ProductAMD Instinct MI355X / AMD MI355X
MI355X
A data-center AI accelerator card from AMD based on the CDNA 4 architecture. It is equipped with HBM3E memory, delivers aggregate memory bandwidth exceeding 6 TB/s, and supports FP8-precision inference. The card is designed primarily for large-model training and inference workloads.
Timeline (last 90 days)
Oct 3
Through request-owned speculative KV, in a DeepSeek-V4-Pro 1P1D, MI355X, concurrency-256 scenario, the average TTFT was reduced from 19.4 seconds to 15.1 seconds.
Unverified50%