Huawei Reconstructs Scaling Boundaries with NPO
bookwormengr · x · 2026-07-14
Discusses Huawei's approach to NPO: attempting to eliminate the traditional distinction between scale up and scale out, unlocking interesting possibilities like cluster-wide CPU RAM sharing for KV caching.
Replies also noted that Innolight and Accelink are playing major roles in NPO-related work targeting US hyperscalers.
More from Infra
- Encrypted GPT-2 speedrun challenges FHE inference: best is 3.031 s/token, goal sub-1s — tensorqt · 2026-10-27
- Open-source Latch compiles PyTorch into training megakernels, 2.75x faster on H100 — sguada · 2026-10-27
- Stanford launches GPU-centric supercomputer Marlowe for data science and AI research — StanfordHAI · 2026-10-27
- NVIDIA Claims HGX Rubin NVL8 Delivers 10x More Tokens per Megawatt Than B200 — thursdai_pod · 2026-10-27
- Together lets multiple LoRA post-training experiments share the same GPU capacity — togethercompute · 2026-10-27
- Commoditization dilemma: either human labor or GPUs become the commodity — _AustinCalvert_ · 2026-10-27