CRISP: Co-designing Pruning and Scheduling for Efficient MoE Inference on Edge Servers
Research framing: expert-weight movement dominates edge MoE inference.
AI Storyline
3 items · 2 sources · 3 days
Operational story trace
MoE Inference
Follow in this browser to see new updates on your Live feed.
Current stateDeveloping
Latest change
A Show HN project (Nov 3) claims faster MoE inference on a Mac with less RAM, following a Nov 2 post on running a 124B MoE on one DGX Spark.
Earlier contextThe story so far
An Oct 27 arXiv paper framed edge MoE inference as an expert-weight movement problem rather than a compute problem. Two Hacker News posts followed within a week, showing the same memory constraint being attacked on a single DGX Spark and on Macs.
editor-curated · source-linked
Arc
Oct 27Nov 3 · now
RESEARCH · Oct 27CRISP: pruning plus scheduling for edge MoE1 source · show source ▾
HANDS-ON · Nov 2124B MoE tuned on a single DGX Spark1 source · show source ▾
Optimizing 124B MoE inference on a single Nvidia DGX Spark
Practitioner report on fitting a 124B MoE onto one DGX Spark.
NOW · Nov 3Show HN: faster MoE on Mac with less RAM1 source · show source ▾
Show HN: Run faster MoE inference on Mac with less RAM
Applies the memory-reduction goal to Mac hardware.
Day 1 Tuesday, Oct 27, 2026
CRISP: Co-designing Pruning and Scheduling for Efficient MoE Inference on Edge Servers
Research framing: expert-weight movement dominates edge MoE inference.
Day 2 Monday, Nov 2, 2026
Optimizing 124B MoE inference on a single Nvidia DGX Spark
Practitioner report on fitting a 124B MoE onto one DGX Spark.
Day 3 Tuesday, Nov 3, 2026
Show HN: Run faster MoE inference on Mac with less RAM
Applies the memory-reduction goal to Mac hardware.
What to watch — open questions
- Do the Mac and DGX Spark projects publish tokens-per-second and RAM figures that others can reproduce?
- Do practitioner projects adopt expert pruning or scheduling ideas like CRISP's?
How this thread was built
editor wrote the arc · 3 beatswatcher 1 status change
Storylines are threaded mechanically from the feed: stories that share a distinctive anchor across multiple days and sources. Each item links to its original source. The evidence trace, current state, and open questions are written by the editor routine and refreshed whenever a new beat lands.