Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
PPC 970 in July?
- Thread starter MacRumors
- Start date
- Sort by reaction score
evoluzione
macrumors 68020
Re: USB 2.0
what's the big deal with USB 2?? now Apple has released FireWire800 I see no reason to NOT have USB2. It was never going to happen before the FW800 as it was a competitor to FW400. So, on all the new Macs with FW800, I would expect to see USB2 as well. (although that doesn't explain why it isn't on the 17" AiBook) 😕
what's the big deal with USB 2?? now Apple has released FireWire800 I see no reason to NOT have USB2. It was never going to happen before the FW800 as it was a competitor to FW400. So, on all the new Macs with FW800, I would expect to see USB2 as well. (although that doesn't explain why it isn't on the 17" AiBook) 😕
MacCoaster
macrumors 6502a
Re: Re: PPC 970 in July?
People around the boards have claimed that the 970 wouldn't take FULL advantage of the 900 MHz bus, probably just 800 MHz because of latency and other stuff, but I know nothing about technical memory stuff.
Also, from what I understand, the PowerPC 970 has a variable FSB. At 1.8 GHz, it's 900 MHz, but at 1.4 GHz, it's 700 MHz. So, if they're going to debut at 2.5 GHz straight up, that'd be 1.25 GHz memory bandwidth.
From what I understand about DDR400, you can have a dual channel DDR400, and therefore effectively 800 MHz
People around the boards have claimed that the 970 wouldn't take FULL advantage of the 900 MHz bus, probably just 800 MHz because of latency and other stuff, but I know nothing about technical memory stuff.
Also, from what I understand, the PowerPC 970 has a variable FSB. At 1.8 GHz, it's 900 MHz, but at 1.4 GHz, it's 700 MHz. So, if they're going to debut at 2.5 GHz straight up, that'd be 1.25 GHz memory bandwidth.
Hmmm...
More exciting than the July release would be the future processor plans. A 4+ Ghz Power5 derived CPU would seriously outpace anything that Intel has planned (i.e. continually cranking the clock on the P4 core) for the next couple years.
But I honestly don't believe any of it (well, conceivably the July release is true, though even that seems too early to be credible). MacBidouille also just posted a rumor that the 17" Powerbook will use the MPC 7457. But the MPC 7457 has 512k of L2 cache (according to published Motorola docs) and the 17" PB has only 256k of L2 cache (according to published Apple specs). So the MPC 7457 rumor is also surely false, as the 17" Powerbooks almost surely use the MPC 7455, just like all the other Powerbooks. Likewise, back in November, MacBidouille said that Apple would be using 64 bit AMD chips any day now. Suffice to say, the probability of that happening is exceedingly low. I have no inside information whatsoever, but based on common sense alone I would gladly bet anyone that Apple will not be selling PowerMacs with chips called "Athlon-64" or "Opteron" in them anytime during 2003.
In short, the MacBidouille site lacks credibility. I believe they had some good predictions earlier in 2002, but their accuracy seems to have really gone down in the last few months.
More exciting than the July release would be the future processor plans. A 4+ Ghz Power5 derived CPU would seriously outpace anything that Intel has planned (i.e. continually cranking the clock on the P4 core) for the next couple years.
But I honestly don't believe any of it (well, conceivably the July release is true, though even that seems too early to be credible). MacBidouille also just posted a rumor that the 17" Powerbook will use the MPC 7457. But the MPC 7457 has 512k of L2 cache (according to published Motorola docs) and the 17" PB has only 256k of L2 cache (according to published Apple specs). So the MPC 7457 rumor is also surely false, as the 17" Powerbooks almost surely use the MPC 7455, just like all the other Powerbooks. Likewise, back in November, MacBidouille said that Apple would be using 64 bit AMD chips any day now. Suffice to say, the probability of that happening is exceedingly low. I have no inside information whatsoever, but based on common sense alone I would gladly bet anyone that Apple will not be selling PowerMacs with chips called "Athlon-64" or "Opteron" in them anytime during 2003.
In short, the MacBidouille site lacks credibility. I believe they had some good predictions earlier in 2002, but their accuracy seems to have really gone down in the last few months.
