Could not run 'prepacked::conv2d_clamp_prepack' with arguments from the 'CUDA' backend.

[FedeAi]

🐛 Bug
I'm using LibTorch API for c++. It seams that the torchscript model ( exported with device 0 ) calls some functions that are not implented for CUDA.

To Reproduce (REQUIRED)

Input:

module_ = torch::jit::load(base_path + torchscriptGpuPath_);
....
torch::Tensor imgTensor = torch::from_blob(cv_frame.data, {cv_frame.rows, cv_frame.cols, 3}, torch::kByte).to(device_);
imgTensor = imgTensor.permute({2, 0, 1});
imgTensor = imgTensor.toType(torch::kFloat);
imgTensor = imgTensor.div(255);
imgTensor = imgTensor.unsqueeze(0);

// Model feeding
torch::Tensor preds =
        module_.forward({imgTensor}).toTuple()->elements()[0].toTensor();

Output:

Traceback of TorchScript, serialized code (most recent call last):
  File "code/__torch__/models/yolo/___torch_mangle_1107.py", line 16, in forward
    _7 = torch.slice(_6, 3, 1, 9223372036854775807, 2)
    input = torch.cat([_1, _3, _5, _7], 1)
    _8 = ops.prepacked.conv2d_clamp_prepack(CONSTANTS.c0, CONSTANTS.c1, [1, 1], [1, 1], [1, 1], 1, None, None)
         ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ <--- HERE
    _9 = ops.prepacked.conv2d_clamp_run(input, _8)
    input0 = torch.mul(_9, torch.sigmoid(_9))

Traceback of TorchScript, original code (most recent call last):
    graph(%input, %weight, %bias, %stride:int[], %padding:int[], %dilation:int[], %groups:int):
        %output_min_max : None = prim::Constant()
        %packed_weight_bias = prepacked::conv2d_clamp_prepack(
                              ~~~~~~~~~ <--- HERE
            %weight, %bias, %stride, %padding, %dilation, %groups,
            %output_min_max, %output_min_max)
RuntimeError: Could not run 'prepacked::conv2d_clamp_prepack' with arguments from the 'CUDA' backend. This could be because the operator doesn't exist for this backend, or was omitted during the selective/custom build process (if using custom build). If you are a Facebook employee using PyTorch on mobile, please visit https://fburl.com/ptmfixes for possible resolutions. 'prepacked::conv2d_clamp_prepack' is only available for these backends: [CPU, BackendSelect, Named, InplaceOrView, AutogradOther, AutogradCPU, AutogradCUDA, AutogradXLA, UNKNOWN_TENSOR_TYPE_ID, AutogradMLC, Tracer, Autocast, Batched, VmapMode].

CPU: registered at ../aten/src/ATen/native/xnnpack/RegisterOpContextClass.cpp:84 [kernel]
BackendSelect: fallthrough registered at ../aten/src/ATen/core/BackendSelectFallbackKernel.cpp:3 [backend fallback]
Named: registered at ../aten/src/ATen/core/NamedRegistrations.cpp:7 [backend fallback]
InplaceOrView: fallthrough registered at ../aten/src/ATen/core/VariableFallbackKernel.cpp:60 [backend fallback]
AutogradOther: fallthrough registered at ../aten/src/ATen/core/VariableFallbackKernel.cpp:35 [backend fallback]
AutogradCPU: fallthrough registered at ../aten/src/ATen/core/VariableFallbackKernel.cpp:39 [backend fallback]
AutogradCUDA: fallthrough registered at ../aten/src/ATen/core/VariableFallbackKernel.cpp:47 [backend fallback]
AutogradXLA: fallthrough registered at ../aten/src/ATen/core/VariableFallbackKernel.cpp:51 [backend fallback]
UNKNOWN_TENSOR_TYPE_ID: fallthrough registered at ../aten/src/ATen/core/VariableFallbackKernel.cpp:43 [backend fallback]
AutogradMLC: fallthrough registered at ../aten/src/ATen/core/VariableFallbackKernel.cpp:55 [backend fallback]
Tracer: fallthrough registered at ../torch/csrc/jit/frontend/tracer.cpp:1027 [backend fallback]
Autocast: fallthrough registered at ../aten/src/ATen/autocast_mode.cpp:250 [backend fallback]
Batched: registered at ../aten/src/ATen/BatchingRegistrations.cpp:1016 [backend fallback]
VmapMode: fallthrough registered at ../aten/src/ATen/VmapModeRegistrations.cpp:33 [backend fallback]

Environment

• OS: [Ubuntu]
• GPU [1060 Ti]

Additional context

I'm just importing the model and calling forward. The strange thing is that the old model I've trained some month ago, and exported to torchscripts, works well ( no error ) in LibTorch. Looking into it I do not see ops.prepacked.conv2d_clamp_run
nodes in netron, while this last one, that gives me errors, does have it.
In order to be sure, I've already tested it with the last version of pytorch (libtorch) and it does not work.

Model exported following the tutorial.
It gives the same error also exporting the default yolov5s.pt to torchscript and the using it in Libtorch

[github-actions]

👋 Hello @FedeAi, thank you for your interest in 🚀 YOLOv5! Please visit our ⭐️ Tutorials to get started, where you can find quickstart guides for simple tasks like Custom Data Training all the way to advanced concepts like Hyperparameter Evolution.

If this is a 🐛 Bug Report, please provide screenshots and minimum viable code to reproduce your issue, otherwise we can not help you.

If this is a custom training ❓ Question, please provide as much information as possible, including dataset images, training logs, screenshots, and a public link to online W&B logging if available.

For business inquiries or professional support requests please visit https://www.ultralytics.com or email Glenn Jocher at glenn.jocher@ultralytics.com.

Requirements

Python 3.8 or later with all requirements.txt dependencies installed, including torch>=1.7. To install run:

$ pip install -r requirements.txt

Environments

YOLOv5 may be run in any of the following up-to-date verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):

• Google Colab and Kaggle notebooks with free GPU:
• Google Cloud Deep Learning VM. See GCP Quickstart Guide
• Amazon Deep Learning AMI. See AWS Quickstart Guide
• Docker Image. See Docker Quickstart Guide

Status

If this badge is green, all YOLOv5 GitHub Actions Continuous Integration (CI) tests are currently passing. CI tests verify correct operation of YOLOv5 training (train.py), testing (test.py), inference (detect.py) and export (export.py) on MacOS, Windows, and Ubuntu every 24 hours and on every commit.

[a28293971]

I solved this problem by deleting the optimization function.
you can have a try.

yolov5/models/export.py

Line 81 in 57b0d3a

optimize_for_mobile(ts).save(f) # https://pytorch.org/tutorials/recipes/script_optimized.html

[glenn-jocher]

@FedeAi @a28293971 thanks for the bug report and the proposed solution. It seems like optimize_for_mobile() is breaking export for c++ use cases unfortunately.

@FedeAi it seems like the best solution would be to only optimize_for_mobile() if an --optimize argument is passed?

[glenn-jocher]

@FedeAi @a28293971 good news 😃! Your original issue may now been fixed ✅ in PR #3093. To receive this update you can:

• git pull from within your yolov5/ directory
• git clone https://github.com/ultralytics/yolov5 again
• Force-reload PyTorch Hub: model = torch.hub.load('ultralytics/yolov5', 'yolov5s', force_reload=True)
• View our updated notebooks:

Thank you for spotting this issue and informing us of the problem. Please let us know if this update resolves the issue for you, and feel free to inform us of any other issues you discover or feature requests that come to mind. Happy trainings with YOLOv5 🚀!

[FedeAi]

Now it's working fine, thanks!!