G5 @ 2.5 GHZ *NOT CONFIRMED* but here's how
I actually got up at 3.30am to figure this out, as I was just lying there dreaming about all the goodness Steve will give us on Monday.
Ok. We know the G4 7450 (not Apollo) scales to 867GHZ
The increase between 7400 & 7450 was 3 extra stages. An increase of 75%. Put into clockspeed this gives us 500(g4) > 875(7450) so we've seen that a %increase in stages gives a near linear %increase in Mhz.
I also assume that a 7450 can clock to about 1GHZ, as there have been people foolish enough to do this, and this is probably what we'd expect (133mhz) from the last revision in July.
Now. Increasing the 7stage pipeline to 10 Stages (rumoured G5) would be an apx 40% increase.
This means our pretend G5 manufactured on the 7450 assy line would yield 1400MHZ.
Now. The die shrink. From 0.18 microns to 0.13 is another %increase of 40%. Now we have our pretend 1400MHZ G5 running at 1960MHZ
Now. Factor in the SOI. FWIW the rumour is that this adds about 30% to clockspeed. So our nice G5 is now running at 2548MHZ (2.5GHZ)
So If all these things are in place, And remember MOTO could be getting others (AMD?) to subcontract these parts, its not totally unreasonable to assume that (in an imperfect world) the G5 could debut somewhere between 1.6-2.2GHZ, and would be spot on with MOTO's PPC roadmap, AND a good reson for Apple to be creaming all over thier front Page.
*Not going back to bed for a bit, wondering how many flames I can amass!*
[ 01-05-2002: Message edited by: MarcUK ]</p>
Ok. We know the G4 7450 (not Apollo) scales to 867GHZ
The increase between 7400 & 7450 was 3 extra stages. An increase of 75%. Put into clockspeed this gives us 500(g4) > 875(7450) so we've seen that a %increase in stages gives a near linear %increase in Mhz.
I also assume that a 7450 can clock to about 1GHZ, as there have been people foolish enough to do this, and this is probably what we'd expect (133mhz) from the last revision in July.
Now. Increasing the 7stage pipeline to 10 Stages (rumoured G5) would be an apx 40% increase.
This means our pretend G5 manufactured on the 7450 assy line would yield 1400MHZ.
Now. The die shrink. From 0.18 microns to 0.13 is another %increase of 40%. Now we have our pretend 1400MHZ G5 running at 1960MHZ
Now. Factor in the SOI. FWIW the rumour is that this adds about 30% to clockspeed. So our nice G5 is now running at 2548MHZ (2.5GHZ)
So If all these things are in place, And remember MOTO could be getting others (AMD?) to subcontract these parts, its not totally unreasonable to assume that (in an imperfect world) the G5 could debut somewhere between 1.6-2.2GHZ, and would be spot on with MOTO's PPC roadmap, AND a good reson for Apple to be creaming all over thier front Page.
*Not going back to bed for a bit, wondering how many flames I can amass!*
[ 01-05-2002: Message edited by: MarcUK ]</p>
Comments
I am ready to drop serious $ for the Mac you just described. I think you have great points.
Apple floundered for years at 500 Mhz, meanwhile they were signing deals with AMD, and cranking out newer chip designs with IBM (SOI) and MOTO to produce killer chips (64 bit maybe)
Apple knows damn well that the Mhz myth spin may be true but it wont sell computers.
I bet we see a new IBM sticker on Macs with the PPC logo again, or maybe a "Powered by IBM" sticker to compete with Inte more.
Can you imagine the TV commercials for a ? 2 GHz G4!
We may be blind-sided with these bad boys Monday AM!
I just hope Steve won't say: "Order today, they will be ready in Spring 02!". I will flip out! That really means June 19th...
WHat would be even cooler is a MP version of this thing, a dualie 2.5 would be a cool .5 terahertz.
Niiiiice.
MSKR
PS I will also kiss you if ti comes true.
Sweeeeeet....
-robo
Eh?
1000MHz = 1GHz
1000GHz = 1THz
5GHz not= .5 THz\\
[grr, option equals didn't translate]
[ 01-06-2002: Message edited by: SkullMac ]</p>