Dont Hurt Me
macrumors 603
Re: Re: Re: PPC 970 in July?
I Strongly doubt we will see a 2.5 anytime soon. They just wanted to see how far they could go with it more then likely or a misprint. Funny how that page was pulled according to the front page of this site. i expect a summer announcement followed with the systems on the shelf by october. This could mean maybe a 1.2 g4 powermac, 1.4 970 powermac, 1.8 970 powermac. If so i would not expect market share to jump to 5%. Now if they come out and say we have the best cpu in the world and we love it so much we are sticking it in all the new imacs,the powermacs and powerbooks THAT could get them to the 5% marketshare.
Frobozz
macrumors demi-god
Re: Re: USB 2.0
There's no reason to have USB 2.0 on a Mac. You have firewire 800, which is much faster. USB is for connecting non-storage or flash-storage peripherals only. The wintel world tried ot make it compete with Firewire by beefing up it's specs. USB will only be included on a Mac if Apple sees an reason to support peripherals that can not be used on USB 1.1. Since this won't be CDRW drives, Hard drives, etc., I don't see the point. It would potentially make it a competitor to their own Firewire standard and, in turn, make USB 2.0 more accepted since it will work on most new PC's and Macs. We don't want this. It's not as good as firewire!
There's no reason to have USB 2.0 on a Mac. You have firewire 800, which is much faster. USB is for connecting non-storage or flash-storage peripherals only. The wintel world tried ot make it compete with Firewire by beefing up it's specs. USB will only be included on a Mac if Apple sees an reason to support peripherals that can not be used on USB 1.1. Since this won't be CDRW drives, Hard drives, etc., I don't see the point. It would potentially make it a competitor to their own Firewire standard and, in turn, make USB 2.0 more accepted since it will work on most new PC's and Macs. We don't want this. It's not as good as firewire!
Frobozz
macrumors demi-god
Re: Re: Re: Re: PPC 970 in July?
2 seperate sources, one being the chipmaker it's self (IBM), has stated they have been able to produce quantities of chips at 2.5 GHz. Now, Apple may not crank out the first revision at 2.5 GHz, but they thought the previous top of the line would be closer to 1.8 or 2.0 GHz. Given this, I think they will throw in a 2 GHz+ chip in the first revision. IBM does not have production problems like Motorola.
Unlike most people, I do believe Apple will intro the new machines in July, even if they don't ship until September. What other large annoucements will they have for MWNY? Software won't do it again, unless it was amazing stuff. And, yes, Apple will be at MWNY 2003. Sheesh. 🙂
2 seperate sources, one being the chipmaker it's self (IBM), has stated they have been able to produce quantities of chips at 2.5 GHz. Now, Apple may not crank out the first revision at 2.5 GHz, but they thought the previous top of the line would be closer to 1.8 or 2.0 GHz. Given this, I think they will throw in a 2 GHz+ chip in the first revision. IBM does not have production problems like Motorola.
Unlike most people, I do believe Apple will intro the new machines in July, even if they don't ship until September. What other large annoucements will they have for MWNY? Software won't do it again, unless it was amazing stuff. And, yes, Apple will be at MWNY 2003. Sheesh. 🙂
GeneR
macrumors 6502a
I'll believe it when I see it.
I want to believe that the speed bump to up to 4.5GHz will be available for the Mac, however, I haven't read anything here so far that confirms that the 980 is also for the Mac.
So, unless they say otherwise, I'm just going to assume that the 970 is the next big thing and the top speed is still 2.5.
However, I really hope I'm wrong. 😀
I want to believe that the speed bump to up to 4.5GHz will be available for the Mac, however, I haven't read anything here so far that confirms that the 980 is also for the Mac.
So, unless they say otherwise, I'm just going to assume that the 970 is the next big thing and the top speed is still 2.5.
However, I really hope I'm wrong. 😀
I've seen quite a few camera companies and photo storage makers coming out with USB2 and no Firewire. It would behoove Apple to offer it - whether they want to or not. It's being adopted and Mac users are going to miss out on any devices that do come out supporting it. I mean, why not have options right? We all know Firewire is better - but might as well give us both.
