sglang rocm

1 article

Prefill Context Parallelism on MI355X: Long-Prompt Gate

Setup and on/off gate for prefill context parallelism on MI355X in 2026: AMD's GLM-5.2 test lost 8% to 18% throughput at 1,024 tokens and gained 43% to 54% at 40,960 tokens or more.

November 21, 2026• 30 min read
prefill context parallelismcontext parallelism mi355xamd instinct mi355x