2

John Carmack tweeted:

For the same given paper spec, a console will deliver twice the perf of a PC, and a PC will deliver twice the perf of a mobile part.

Normally this is the sort of claim I would take with a really big grain of salt, but considering the man's pedigree. I'm inclined to take at least give him the benefit of the doubt.

I'm mostly interested about the first part of said claim, the one involving PCs and consoles, I'm well aware of the power and heat issues that mobile parts have to deal with so the latter part is more believable.

To formulate this concisely: What factors could explain that a specific (and fixed) hardware setup can have twice the performance as generic setup made from off the shelf parts considering that on paper both of these are deemed to be equivalent?

Can it really be down to just really optimised drivers?

P.S: Any resemblance to this question may or may not be intended.

Community
  • 1
Borgleader
  • 149
  • 6
  • 1
    Is there any other context? Obviously all specs aren't created equal. There are games ported between the three classes of platforms, so good comparisons *are* possible. But there's no way this is a general rule, or that performance differences can't be explained in terms of hard specifications of the respective machine architectures. – Potatoswatter Feb 19 '14 at 06:58
  • 2
    The tweet is wrong; this might be due to the length one allows to type, but the words that are missing out aren't helping. I can pick one console (perhaps a Wii) and one PC and show that this is wrong; so, there are definitely details missing that limit the possibilities that you choose, or that allow a (ac)cumulative view on it. A very first question would be: Is this on average, or mean? But realistically, John needs to provide which statistics he is basing himself on; otherwise it's nothing more than a weightless opinion, which indeed is speculation given all the factors that are involved. – Tamara Wijsman Feb 19 '14 at 08:08
  • Other questions include: How is the system set up? Which benchmark or game test is used? Does that particular benchmark or game test use the same graphical features and graphical memory size? (Less features or size being used on the console can yield a higher FPS in trade of worse quality) Which cooling was used for which devices? Which devices does this pertain to and/or how is the mean/average of that calculated? (consoles, PCs and mobiles differ in speed) Is it just a broad view or where actual proper multiple runs done to gather statistics on this? I see now John made two follow-up tweets: – Tamara Wijsman Feb 19 '14 at 08:15
  • https://twitter.com/ID_AA_Carmack/status/436012724791681024 https://twitter.com/ID_AA_Carmack/status/436020810508021760 – Tamara Wijsman Feb 19 '14 at 08:15

1 Answers1

-1

I believe what he was referring to is a broad view on the amount of additional software that is necessary on today's pc's which creates more hoops for processes to jump through (unlike the days of playing Double Dragon in DOS).

On the other hand a console that was manufactured for a sole purpose (and doesn't have any extra hoops to jump through) thrives in its own environment because it is lacking unnecessary steps on the way.

I came to this conclusion after reading his response to a comment on this tweet. https://twitter.com/ID_AA_Carmack/status/436012724791681024

Trae Abell
  • 171
  • 5
  • Down voting it does not change the fact that this is the actual answer. From the Twitter link I provided: "William: was it different back in the DOS days?" "John Carmack: Definitely, you could get all the perf out of the PC back then, it was more console-like than today's consoles." If you don't like the answer go argue with John Carmack. – Trae Abell Feb 20 '14 at 23:45