0

We currently have a trained ResnetV250 image classification model that performs as expected on CPU and with GPU support on Turing based cards with CUDA 10.1 and cudnn 7.6.4.

When transferring this to machines with Ampere based cards, we get similar expected results for CPU runs but when enabling GPU support we get fixed value results no matter what inputs are applied.

The drivers on all cards are up to date and relevant CUDA and cudnn versions are installed. On first start up with the Ampere cards we see the Compute cache being constructed and used on subsequent classification runs.

Could there be an issue with nvcc in constructing the PTX?

Improve this question
Related questions
5
C# ML.Net Image classification: Does GPU acceleration help improve the performance of predictions and how can I tell if it is?
2
Why won't ML.NET Model Builder GPU extension install properly?
9
Non-OK-status: GpuLaunchKernel(...) status: Internal: no kernel image is available for execution on the device
Load 7 more related questions Show fewer related questions

0

Reset to default

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.