News for hackers. Stuff that matters.
Re:M5 Ultra Mac Studio Review: The Dream Mac for Local AI Ag... by bigyabai on Monday September 21, 2026 @12:58PM (#49789948)
It's likely that a stack of Macs will draw more power for slower prefill/decode than equivalently priced Nvidia GPUs. If power efficient inference is the goal, Macs are a non-starter.
Re:M5 Ultra Mac Studio Review: The Dream Mac for Local AI Ag... by _hugerobots_ on Monday September 21, 2026 @01:25PM (#49790351)
So if it isn't a comparative ability, now it's a power cost issue? This reads like goal post moving.
Re:M5 Ultra Mac Studio Review: The Dream Mac for Local AI Ag... by bigyabai on Monday September 21, 2026 @03:58PM (#49792541)
Oh, it's absolutely both. The power you waste waiting for TFTT on prefill will absolutely compound at the "medium sized labs and businesses" scale.