DFlash Accelerates Blackwell Inference
PyTorch · x · 2026-07-10
PyTorch introduced DFlash, a PyTorch-based block-diffusion model for speculative decoding aimed at accelerating LLM inference on NVIDIA Blackwell.
The post notes that researchers released a related paper and implementation details: DFlash uses a lightweight block-diffusion drafter to generate candidate tokens in parallel, which are then verified by the target model. NVIDIA states that at the same interaction level, it can boost the throughput of gpt-oss-120b by up to 15x, and it is now available in TensorRT-LLM, SGLang, and vLLM.
More from Infra
- RTK Terminal Compression Cuts Tokens but Leaves Your AI Coding Bill Unchanged — Bartaseth · 2026-09-11
- SF Compute founder: buying compute is 'an absolutely awful experience' right now — IgorCarron · 2026-09-11
- SmolVM open-sources persistent computer infrastructure for agents that outlive chat sessions — aniketmaurya · 2026-09-11
- PyTorch Day Korea 2026 launches first offline conf, CFP closes Sept 13 — PyTorch · 2026-09-11
- Local LLM server dilemma: 4x CMP-170HX (price up 53% in 20 days) vs Mac Studio M5 Ultra — rumboll · 2026-09-11
- llama.cpp lands Flash Attention tuning for RDNA4, big prefill gains on AMD — pmttyji · 2026-09-11