Huawei's Ascend 950DT TDP hits 950W per card; B200 ~4.4x denser in FP8/W
teortaxesTex · x · 2026-10-04
Based on Huawei's own figures, the Ascend 950DT NPUs in the 950 SuperNode (1024 NPUs) have a TDP of 950W each. At dense FP8/W, the author estimates Nvidia's B200 delivers roughly 4.4x the energy efficiency.
System-level power figures haven't been disclosed; extrapolating from Nvidia's NVL72, the supernode overhead is likely around 40-50%.
More from Infra
- Musk: xAI to hit 10GW of compute by end of next year, sees AI inference moving to space — beffjezos · 2026-10-04
- BF16 rounding breaks a conservation law, blowing up FlashAttention gradients late in training — HongyiWang10 · 2026-10-04
- Getting PyTorch CUDA training running on BC-250 boards, captured as an image — redfoxkiller · 2026-10-04
- Huawei 950 super-node claims seamless scaling from 550B/1.6T up to 10T-class models — teortaxesTex · 2026-10-04
- Huawei: 1000+ Ascend 910C supernodes deployed, 40+ LLMs natively pretrained — teortaxesTex · 2026-10-04
- The curse of 64GB RAM: Strata pushes local Qwen3.8-Flash-Next to 60 t/s but hogs system memory — Cautious_Chicken_604 · 2026-10-04