Moreover, the prosumer (and below) cameras are nearly universally USB. USB2 is necessary for digicam support now and increasingly in the future. (I still would buy a firewire digicam, but they're only available as pro models.)
Cheese
macrumors 6502
July? Of what year? Who knows what is up with the next big silicon 5v brain? As long as it doesn't try to speak french it should be all good and stuff. (LOL) From my perspective, it seems that as processors got faster, programs and their creators simply got fatter.. the dumbing down of our computer society. Remeber when a IIe was fun because we found out things that we could make it do that no one else had yet discovered? As things get more complicated, they have the appearance of being simpler, only to leave the firey question... what will you do if you had no choice but to go back to the way it used to be done? Does anyone even remember what we called the way it used to be done? We have made great progress, and we will continue to make strides towards faster, bigger, and better as long as we can, but, are there not people who think that there is more that can be done with a G3? isn't 400 MHZ enough? We don't all have Fortran running highly sophisticated calculations that take more than a week to solve, and although I sympathize with those that do, I wonder if there are yet many possibilities that can be found within the capability of lesser than top shelf processors. I still say bring on the 970 and it's baby sister or brothe too!
dongmin
macrumors 68000
2.0 ghz in July (introduced, not shipping) would be awesome. I was gonna say it's too optimistic--it would represent a 40%+ jump in 6 months--but given hints that IBM will introduce the 970 at up to 2.5 ghz in their blades, maybe 2.0 is actually conservative.
The big question for me is whether Apple will be bringing out dualies with the new chip. The performance gain will be significant enough that they might not have to, topping out with a single 2.0 ghz 970. But then again, maybe to counter the market perception, they'll bring out a dual 1.8 ghz system as the ultimate machine.
Or maybe, following AMD's lead, they'll introduce some totally new numbering convention, like G5 2000, the number referring to their combined SPEC scores. The new chips raise some interesting marketing questions.
The big question for me is whether Apple will be bringing out dualies with the new chip. The performance gain will be significant enough that they might not have to, topping out with a single 2.0 ghz 970. But then again, maybe to counter the market perception, they'll bring out a dual 1.8 ghz system as the ultimate machine.
Or maybe, following AMD's lead, they'll introduce some totally new numbering convention, like G5 2000, the number referring to their combined SPEC scores. The new chips raise some interesting marketing questions.
I can see the powermac line up as being one of the 3 things if apple do use the PPC970 in a range of macs this year. It probably just depends on if they actually have a motherboard that works with them and if the higher clocked chips are available in high enough quantities to allow dual models.
a repeat of the digital audio line up
1.2 Ghz single
1.6 Ghz single (dual as BTO option)
1.8 Ghz single
2 Ghz single
a repeat of the first Quicksilver G4 line up
1.4Ghz single
1.8Ghz single
1.6Ghz dual
or maybe just the current one just speed bumped
1.4Ghz single
1.6Ghz dual
2Ghz dual
I just hope we do see significant speed increases similar to the difference between the final sawtooth lineup and the digital audio lineup. Jumping from 500Mhz / 100Mhz FSB to 733Mhz / 133Mhz FSB was signficant enough to make it appear to be way faster. Going from dual 1.42Ghz / 167Mhz FSB to 2GHz / 800Mhz FSB would be enough to make a lot of people very impressed. When you concider how the current Athlon systems are only around 2Ghz with a 167Mhz x 2 FSB but stand up very well against the far higher clocked P4 chips with a 133 x 4 FSB.
a repeat of the digital audio line up
1.2 Ghz single
1.6 Ghz single (dual as BTO option)
1.8 Ghz single
2 Ghz single
a repeat of the first Quicksilver G4 line up
1.4Ghz single
1.8Ghz single
1.6Ghz dual
or maybe just the current one just speed bumped
1.4Ghz single
1.6Ghz dual
2Ghz dual
I just hope we do see significant speed increases similar to the difference between the final sawtooth lineup and the digital audio lineup. Jumping from 500Mhz / 100Mhz FSB to 733Mhz / 133Mhz FSB was signficant enough to make it appear to be way faster. Going from dual 1.42Ghz / 167Mhz FSB to 2GHz / 800Mhz FSB would be enough to make a lot of people very impressed. When you concider how the current Athlon systems are only around 2Ghz with a 167Mhz x 2 FSB but stand up very well against the far higher clocked P4 chips with a 133 x 4 FSB.
