Fetching from the wire…
01
02
03
04
05
06
07
08
Infra2026-08-16 · source-backed
Measuring expert time on two datacenter GPU generations shows it is linear in neither token count (EPLB, LPLB, UltraEP) nor activated-expert count (METRO). Below roughly 156 to 168 tokens, HBM weight streaming dominates so cost attaches to activated replicas. Above it, grouped GEMM rounds tokens to 128-tile M-tiles so splitting an expert adds padded compute (arXiv). Real decode batches hold hot and cold experts in both regimes at once, and recorded batches show proxy dispatchers differ by 1.4 to 1.6x in modeled block time. TEMPO's makespan-aware dispatcher gains 4 to 6% throughput and cuts p99 latency about 15.6% on Qwen3-235B, while showing only mechanism cost on communication-dominated DeepSeek-V3. The authors claim a phase diagram, not a universal win.
Each link below shares sources, entities, or timing with this story.
DeepSeek competes with Anthropic / Shared entities / Shared topic / Earlier coverage / Tension
Linked by a graph relationship (DeepSeek competes with Anthropic); both cover GPU, MoE, Qwen3; overlapping topics (cost, token).
DeepSeek competes with Anthropic / Shared entities / Earlier coverage
Linked by a graph relationship (DeepSeek competes with Anthropic); both cover DeepSeek, MoE, Qwen3; earlier DeepSeek coverage from 2026-08-10.
DeepSeek partners with Huawei / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (DeepSeek partners with Huawei); both cover DeepSeek, MoE; overlapping topics (cost, token).
DeepSeek competes with Anthropic / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (DeepSeek competes with Anthropic); both cover DeepSeek, MoE; overlapping topics (cost, token).
DeepSeek competes with Anthropic / Shared entities / Earlier coverage / Tension
Linked by a graph relationship (DeepSeek competes with Anthropic); both cover MoE, Qwen3; earlier MoE coverage from 2026-04-21.
Linked by a graph relationship (DeepSeek competes with Anthropic); both cover DeepSeek, MoE; earlier DeepSeek coverage from 2026-04-20.
DeepSeek criticizes Claude / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (DeepSeek criticizes Claude); both cover DeepSeek, GPU, Qwen3; overlapping topics (cost, token).
GPT competes with DeepSeek / Shared entity: Qwen3 / Same source domain / Shared topic / Earlier coverage
Linked by a graph relationship (GPT competes with DeepSeek); both cover Qwen3; reported by the same outlet (arxiv.org).