ArsTechnica's HUGE article on PowerPC 970, Part II

[Eug]

Here. :wine:

I want my Apple 15" AlumaBook with PPC 970 NOW!

 

[Eug]

Hmmm... He keeps saying that Apple is to blame for a bad memory bus design in their mobos.

AFAIK (which isn't much), the blame is with the G4 chip, not Apple. It is an innate limitation of the CPU.

 

[EdipisReks]

if you read the discussion over at ars about the review, you'll find your view about who is at fault for the G4 bus to not be totally accurate. personally, i can't wait for 970 based powerbooks.

 

[zephyrprime]

The 970 is really interesting. So what's the deal with motorola now? Have they just given up on processors to focus on cell phones or something?

 

[EdipisReks]

Originally posted by: zephyrprime
The 970 is really interesting. So what's the deal with motorola now? Have they just given up on processors to focus on cell phones or something?

motorola has always been mainly an embedded CPU company. it appears that the R&D necessary to keep Apple supplied with decent chips has outstriped the profit margins as far as Motorola is concerned. the G5 was canned quit a while ago.

 

[Eug]

Originally posted by: EdipisReks
if you read the discussion over at ars about the review, you'll find your view about who is at fault for the G4 bus to not be totally accurate. personally, i can't wait for 970 based powerbooks.

Well, I went thru that Ars thread, and the only thing really mentioned was:

"Thus, Apple has been forced to come up with their own memory controller for the G4, lest they be stuck with the previous generation's terrible SDR stuff. At worst their offerings are a pointless marketeing-inspired hack, and at best they're a brilliant stopgap measure. IN any case, Apple is trying their best to make do with motorola's offerings, but the point remains that Apple, not motorola, designed the current setup."

and

"Narrowly speaking, true. But in CPUs, as in all other things ... you get what you pay for... IF you are a savvy buyer.

Grafting a faster FSB onto the G4+ core would not have been unreasonably hard... if Apple wanted to pay for it. A big part of the problem is that APPLE didn't give a rip about memory bandwidth back in the days when the MPXbus wasn't the limiting factor... Apple was always late with better memory bandwidth ... and by the time the situation got really, really embarrassing Apple needed more than just a G4+ with some more bandwidth, it needed a whole new CPU architecturally... sooooo ..."

What that suggest to me is that my contention that the current G4 cannot (properly) make use of a DDR bus is still valid.

It would take a new chip design. That chip would be 7457-RM, but alas, it does not yet exist. It sounds like from the above argument is that Apple simply didn't pay Motorola enough to make that design change earlier. Given Motorola's focus on the embedded market and track record with Apple, I can understand why Apple would balk at this.

 

[Eug]

IBM unveils new-chip details (May 15, 2003)

Nothing really, new, but now it's from the horse's mouth.