You're doing this wrong.
What we're doing with the Wii U is to compare raw horsepower, not what visual differences it might originate from it.
What you're saying, on the other had, is simply absurd. A 500x generation leap, really? Then please tell us what color pallet differences are between generation 6 and 7 (current) with those multipliers of yours so we can see if that was a "true" generation leap too.
In terms of horsepower the differences between the NES and SNES don't even come close to 3x, at least in the CPU department. And it was a giant leap in terms of what you'd visually get. There were other improvements as well, like a lot more RAM and better audio, so you can't simply determine a system's power by comparing it to other that way (with silly multipliers).
Have POWER7 CPU's for example. Although clocked lower compared to other CPU types, you get a lot (and I mean really really a lot) more performance out of it simply because it has a greatly optimized architecture (Wii U CPU).
I never intended a single performance jump to be used as the sole multiplier determining system power differences.
Being that, in raster graphics a single sprite or tile is the closest comparative to the primitives we have today (polygons), how many different colors and tiles and sprites you can have is the biggest indication of horsepower around. MUCH more telling than something as stupid as cpu bus size.
If I did I wouldnt have brought up other things like simultaneous colors displayed, advanced computational overhead like scaling and rotation, and number of simultaneous sprites displayed.
I also wouldnt have asked people to compare the otger performance measures I didnt feel like bringing up, things like resolution ram and simultaneous tiles displayed.
Typically, if you want a multiplier to be anything more than a worthless number meaning nothing, you make note of all the performance increase multiples, and make an average out of them. You usually end up with about 20-30x each generation.
Since you didnt make the meerest attempt to compare any of the metrics i suggested, your rebuke is worthless.
Since you refused to make any effort in obtaining incredibly easy to find facts, and you rebuked the hardest evidence of observable increases in performance to hide behind the magical nonterm 'horsepower', with absolutely no supportingvevidence, i will do it for you.
SNES
CPU reference
Processor Ricoh 5A22, based on a 16-
bit 65c816 core
Clock rates
(NTSC)
Input: 21.47727 MHz
Bus: 3.58 MHz, 2.68 MHz,
or 1.79 MHz
Clock rates
(PAL)
Input: 21.28137 MHz
Bus: 3.55 MHz, 2.66 MHz,
or 1.77 MHz
NES
CPU 2A03 - customized 6502 CPU - audio -
does not contain support for decimal
The NTSC NES runs at 1.7897725MHz, and
1.7734474MHz for PAL.
snes, 21MHz (requires fx2 for max effectiveness)
nes 1.7 MHz (requires c3 for max effectiveness)
Thats 13x The number of cycles a second.
I will leave it up to you to guess whether the snes cpu performed more instrucions a cycle than the nes, (this is your p7 example, by the way)
Protip, from 1975-2003/4, processors regularly improved both cycles per second and instructions per cycle. they didnt have to choose because dennard scaling was in full effect.
Snes gave 128KB ram to the cpu, and 64KB of ram for the video buffer. 192 KB ram.
Nes 2KB ram for cpu 2KB ram for video buffer 4KB Ram.
Snes had 48x the ram capacity as snes.
I will leave it up to you to figure out which systems ram was faster, and had lower latency, and thus the higher bandwidth.
protip, it was the snes.
Yeah, the raw specs are saying a hellova lot more than 'nowhere near 3x'.
Actually, they are nowhere near 3x, just in the opposite direction of what you postulated.
Edited by 3Dude, 25 September 2012 - 06:55 AM.
