dist-packages (from packaging->onnxslim>=0.1.71) (3.2.5)
Downloading onnxslim-0.1.78-py3-none-any.whl (166 kB)
Downloading colorama-0.4.6-py2.py3-none-any.whl (25 kB)
Installing collected packages: colorama, onnxslim

Successfully installed colorama-0.4.6 onnxslim-0.1.78
WARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager, possibly rendering your system unusable. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv. Use the --root-user-action option if you know what you are doing and want to suppress this warning.

requirements: AutoUpdate success ✅ 3.5s
WARNING ⚠️ requirements: Restart runtime or rerun command for updates to take effect


ONNX: starting export with onnx 1.19.1 opset 17...
ONNX: slimming with onnxslim 0.1.78...
2025-12-08 13:14:39.070489513 [E:onnxruntime:Default, env.cc:234 ThreadMain] pthread_setaffinity_np failed for thread: 33, index: 0, mask: {10, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.
2025-12-08 13:14:39.070534505 [E:onnxruntime:Default, env.cc:234 ThreadMain] pthread_setaffinity_np failed for thread: 34, index: 1, mask: {9, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.
2025-12-08 13:14:39.078332323 [E:onnxruntime:Default, env.cc:234 ThreadMain] pthread_setaffinity_np failed for thread: 35, index: 2, mask: {8, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.
ONNX: export success ✅ 10.1s, saved as 'yolov8m.onnx' (99.0 MB)

Export complete (11.7s)
Results saved to /ultralytics
Predict:         yolo predict task=detect model=yolov8m.onnx imgsz=640
Validate:        yolo val task=detect model=yolov8m.onnx imgsz=640 data=coco.yaml
Visualize:       https://netron.app
Loading yolov8m.onnx for ONNX Runtime inference...
Using ONNX Runtime 1.20.0 ('CUDAExecutionProvider', {'device_id': 0})
2025-12-08 13:14:40.004287719 [E:onnxruntime:Default, env.cc:234 ThreadMain] pthread_setaffinity_np failed for thread: 40, index: 0, mask: {10, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.
2025-12-08 13:14:40.004343654 [E:onnxruntime:Default, env.cc:234 ThreadMain] pthread_setaffinity_np failed for thread: 42, index: 2, mask: {8, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.
2025-12-08 13:14:40.004346566 [E:onnxruntime:Default, env.cc:234 ThreadMain] pthread_setaffinity_np failed for thread: 41, index: 1, mask: {9, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.
Loading yolov8m.onnx for ONNX Runtime inference...
Using ONNX Runtime 1.20.0 ('CUDAExecutionProvider', {'device_id': 0})
2025-12-08 13:14:42.695232870 [E:onnxruntime:Default, env.cc:234 ThreadMain] pthread_setaffinity_np failed for thread: 48, index: 0, mask: {10, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.
2025-12-08 13:14:42.695763711 [E:onnxruntime:Default, env.cc:234 ThreadMain] pthread_setaffinity_np failed for thread: 50, index: 2, mask: {8, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.
2025-12-08 13:14:42.695795807 [E:onnxruntime:Default, env.cc:234 ThreadMain] pthread_setaffinity_np failed for thread: 49, index: 1, mask: {9, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.
Setting batch=1 input of shape (1, 3, 640, 640)

WARNING ⚠️ Dataset 'coco8.yaml' images not found, missing path '/datasets/coco8/images/val'
Downloading https://ultralytics.com/assets/coco8.zip to '/datasets/coco8.zip': 100% ━━━━━━━━━━━━ 432.8KB 1.4MB/s 0.3s
Unzipping /datasets/coco8.zip to /datasets/coco8...: 100% ━━━━━━━━━━━━ 25/25 1.3Kfiles/s 0.0s
Dataset download success ✅ (1.1s), saved to /datasets

val: Fast image access ✅ (ping: 0.0±0.0 ms, read: 884.5±425.8 MB/s, size: 54.0 KB)
val: Scanning /datasets/coco8/labels/val... 4 images, 0 backgrounds, 0 corrupt: 100% ━━━━━━━━━━━━ 4/4 80.3it/s 0.0s
val: New cache created: /datasets/coco8/labels/val.cache
                 Class     Images  Instances      Box(P          R      mAP50  mAP50-95): 100% ━━━━━━━━━━━━ 4/4 2.3it/s 1.8s
                   all          4         17      0.902      0.776       0.94      0.728
Speed: 2.2ms preprocess, 94.7ms inference, 0.0ms loss, 4.5ms postprocess per image
Setup complete ✅ (8 CPUs, 61.4 GB RAM, 215.5/1759.4 GB disk)

Benchmarks complete for yolov8m.pt on coco8.yaml at imgsz=640 (20.48s)
Benchmarks legend:  - ✅ Success  - ❎ Export passed but validation failed  - ❌️ Export failed
+------------------------------------------------------------------------------------------+
|     Format   Status❔   Size (MB)   metrics/mAP50-95(B)   Inference time (ms/im)   FPS   |
+==========================================================================================+
| 1   ONNX     ✅         99.0        0.7285                94.71                    10.56 |
+------------------------------------------------------------------------------------------+

💡 Learn more at https://docs.ultralytics.com/modes/benchmark
free(): invalid pointer
Aborted (core dumped)

Hi,

This is a known issue from ONNXRuntime:

The error happens if there are some CPU statuses set to off.
Thanks.

Hi @AastaLLL ,

Thanks :)

So until the bug is fixed, is there any other benchmarks to be run on 15W and 30W ?

Hi,

You can check below link for the benchmark sample:

For YOLO, please try Ultralytics example, which deploys the model with TensorRT:
https://www.jetson-ai-lab.com/tutorial_ultralytics.html

Thanks.

Many thanks ! :)