LLM Digest

Story

arxiv_cs_lg · Oct 3, 2026 · paper

Source brief

PhaseGate: Phase-Aware CPU Retrieval Scheduling for On-Device LLMs on Unified Memory

arxiv.orgOct 3, 2026
original source linked

In brief

On-device assistants run GPU-based LLM inference alongside CPU retrieval on unified-memory systems. Under a saturated local-retrieval workload, four concurrent retrieval workers raise 95th-percentile (p95) decode late...

Feed lens

eval

Read the original at arxiv.org →Open in live feed Read that day’s brief

PhaseGate: Phase-Aware CPU Retrieval Scheduling for On-Device LLMs on Unified Memory

Earlier in this thread 4 items

HHR: Hierarchical Hash Retrieval for Efficient LLM Generation

Show HN: RouteMind – routing instead of retrieval for agent knowledge

MERGE: Multi-LLM Ensemble for Retrieval via Generative Enrichment

RAMP: Robust Adaptive Mixed-Precision Quantization for Edge CPU Vision Models