Tag: #cuda-alternative
Writing on GPUs, LLMs, MLOps, Kubernetes — and mindset · 1 posts
AMD GPU & ROCm Deep Dive: Can It Challenge CUDA for LLM Inference?
A thorough technical analysis of AMD MI300X with 192GB HBM3, the ROCm software stack, and HIP programming model. Includes real LLM serving benchmarks with vLLM and llama.cpp, and an honest assessment of strengths and wea
2026-03-18 · 14 min read #amd#rocm#gpu#mi300x#model-serving