M5 Ultra LLM test: 4x faster prompt processing, but double the power draw
DigitalguyCH · reddit · 2026-09-23
A Reddit user posted early benchmarks of running LLMs on Apple's M5 Ultra (video linked in original):
- Prompt processing: up to 4-4.5x faster than M3 Ultra depending on context length
- Generation: 1.5x faster tok/s
- Trade-offs: the machine draws twice the power (400W vs 200W), runs louder fans, and runs much hotter
Big performance gains, but energy efficiency regressed — a real trade-off for quiet, low-heat local deployment.
More from Embodied
- CoRL results show sim-only training now delivers top real-world robot performance — carlosdponx · 2026-11-12
- Robot fails at folding a towel, reassuring roboticists frontier models won't eat robotics yet — chris_j_paxton · 2026-11-12
- Nuro podcast: manufacturing, charging, and regulation gate robotaxi mass adoption — R_D · 2026-11-12
- Rodeo Robotics demos robot that changes your car tire — ycombinator · 2026-11-12
- Cybercab rear-ended by human driver, researcher quips: the problem is the humans — chris_j_paxton · 2026-11-12
- HT-Policy Uses Heavy-Tailed Student-t Regression to Close Robot Policy Gap — chris_j_paxton · 2026-11-12