{"type": "article", "title": "Benchmarking Speculative Decoding on DGX Spark: 11.5 → 29.5 tok/s", "publisher": "Web Pulse", "url": "https://wpnews.pro/news/benchmarking-speculative-decoding-on-dgx-spark-11-5-29-5-tok-s", "original_source": "https://alephinitesimal.com/posts/bandwidth-ceiling.html", "published": "2026-08-20T18:59:52+00:00", "accessed": "2026-08-20", "id": "benchmarking-speculative-decoding-on-dgx-spark-11-5-29-5-tok-s"}