Crolles already at 90 nm
http://siliconstrategies.com/article...equestid=73972
How soon do you think we will see the new Freescale chips
at this size? anyone? Mr. Macphisto whats your take on this news?
exciting times over at Freescale e600 e700 in macs by the end of this
year?
\ 
How soon do you think we will see the new Freescale chips
at this size? anyone? Mr. Macphisto whats your take on this news?
exciting times over at Freescale e600 e700 in macs by the end of this
year?
\
Comments
Originally posted by ghstmars
http://siliconstrategies.com/article...equestid=73972
It looks like registration is needed. Could you post here what you have read?
" Royal Philips Electronics Monday (May 10) claimed it has produced a 90-nm CMOS device within two separate fabs, including within Crolles2 and Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC).
The Philips device, dubbed the CMOS090LP, is a system-on-chip solution, based on a low-leakage, 90-nm CMOS process. This silicon integrates Arm Holdings plc's processor core with SRAM, ROM and analog signal circuitry for use in wireless applications.
The device came from the Crolles2 fab in Crolles, France and at TSMC. Partners in the 300-mm plant in Crolles are Philips, Motorola and STMicroelectronics.
"In addition to achieving right-first-time silicon, one of the most satisfying results is that we are seeing consistent performance between devices produced in the Crolles2 fab and those that have come from the TSMC fab," said Jan-Marc Luchies, 90-nm CMOS program manager at Philips Semiconductors, in a statement. "This means that the early work done by the Crolles2 partners and TSMC in aligning the processes at the two fabs, both in terms of design rules and electrical parameters, has really paid off."
It's good to hear that at least one fab (unlike IBM, AMD and Intel) has got through the minaturisation process without facing significant delays. It does tend to suggest that the fab is ready - the main variable is in Moto's ability to develop the SoC process to incorporate the G4 core (for the e600). The timetable is up in the air. To speculate, if Freescale is to be floated in three or four months, what better way to stimulate investor interest than to release the SoC 'e'-line of chips to developers/consumers
A couple of questions though. How does this share-fab setup work? Who gets to use the fab when? Is it done on a rotation basis or on the partner who has the highest demand to satisfy? You could be left with the possibility that history will repeat itself for Moto; a decent processor for the time (as in 99/00) but an acute inability to satify demand. You would hope Moto has factored this into their analysis.
While promising, I can't get too excited anticipating any RSN MPC 74XX processors. Hope Freescale surprises me though.
That said, the fact the the 90nm fab is ready to go is good news if FreeScale has got a good design that is ready. That may mean upgraded laptops for Apple come September or October - and I do think that the e600 will be around by then.
IBM has gotten the kinks worked out on 90nm. I'm not too sure the 970FX is meant for the PowerMac though. I've not heard if Apple has any more orders in for faster versions. I still tend to lean on the Power5 derivative finding its way into the next PowerMac, though the 970FX may go elsewhere.
Originally posted by Mr. MacPhisto
IBM has gotten the kinks worked out on 90nm. I'm not too sure the 970FX is meant for the PowerMac though. I've not heard if Apple has any more orders in for faster versions. I still tend to lean on the Power5 derivative finding its way into the next PowerMac, though the 970FX may go elsewhere.
975 based Power Macs at up to 3GHz and 970fx based iMacs at up to 2.0Ghz at WWDC? Pretty please?
Originally posted by KANE
You'd think this thread would create more buzz than it actually did. Has everyone already given up on Freescale?
It is not that, but, as a j stev and Mr. MacPhisto pointed out, we cannot tell nothing more than pure speculation words, and this on a not very solid basis. Until we hear a day that finally Freescale delivers the e600-e700 processors, we can only hope that they are on schedule, although seemingly things go right.
Look at the example of 7447A. From somewhere last year until its announcement this year, the G4 was considered as a lost affair, a dead horse, that Apple should abandon the soonest possible. Then suddenly, the 7447A variant made its appearance and people of course could not believe that Apple was going to use this processor anywhere in its product lines (except perhaps the eMac and iBook, and temporarily), notwithstanding the evidence that Apple had no other options. Now Apple has this 7447A in its Powerbook line and people are happy with this (well, there are some exceptions
I believe (or want to hope) the same will happen with the e600 chips. But this time it will be a much nicer chip, capable to go dual core, eventually (in a 17" Powerbook), and with system integration on it. If the rumors that the 750VX has been abandoned hold any truth, then this would be a *confirmation* that something strong is coming for Apple's low power machines. And according to current evidence, it should be from Freescale.
Crolles 2 is the real thing. If they're up to SoC designs with CPUs (even if they're ARM CPUs) then the capability to fab a G4-scale CPU reliably shouldn't be far off.
And if they've got this partnership with TSMC, then they aren't limited by Crolles' volume, either.
So far everything about this partnership seems to be well executed. Time will tell, but I think Freescale will make a real comeback this summer.
I've seen all sorts of explanaitons as to why Motorola has failed so badly, everything from dirty fabs to poaching by Intel with respect to the engineering staff. Frankly all BS from my standpoint, Motorolas problems are management related, Until we see signs that Freescale has been fully and completely purged of these mindless bitches we shuold not expect much from Freescale as a company.
I always look at things this way: would I invest my hard earned money in Freescale stock? Not at the moment, I would need a signifcant indication that the management practices that so damaged Motorola have been left behind.
