Calebj14 said:
OMGosh! did thatwendigo and DHM just agree???? LOL
!
You want to know the kicker, the one that makes this even more amusing? I was agreeing because of past prejudice, and since have slightly changed my opinion because other people have shown me some new information.
Go figure.
On a side note, i do not want another G4. Moto had its chance and they majorally blew it! I want the G5 to come screaming down into at least PBs and iMacs! Also, and G5 Cube would be nice!
The G5, as things stand right now, is not much faster at the lower clockspeeds that would be necessary to cram it into the smaller formfactors. This is why something like the VX or e600/7000 is going to be a better choice for a while. Most of the advantages of 64-bit computing are denied in a system with limited space for RAM, in any case.
The VX was not merely a 750 with Velocity Engine tacked on. It was based on the 750 - that was the starting point. The original idea was that it was going to be available in March for mass distribution @ 90nm ... Apple couldn't have kept up if the 970 was not viable in a laptop - and their consumer end would languish for about two years as they waited for the G5 to filter down to every Mac -- too long when Moto was not to be trusted.
Interesting stuff.
Do you know what alterations they made to the chip, in general? The Centrino, which I've been using as a comparison in terms of back-stepping to older technology and then adding new features, keeps the Pentium-3 execution core and a lower staged pipeline in order to drive heat down, but has the faster Pentium-4 memory bus and what, by my reading, is a fairly interestingly power managed L2 cache. Fixing up the memory and front-side
buses in the G3, tacking some extra cache on, and tuning the power management sounds like it would have been the way to go with the VX. After all... Apple's been against the megahertz myth for years, so it shouldn't be too hard to sell a chip to professionals that is clocked the same as the curent ones, as long as it can be shown to perform better.
Even though it was killed off, it was far enough along that it could be fabbed in a couple of months; they didn't destroy their work. But there is no fabbing scheduled for the design as of the last report I received a couple of weeks ago.
Good to know it's at least waiting in the wings, if it turns out that Freescale is going to pull the same mistakes and unfortunate prejudices that their predecessor did. I'm of the opinion that Apple's better with one
good manufacturer supplying their stuff, but if we get the kind of competitive pushing that the PC world has with Intel/AMD, then that can only benefiet us as consumers.
As for the 970FX - it can go into a laptop and may have been planned to if not for the 90nm delays, and the fact that a 1.6GHZ G5 is not too much faster than a 7447A running at 1.5GHZ (there was a FCP4 benchmark posted earlier that shows this - imagine if their bandwith and memory were equal).
I spent a not-inconsiderable time pointing this out and extrapolating the numbers, actually. I've never been in favor of the G5 in a portable.
Another thing I want to say is to those who want a firm schedule from FreeScale (this is not aimed directly at thatwendigo or anyone else) - why should they release a timeline publicly? Remember that Apple contributes a significant amount of revenue to them (my guess is a couple hundred million a year in revenue comes from Apple). Apple just introduced new PowerBooks - and PB sales were in a slump before the refresh. If FreeScale announces a few weeks later that there will be 2GHZ G4s with on-board memory controllers, true DDR support, and 1MB L2 cache fabbed at 90nm available in the middle of the summer, Apple's PowerBook sales would drop like a rock because everyone would anticipate an update not long after the introduction - and 2GHZ is a big jump, especially when combined with increased bandwith and an enlarged cache. There's also the problem of getting everyone's hopes up.
Point conceded. That's actually a damn good set of reasons not to say
anything about when they'll be available. I don't think that we'll see major updates until after the summer, at the very earliest. Whatever way they go, Apple's going to have to redesign boards for a chip transition to anything but the MPC 74xx series.
nek said:
If IBM is making a PowerPC version of the Power5 as I'm sure they are, then it will likely be designed as a 90nm part, eliminating any problems in moving from 130nm to 90nm as they are having with the PPC970.
In a word, no.
At least one of the reported problems with the switch to 90nm is the usage of SSOI and the failure of a layer-bonding agent at operating temperatures in the completed chips. If IBM sticks with the older SOI technology and avoids their tendency to integrate new materials, then you might be right that a Power5 would leap in without issue. At this point, the problem is in fabbing, not the design.
I realize that IBM's problems with 90nm likely go beyond this issue, but it seems quite possible that there will be a sudden and large increase in speed in the Powermacs this summer.
This I agree with. They've been ironing out the issues, and if anyone in the industry has the money to throw at a problem like this, it's IBM. Odds are they're patenting as they go, too.
This will then allow the iMac to get a G5 a couple months later. Or the iMac and Powermac could get speed increases at the same time since the Powermac will be using a newer 3+GHz 97x and the iMac will get the current 2GHz 970FX. I think the PowerBook will begin using the G5 in january.
Hardly. There are issues besides the chip that will need to be solved before the G5 can go into an iMac that's even remotely the formfactor that's currently in use. Maybe it will end up being resolved in a some unforseen way, maybe it's possible that Apple's got something up their sleeves and I'm about to eat my words about the heat being too great when they release a 2.0ghz iMac in the gumdrop white machines.
If motorola/freescale actually create their e700 within the next year, then the iBook and eMac will likely use that processor. Or they might move to the e600 first, then the e700.
The e700 looks quite a bit more appealing than the current crop of 970s, and unless there's some revolutionary addition of power management and bus slewing that cuts the overall system down on usage, the e-series could prove to be a better line to follow with portables. A pin-compatible enhanced G4 that loses the restricting bus, has on-die memory, at least 1MB of L2 cacde, and advanced power control, staging over 2.0ghz, all under 20watts of heat at peak operation? Where do I sign up, since it's unlikely that the 2.0ghz G5s are going to get under 20watts?
