Serving Kimi K3 on AMD MI355X: 3.8x Throughput and 71% Lower Cost vs B200
SumitGup · x · 2026-08-04
The Wafer.ai team successfully deployed the massive 2.8T parameter Kimi K3 model on AMD MI355X chips. Through specific optimizations, the MI355X achieved an aggregate throughput of 952 tok/s per node.
Compared to an NVIDIA B200 node, the AMD setup delivers 3.8x higher aggregate throughput and 1.3x faster single-stream performance, while reducing inference costs by 71%. This breakthrough was driven by MI355X's 288GB VRAM capacity and AMD's day-0 software support, proving its price-to-performance advantage over NVIDIA's flagship GPUs.
Related event: AMD MI355X Runs Kimi K3 at 952 tok/s, Outperforming B200(6 posts)→
More from Infra
- GitHub rewrote the 800k-line Copilot runtime in Rust using AI agents, cutting CPU and memory by over 90% — bibryam · 2026-10-16
- NVIDIA quietly delists the entry-level RTX 5050 from its marketplace, following the RTX 5090 — chemist_slime · 2026-10-16
- Grok claims 816B MoE model runs locally on dual M5 Ultra Mac Studios via MLX — grok · 2026-10-16
- Same 1,000 Cloudflare Durable Objects: $4,050/month cut to $22 with one trick — threepointone · 2026-10-16
- NVIDIA ships FP4-quantized Qwen3.8-Flash for local use, trending on Hugging Face — nvidia · 2026-10-16
- Baidu's Virtual Head Attention tames the KV cache, FLOPs, quality triangle in long-context LLMs — baidu · 2026-10-16