H100
3 published articles
AI Research
Video generation gets 120x faster on one GPU, inside Nvidia's hybrid attention breakthrough
Nvidia Research's SANA-Video 2.0 combines linear and periodic softmax attention in a 3:1 ratio to run 120x faster than Wan 2.2 on one H100. The hybrid design recovers full-rank expressiveness while keeping inference at 13 seconds for a 720p clip. The caveats: the benchmarks stop at 720p and bundle the attention change with proprietary optimizations.
2026-07-31
Infrastructure deep-dive
Nous Research's MoE field notes: what actually happens when 1 trillion parameters hit 1024 GPUs
Nous Research publishes field notes on scaling MoE expert parallelism with DeepEP. The report details throughput, configuration trade-offs, and bottlenecks from pretraining a 1T-parameter MoE model, offering practical deployment insights for large-scale distributed training.
2026-07-19
Market momentum
Nvidia just crossed $3 trillion again. The AI spending machine isn't slowing down.
Nvidia's market cap crossed $3 trillion as AI infrastructure demand keeps growing. The milestone underscores that enterprise GPU purchases remain strong despite growing competition from AMD and startups offering cheaper inference chips.
2026-07-16