Re: Re: PPC 970 in July?
RDRAM (RamBus) is 16bits wide. Dual channel RDRAM 1066 has a max bandwidth of 33Gbit or a bit over 4GB/sec (1066x16x2/1024/8)
DDR DIMMS are 64bits wide. Dual Channel DDR 333 has a max bandwidth of 41 Gbit or about 5.2 GB/sec. (333x64x2/1024/8)
RamBus has more latency also so it takes more ticks to get data in or out.
Rambus costs more, to add insult to injury.
rambus sucks. DDR is the way to go. I've got Rambus PCs in my office. they cost a [relative] fortune to upgrade because they are all dual channel and the memory is too expensive.
RDRAM (RamBus) is 16bits wide. Dual channel RDRAM 1066 has a max bandwidth of 33Gbit or a bit over 4GB/sec (1066x16x2/1024/8)
DDR DIMMS are 64bits wide. Dual Channel DDR 333 has a max bandwidth of 41 Gbit or about 5.2 GB/sec. (333x64x2/1024/8)
RamBus has more latency also so it takes more ticks to get data in or out.
Rambus costs more, to add insult to injury.
rambus sucks. DDR is the way to go. I've got Rambus PCs in my office. they cost a [relative] fortune to upgrade because they are all dual channel and the memory is too expensive.
Analog Kid
macrumors G3
Power5:
Any thoughts on whether Apple might try to differentiate the Xserve line from the PowerMac line by debuting higher end processors (such as the Power5 derivatives, or full blown Power4s) in the Xserve before they filter down to the PowerMacs?
USB2:
USB is slightly cheaper because FW requires every peripheral to be capable of acting as a host.
USB2 is a lousy standard, and USB-on-the-go is just making matters worse. A reasonable industry would have stuck with USB1.1 for keyboards and mice and FW for dataheavy peripherals where the higher margins would dwarf the incremental silicon costs.
All of that said, I want to be able to use the newer devices even if they make the mistake of going with USB2, so Apple's going to have to support it.
GHz and Marketing:
It seems to me that Apple tries to keep a steady increase in CPU power rather than pushing the technology at every opportunity. I wonder if the marketing analysis tells them that a sudden jump in CPU performance won't bring in any more profit than ensuring steady updates over the long haul.
Any thoughts on whether Apple might try to differentiate the Xserve line from the PowerMac line by debuting higher end processors (such as the Power5 derivatives, or full blown Power4s) in the Xserve before they filter down to the PowerMacs?
USB2:
USB is slightly cheaper because FW requires every peripheral to be capable of acting as a host.
USB2 is a lousy standard, and USB-on-the-go is just making matters worse. A reasonable industry would have stuck with USB1.1 for keyboards and mice and FW for dataheavy peripherals where the higher margins would dwarf the incremental silicon costs.
All of that said, I want to be able to use the newer devices even if they make the mistake of going with USB2, so Apple's going to have to support it.
GHz and Marketing:
It seems to me that Apple tries to keep a steady increase in CPU power rather than pushing the technology at every opportunity. I wonder if the marketing analysis tells them that a sudden jump in CPU performance won't bring in any more profit than ensuring steady updates over the long haul.
I think it's half ensuring steady updates and half that they're at the mercy of Motorola.
I imagine the year the 733Mhz G4 came out that the sudden 47% leap in clockspeed wasn't the most they could have managed and they were holding back on the 867Mhz G4 so they had a definate MHz leap available for the next range.
The next range was an even bigger leap comparatively. The entry level model jumped from 466 to 733Mhz, a massive 57% increase for anyone who's priced out of having a higher end system. The dual model in the range had a 50% Mhz increase going from 533Mhz to 800Mhz aswell. This all looked very impressive on paper, especially for the prices but the G4 itself only really got a small increase in clockspeed with the fastest CPU only being 18% (867Mhz vs 733Mhz) than the previous fastest CPU.
