CUDA Graphs
PulseAugur coverage of CUDA Graphs — every cluster mentioning CUDA Graphs across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
ShardFlow framework achieves 28 TPS on Qwen2.5-7B using speculative decoding
A developer has built ShardFlow, a distributed LLM inference framework that splits models across multiple machines and uses speculative decoding to mitigate WAN latency. Benchmarks on Qwen2.5-7B demonstrated a significa…
-
New WAM-Diff2 framework boosts autonomous driving VLA model efficiency
Researchers have developed WAM-Diff2, a new framework designed to improve the efficiency of Vision-Language-Action (VLA) models for autonomous driving. This framework uses a hierarchical distillation strategy to convert…
-
HARD-KV framework boosts LLM inference speed by 2x
Researchers have developed HARD-KV, a novel framework designed to optimize long-context Large Language Model (LLM) inference. This system addresses the conflict between head-adaptive compression algorithms, which offer …
-
AI inference latency limited by more than memory bandwidth, study finds
A new paper reveals that the inference performance of physical AI systems, such as robots and autonomous vehicles, is not solely limited by memory bandwidth as previously assumed. The research demonstrates that while ba…