Interesting hardware comparison with perf/watt

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

Jouni
Posts: 3759
Joined: Wed Mar 08, 2006 8:15 pm
Full name: Jouni Uski

Interesting hardware comparison with perf/watt

Post by Jouni »

Jouni
smatovic
Posts: 3472
Joined: Wed Mar 10, 2010 10:18 pm
Location: Hamburg, Germany
Full name: Srdja Matovic

Re: Interesting hardware comparison with perf/watt

Post by smatovic »

Apple M1 MAX CPU maxes out at 30W (w/o GPU) afaik, so it is close to AMD ryzen-7950x in ECO mode...

ryzen-7950x (ECO mode): 598.27
Apple M1 MAX: 619.16

--
Srdja
wickedpotus
Posts: 153
Joined: Sun May 16, 2021 5:33 pm
Full name: Aron Rodgriges

Re: Interesting hardware comparison with perf/watt

Post by wickedpotus »

smatovic wrote: Tue Dec 27, 2022 8:55 pm Apple M1 MAX CPU maxes out at 30W (w/o GPU) afaik, so it is close to AMD ryzen-7950x in ECO mode...

ryzen-7950x (ECO mode): 598.27
Apple M1 MAX: 619.16

--
Srdja
There are multiple issues with these kinds of comparisons...
  • It is comparing apples and oranges. Single-chip SOCs where GPU + RAM etc. cannot be changed disabled or removed, just breaking out parts of mandatory power-consuming stuff on a chip, makes comparison flawed, and rather meaningless.
  • A mobile phone with a Dimensity 9000 ARMv8 chip gets about 1200.00 on the "perf/watt" ratio used in sheet, but what's the usefulness of these measurements???
  • It is as useless as claiming a 1fps system running Crysis, is somehow "better" than a 30fps one if it just draws 1/40 of watts when running the game.
  • Or as useless as comparing only dB/watt when comparing speakers, disregarding the quality of sound they put out...
Werewolf
Posts: 2058
Joined: Thu Sep 18, 2008 10:24 pm

Re: Interesting hardware comparison with perf/watt

Post by Werewolf »

wickedpotus wrote: Wed Dec 28, 2022 9:34 pm
  • A mobile phone with a Dimensity 9000 ARMv8 chip gets about 1200.00 on the "perf/watt" ratio used in sheet, but what's the usefulness of these measurements???
  • It is as useless as claiming a 1fps system running Crysis, is somehow "better" than a 30fps one if it just draws 1/40 of watts when running the game.
A strong argument.