AMD MI355X serves 32 concurrent Kimi-K3 agents per node, 1.6x B300's headcount
ryanshrout · x · 2026-09-23
Signal65's PINNACLE GPU testing shows AMD's Instinct MI355X beating Nvidia's B300 on agentic serving of Kimi-K3 (2.78T parameters, the largest open-weight model, in MXFP4): 32 concurrent agents per node vs 20 on B300 (1.6x), with sub-second first token at the ceiling, and faster from 16 agents up. AMD's engineering team says more optimization is still coming.
More from Infra
- $500 of Dell OptiPlexes become a diskless netboot lab where AI agents can't brick the hardware — colinmcnamara · 2026-09-23
- AMD MI355X beats NVIDIA B200 by 2.25x at scale with ~40% lower cost per GPU hour: Signal65 — ryanshrout · 2026-09-23
- 2.3B MoE hybrid Mamba model matches Llama-3.2-3B with <1% of its pretraining FLOPs — tri_dao · 2026-09-23
- ByteDance Seed fixes full-pipeline FP8 RL instability with Calibrated Clipping across 8B-32B models — ByteDance-Seed · 2026-09-23
- DigitalOcean Managed Agents Enters Public Preview: Pause-When-Idle Cloud Claude Code and Codex — _AustinCalvert_ · 2026-09-23
- ARK analyst: AI is the most powerful joule in history, converting energy to GDP ~10x better than humans — DMaguireARK · 2026-09-23