Source item
Open original source ↗InferenceX, formerly InferenceMAX, is changing the industry
Performance of hardware + software is constantly evolving with new models and new ML Systems techniques
All major GPUs from AMD and Nvidia + soon TPUs / Trainium
AMD said they'd send a pic of Lisa with their belt soon! https://t.co/NauZyBPr33
Quoted context
These are the quoted author’s words.
InferenceX v2: NVIDIA Blackwell Vs AMD vs Hopper - Formerly InferenceMAX, GB300 NVL72, MI355X, B200, H100, Disaggregated Serving, Wide Expert Parallelism, Large Mixture of Experts, SGLang, vLLM, TRTLLM https://t.co/TG9lBwL6iL
SemiAnalysis_ ↗