Tech Meridian ← ENTITY INDEX
PROMY MERIDIAN RU

COMPANY · ENTITY #5706

FlashAttention

Related event timeline, sources and context from the news index.

EVENT TIMELINE

2

CODING · 1 SOURCE · InfoQ AI, ML & Data Engineering

Meta develops Triton Low-Level Extensions (TLX) to expose advanced GPU execution controls

Meta developed Triton Low-Level Extensions (TLX), a native extension to the Triton compiler that exposes low-level GPU execution features via a new Multiple Instructions, Multiple Warps (MIMW) model. Meta says TLX expands Triton across warp-group orchestration, synchronization, memory orchestration, cluster execution and optional layout, and reports production gains (e.g., 20–25% MFU and up to 2× forward-pass speedup on Blackwell GPUs; 4× for an H100 linear kernel in IKBO), with plans to broaden hardware support and integrate with torch.compile.

8.0

RESEARCH · 1 SOURCE · arXiv cs.AI

RBS-Attention: radius-adaptive dual-branch sparse prefill for long-context LLMs

arXiv preprint arXiv:2609.20971v1 introduces RBS-Attention, a training-free sparse-prefill method that combines a centroid base branch with a radius-based rescue branch to avoid 'mean dilution' when selecting sparse blocks for long-context self-attention. On H100 GPUs the method reports up to 20.65× standalone prefill-attention speedup, 11.92× vLLM prefill-attention speedup and 5.97× end-to-end time-to-first-token speedup at 128K on Qwen3-30B-A3B-Instruct-2507-FP8, while retaining near-dense accuracy (88.65 vs 89.52 RULER on dense Qwen3-32B); evaluations include LongBench-v2, InfiniteBench and Video-MME.

7.0