MinDXNN is a native HLSL and DirectX 12 library for lightning-fast MLP inference leveraging AMD Radeon™ RX 9000 series matrix cores via cooperative vector APIs, delivering optimized kernels, samples, full source and docs to remove compute interop friction.

AMD GPUOpen

AMD has released MinDXNN, a native HLSL and DirectX 12 library for multi-layer perceptron (MLP) inference. It leverages AMD Radeon RX 9000 series GPU matrix cores via cooperative vector APIs, delivering performance comparable to dedicated ML frameworks without requiring compute API interop (e.g., HIP). The initial release includes optimized MLP inference kernels, unit tests, sample applications for 2D image reconstruction, and full source code. Training support is planned for the next release. Currently requires AMD developer drivers and RX 9000 series hardware.

Introducing MinDXNN: MLP library for DirectX 12