# AMD BC-250 PyTorch Patches Hit 2,992 Img/S on ResNet-9

> New patches for PyTorch's OpenCL backend push the AMD BC-250 mining board past a cloud T4 in ResNet-9 training speeds.

- Published: 2026-11-05T00:28:03Z
- Score: 7/10
- Tags: bc250, tool, ai
- Port / platforms: Linux / Steam Deck -> Linux / Steam Deck (Tool)
- Web page: https://emu-radar.customfw.xyz/news/pytorch-dlprim-bc250-patches/
- Sources: https://github.com/mxreyer/pytorch-dlprim-gfx1013/commit/e8d9bbd1eb7d28c6a379655dcbf8e8c93acdeb94

A new patch collection for [pytorch_dlprim](https://github.com/artyom-beilis/pytorch_dlprim) makes the OpenCL backend work correctly and run faster on the gfx1013 GPU of the AMD BC-250 under Mesa's rusticl driver.

Author mxreyer reports that Claude-coded optimizations lift stock performance from 915 images per second to 2,261 in fp32 on a ResNet-9 training step. The opt-in fp16 inner loop reaches 2,918 in the same test. These figures time the GPU step on random input, so data loading and augmentation are not included. In a full 16-epoch CIFAR-10 run, the patched board reaches 2,337 img/s in fp32, just past a Colab T4 at 2,294, and 2,992 with fp16.

Prebuilt `pt_ocl.so` files are attached to the [releases](https://github.com/mxreyer/pytorch-dlprim-gfx1013/releases) of the repository. They replace the `pt_ocl.so` in the stock pytorch_ocl 0.2.0 wheel for torch 2.4.0, Python 3.12, x86-64 Linux.