As to buzz, well I certianly do hope that Freescale does well with their SoC focus. Frankly that is the way most single core processors will be manufactured in the future. In simply produces more economical systems. With in a year to a year and a half the rest of the micro processor world will switch over to dual core SMP products. So Freescale does seem to have a road map that will keep it successful in a number of sectors. How that works out with Apple remains to be seen.
Dave
Originally posted by KANE
You'd think this thread would create more buzz than it actually did. Has everyone already given up on Freescale?
Originally posted by wizard69
Given up no. Having low expectation related to anything that involves Motorola yes.
I've seen all sorts of explanaitons as to why Motorola has failed so badly, everything from dirty fabs to poaching by Intel with respect to the engineering staff. Frankly all BS from my standpoint, Motorolas problems are management related
Where do you get the idea that dirty fabs and lost talent are because of anything other than incompetent management?
Bad management has consequences, you know. The attention to dirty fabs was necessary because people were blaming the G4's yields and performance on bad designs.
I always look at things this way: would I invest my hard earned money in Freescale stock? Not at the moment, I would need a signifcant indication that the management practices that so damaged Motorola have been left behind.
Their last quarter looked a lot like a turnaround - who'd have guessed that Motorola would be eating Nokia's lunch so quickly?. And a friend of mine who works in satellite communications went to a big industry meeting that included a Mot employee (1, not 12, like in the bad old days), and he remarked that if the rest of the company was like that one guy, they were going places.
The point of that anecdote, actually, is that Mot only sent one engineer, and they sent an engineer with his ducks in a row, and he was free enough of adminispheric pollution to be effective. This alone signals a significant - and favorable - change in management style.
The crucible is this summer. Freescale has been upbeat and confident about what's coming this summer. If 9/21 hits and their flagship is still the 7447A at 1.42GHz on 130nm, then we'll know that not enough has changed. Until then, though, the signs are promising.
As to buzz, well I certianly do hope that Freescale does well with their SoC focus. Frankly that is the way most single core processors will be manufactured in the future. In simply produces more economical systems.
Don't overlook the benefits of SoC designs to dual core CPUs. It's a lot cheaper and more efficient to have all the support logic right on board with the cores and the memory controller than to run it all across a hot, expensive, difficult-to-engineer system bus.
Then there's Cell, and who knows what that is? But from what few trickles of information we have, it sounds like it needs the kind of dense, high-bandwidth, low-latency fabric that SoC can provide to be effective at all.
Originally posted by Amorph
. . . Then there's Cell, and who knows what that is? But from what few trickles of information we have, it sounds like it needs the kind of dense, high-bandwidth, low-latency fabric that SoC can provide to be effective at all.
Back on another thread, Mr. MacPhisto mentioned that IBM is continuing to work on a SoC design for Apple, but had cancelled the VX or Mojave chip. Since the Cell is something IBM and Sony have been working on, it suggests some possibilities. Just a guess, but maybe the IBM design is some variation on Cell, while Freescale is simply putting conventional functions on a chip, like memory controller.
Originally posted by Amorph
Where do you get the idea that dirty fabs and lost talent are because of anything other than incompetent management?
Bad management has consequences, you know. The attention to dirty fabs was necessary because people were blaming the G4's yields and performance on bad designs.
Apparently I didn't express my thoughts well. You have done a much better job of it. Motorola semiconductor divisions problems have been management related which resulted in some of the problems mentioned. Frankly it is not a problem that is Motorola/Freescales alone, and is something that deserves more public focus.
Quote:
Their last quarter looked a lot like a turnaround - who'd have guessed that Motorola would be eating Nokia's lunch so quickly?. And a friend of mine who works in satellite communications went to a big industry meeting that included a Mot employee (1, not 12, like in the bad old days), and he remarked that if the rest of the company was like that one guy, they were going places.
Yes but that is not Freescale.
It is good to see that some of the company has changed for the better but I don't want to jump on the band wagon yet and say the worst day are past for the semiconductor business.
Quote:
The crucible is this summer. Freescale has been upbeat and confident about what's coming this summer. If 9/21 hits and their flagship is still the 7447A at 1.42GHz on 130nm, then we'll know that not enough has changed.
What they have promised on their web site is very nice and certainly will appeal to a large part of their customer base. As we have seen before though there is the issue of execution and the abiltiy to atleast stay with the pack.
Quote:
Don't overlook the benefits of SoC designs to dual core CPUs. It's a lot cheaper and more efficient to have all the support logic right on board with the cores and the memory controller than to run it all across a hot, expensive, difficult-to-engineer system bus.
I thought I was endorsing SOC fairly well. What I see happening in the future is the high performance end of the business going to dual core and more technology. Where performance is not paramount SOC will be at the forefront. Initially I see SOC being delivered with single cores but that could easily change to SMP, with the foucs remaining at the low end.
That low end may be well above what we consider low end today. At the high end though I do not see huge integration. Memory interfaces and the like yes but I would not expect a complete peripherial package at the high end. What Motorola suggest is coming would result in a complete Mac on a chip excluding the GPU and memory chips. That could be fantastic for the portable lines and lower cost machines.
Quote:
Then there's Cell, and who knows what that is? But from what few trickles of information we have, it sounds like it needs the kind of dense, high-bandwidth, low-latency fabric that SoC can provide to be effective at all.