I know Mhz isn't everything, this is proved by the fact a dual 1Ghz G4 isn't anywhere near twice the cpu power of a dual 500Mhz G4 and neither is a dual 1.25Ghz G4, only a change in CPU type can change that, without the bandwidth it needs the G4 is a pretty lame CPU, even increasing the L2 to 512K and adding a faster FSB would probably only boost things up a little.
Going from the original G4 to the current one we've only seen a few advancements. Even the big ones don't do enough to push the realworld performance up enough over the older models, let alone recent Athlon or P4 based PCs.
The Original G4 vs The Current G4
100Mhz FSB vs 167Mhz FSB 🙂
4 stage pipeline vs 7 stage pipeline 😡
1Mb backside 2:1 L2 vs 256K onchip 1:1 L2 😡
No need for a L3 vs 1 or 2Mb L3 😡
4 IPC vs 5 IPC 🙂
533Mhz vs 1.42Ghz 🙂
Dont Hurt Me
macrumors 603
Nice post barkmonster, one thing to point out though is that apple knew at the time that motorola wasnt going anywhere and still have not. So they had to get the most out of the little bumps. on the other hand here is the 970 which is scaling up even in testing. I have 1 of those 733's you are talking about. no l2. this thing could run at 733,800,867 just by telling the pll what to do. If apple does release the 970 this thing will scale up fast or apple could surprise everyone and introduce the higher clocks. It will boil down to what these chips cost apple, and to how many computers they think the market will take. Still find it interesting that ibm releases this post just to yank it back. intentional or mistake?
dongmin
macrumors 68000
I agree. But we're not running Apple. Who knows what the braintrust at Apple are up to?
It seems that IBM is having better-than-expected results with yields. And that the 970/980/990 line has more headroom to scale than the G4. Given those factors, I would push the cpu speed as much as it's cost effective.
Does anyone have any solid figures on how much the 970s cost, and how it compares to the G4 7457? I read some speculation that it's cheaper b/c of the smaller dye size but I'm not so sure.
As for the leaked IBM press release, I say it's probably just the case of one IBM group (the blade group) not talking to another group (the ppc group) and not knowing what's kosher to announce to the public.
how exactly do you figure that a dual 1GHz or a dual 1.25 GHz isn't twice the computer of a dual 500MHz? They are different chips. Altivec has been expanded, IPC expanded, Cache expanded. Having more steps in a pipe doesn't always mean a decrease in performance either... RISC instructions are supposed to execute in fewer steps than CISC, but many of them require multiple steps. Longer pipes have advantages other than just tending to raise clock speed. Their main detraction is that you have to backup more when a processor blows a branch prediction.
I think you should back up a blanket statement like that with some figures IMHO.
don't forget the high speed L3 bandwidth also
as stated, this isn't always a problem.
this is a mis-representation. The comparison is actually 512k L2 at half proc speed vs. 256K on-die L2 cache (full speed) AND 1-2 MB of L3 cache on dedicated high speed bus.
the first G4s didn't have on die L2 and they were offered in 512K to 1024K backside L2 configs. All major consumer processors moved from larger off die L2 caches to smaller on die caches at around the same time [athlon, P3, G4]. They benefited from reduced cost AND increased performance. The data in L2 needs fast access over size in most cases. The current L3 takes the place of your beloved 1MB off die L2... so by your reasoning, the current machines should be better in every way, right?
This doesn't make any sense. That's like saying a newer sports car model sucked because the new one 'needs' a supercharger. L3 cache is a feature... It's not a matter of the previous processor not needing a L3. Alphas running at around 500MHz have enormous L3 caches because they increase performance.
July
If it is July, it won't be a minute too soon. I've been waiting a while for this 'til I get my next mac and I guess I'll jsut have to wait with outdated hardware until it comes.
If it is July, it won't be a minute too soon. I've been waiting a while for this 'til I get my next mac and I guess I'll jsut have to wait with outdated hardware until it comes.