Accelerated LLM Inference on AMD ROCm via Isolated MLP Compilation
Build and run your first accelerated benchmark on AMD GPUs
How does docc accelerate LLM inference on AMD GPUs
Compare latency and throughput against stock vLLM on standard workloads
Validate compiled kernel mathematical fidelity against eager PyTorch
Validated Qwen 2.5 checkpoints, planned models, and parameter reference