The Daily Diff ships PDF edition; highlights GPU perf engineering and suffix KV cache reuse
arpit_bhayani · x · 2026-10-03
Arpit Bhayani's curated engineering newspaper The Daily Diff (tdd.cat) now supports PDF downloads — a user-requested feature he built over a long weekend. The paper curates applied AI, databases, and system design reads.
Highlights from the edition:
- GPU performance engineering curriculum (wafer-ai, GitHub): a bottom-up roadmap from GPU hardware architecture and Triton/CUTLASS kernel optimization to KV cache memory layout, continuous batching, and tensor parallelism — noting naive implementations can lag tuned engines by orders of magnitude
- Suffix KV cache reuse (Rulin Shao, Oscar Yin): vLLM/SGLang's radix-tree reuse assumes append-only context; agent edits mid-context break prefix matching, and suffix cache reuse eliminates redundant token recomputation during edits
More from Infra
- AI gateways emerge as key control layer, with eight core capabilities to govern production AI stacks — goyalshaliniuk · 2026-10-03
- Sentdex straps 4x RTX Pro 6000 cards onto an Asus GB300 DGX Station — Sentdex · 2026-10-03
- Qwen 3.8 Flash Next q2_0 hits 10 tok/s on a 6GB VRAM laptop via Strata engine — dampflokfreund · 2026-10-03
- Strata hits 5-10x prefill speedup in two weeks, shaming llama.cpp maintainers — Training_Visual6159 · 2026-10-03
- One RTX 4080, One Night: Full AI Short Film With Qwen Images and Minimax H3 Video — CupQuakeBE · 2026-10-03
- Data centers are unpopular across America, but Loudoun County, their capital, thrives — arian_ghashghai · 2026-10-03