{
  "id": 14092,
  "url": "https://arxiv.org/abs/2607.23402",
  "title": "Characterizing Warp Divergence from Pascal to Blackwell",
  "summary": "Since Volta introduced Independent Thread Scheduling (ITS), NVIDIA GPUs have been widely assumed to handle warp divergence in a fixed manner. We test this assumption across Ampere, Hopper, and datacenter and consumer Blackwell GPUs, using pre-ITS Pascal as a baseline. Combining cycle-accurate microbenchmarks, hardware counters, and static analysis of compiler-generated SASS, we separate stable behavior from architectural change. Across all tested generations, divergent paths serialize linearly w",
  "authors": "Alpin Dale",
  "category": "research",
  "topics": "environment",
  "orgs": "nvidia",
  "regions": null,
  "published_at": "2026-07-25T20:00:00.000Z",
  "fetched_at": "2026-07-29T05:10:12.205Z",
  "source_slug": "hf-daily",
  "source_name": "HuggingFace Daily Papers",
  "source_homepage": "https://huggingface.co/papers",
  "ethics_ai_record_url": "https://ethics.ai/record/14092",
  "original_url": "https://arxiv.org/abs/2607.23402",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}