An HN post details 11-16x faster LLM inference using Llama.cpp with Apple Silicon and macO

Prewire · case file · public copy of the record

FILE P334 PREWIRE receipts archive the live wire

CATCH #634 DETECTED AUG 11, 2026 15:40 UTC

THE CATCHscore 7/10

EXHIBIT 634 · URGENT

Aug 11, 2026 15:40 UTC hacker news https://news.ycombinator.com/item?id=49259339

An HN post details 11-16x faster LLM inference using Llama.cpp with Apple Silicon and macOS VMs.

READY TO POST:

llama.cpp users are seeing huge speedups (11-16x!) on apple silicon with macos vms. pretty wild performance gains.

if you're running llms locally on apple silicon, there's new research showing massive speed improvements with llama.cpp and macos vms. worth a look.

EVIDENCE — the receipts

HN story climbing fast:
title: Apple Silicon and macOS VMs: 11–16× Faster LLM Inference with Llama.cpp
points: 45 in 0.8h = 54 pts/hr
link: https://github.com/trycua/cua/blob/main/blog/gpu-passthrough-macos-vms.md
discussion: https://news.ycombinator.com/item?id=49259339

caught by the 20-minute sweep · the live wire · all receipts

THE WIRE RECEIPTS ACCESS @VEDOLOS TERMS PRIVACY break the story before it breaks