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 Apple Motherboards?
- Thread starter MacRumors
- Start date
- Sort by reaction score
I am curious as to whether most people here would rather see dual 970's or a price drop in the powermac line with single 970's only. Not that either will happen, but I am just curious as to how much people really want that second processor. If the priceing/lineup stayed the same with a single for the lowend I think I might just go ahead and buy the single 970 for the 1500 it cost.
Look again:
Page 1: Post by nuckinfutz "1. A Memory Bandwidth of 7.2Gbps(6.4 realized)
Page 3: Post by eric_n_dfw "From the spec's about the 970 (900Mhz FSB!!!!) that thing should shut up even the staunchest Apple critics"
You lose, thanks for playing.
Re: What are we talking...
Here's some info
The PPC 970 is a "8" Superscaler chip as opposed to the G4's "4".
PPC 970= 8 Instructions 5 dispatch
G4= 4 Instructions 3 dispatch(IIRC)
This simply means even at a base leve the PPC 970 can dispatch 66% more date per clock cycle.
Memory subsystems are crucial to feeding the various components of a computer
The G4 supports a Front Side Bus of 1.3Gbps(Gigabits per second). This is the maximum speed at which the processor directly access the memory.
The PPC 970 supports 7.2Gbps(6.4Gbps when overhead is accounted for). This 4x the throughput of the G4. This will make Altivec and memory bound applications simply fly.
Next we have L2 ondie cache.
PPC 970= 512k G4= 256k
Die Size
G4 .18 Micron= 106mm squared
PPC 970 .13Micron= 121mm squared
This is less than %15 between the two meaning that the PPC 970 should be damn near equal to the G4 in production costs. If IBM produces high yield rates the PPC 970 could be cheaper.
The 64bit part won't affect a majority of consumers. High End apps will benefit from 64bit code and processing but you won't see much improvement at the consumer level IMO.
Hell I say Apple should do both. Make the low end Powermac a single 1.6Ghz and then the midrange could use Dual 1.6Ghz and top would be Dual 1.8Gz.
Apple needs to stay dual. It's an advantage over Intel right now(as far as Pentium4) Multitasking is very important. and buy keeping Dual Processor based system constantly available you keep your Developers focused on making their software multithreaded to take advantage of Dual Proc systems. In the future you will see Dual Cores running multiple threads to the processor itself.
Now imagine this. What would be easier. Stuffing 4 seperate chips in a box or running a Single Processor with Dual Cores and again bisecting each core with Multithreading? That gives you the logical equivalent of a Quad Processor system yet you've only have to use one physical chip. This IS the future!
Here's some info
The PPC 970 is a "8" Superscaler chip as opposed to the G4's "4".
PPC 970= 8 Instructions 5 dispatch
G4= 4 Instructions 3 dispatch(IIRC)
This simply means even at a base leve the PPC 970 can dispatch 66% more date per clock cycle.
Memory subsystems are crucial to feeding the various components of a computer
The G4 supports a Front Side Bus of 1.3Gbps(Gigabits per second). This is the maximum speed at which the processor directly access the memory.
The PPC 970 supports 7.2Gbps(6.4Gbps when overhead is accounted for). This 4x the throughput of the G4. This will make Altivec and memory bound applications simply fly.
Next we have L2 ondie cache.
PPC 970= 512k G4= 256k
Die Size
G4 .18 Micron= 106mm squared
PPC 970 .13Micron= 121mm squared
This is less than %15 between the two meaning that the PPC 970 should be damn near equal to the G4 in production costs. If IBM produces high yield rates the PPC 970 could be cheaper.
The 64bit part won't affect a majority of consumers. High End apps will benefit from 64bit code and processing but you won't see much improvement at the consumer level IMO.
Hell I say Apple should do both. Make the low end Powermac a single 1.6Ghz and then the midrange could use Dual 1.6Ghz and top would be Dual 1.8Gz.
Apple needs to stay dual. It's an advantage over Intel right now(as far as Pentium4) Multitasking is very important. and buy keeping Dual Processor based system constantly available you keep your Developers focused on making their software multithreaded to take advantage of Dual Proc systems. In the future you will see Dual Cores running multiple threads to the processor itself.
Now imagine this. What would be easier. Stuffing 4 seperate chips in a box or running a Single Processor with Dual Cores and again bisecting each core with Multithreading? That gives you the logical equivalent of a Quad Processor system yet you've only have to use one physical chip. This IS the future!
Any one have an idea much are these 970s supposed to cost -- ballpark? Dramatically more or less than the G4 or about the same?
We're hearing this the 970 the same week the last of the CRT iMacs are finally de-listed. One of the great things about the original iMac was that it was positioned as a low cost machine. In my opinion, the low-cost lineup from Apple today is too under-powered.
I hope 2003 and 2004 bring some power-boosts to the low end as well as to the Power Macs. If we have to wait until 2005 to get an iMac or iBook that can really compete with Intel (perfomance, not design of course), I'm afraid that will be too late.
We're hearing this the 970 the same week the last of the CRT iMacs are finally de-listed. One of the great things about the original iMac was that it was positioned as a low cost machine. In my opinion, the low-cost lineup from Apple today is too under-powered.
I hope 2003 and 2004 bring some power-boosts to the low end as well as to the Power Macs. If we have to wait until 2005 to get an iMac or iBook that can really compete with Intel (perfomance, not design of course), I'm afraid that will be too late.
The problem is knowledge that the G4 is better optimized. The problem is software that isn't optimized. The G4 presently can outpace a PC 4 to 5 fold. Look at the MTOP numbers here:
http://forgetcomputers.com/~jdroz/09.html
PCs faster than Macs? Maybe in Adobe Premier, but that's because Adobe doesn't optimize Premier as well as they should for the Mac. Final Cut Pro is more optimized.
I'd put my Dual 867 up against a 3.06 P4 anyday...
... and watch my mac's ass get kicked.
Sorry gopher, but no way.
... and watch my mac's ass get kicked.
Sorry gopher, but no way.
Two Motherboard Prototypes
No one's mentioned this, so I'll raise the question. The rumor is stating that two new motherboards have gone out for bids, one for the XServe and one for the PowerMacs (that's new speculation I first noticed above). If this is the case, and since the original rumor stated that one of these boards was for a dual processor and one for a single processor configuration, doesn't that suggest that the dual is for the XServe and the single is for the PowerMac? Or, can the PowerMac use the XServe's motherboard in a dual CPU setup? If it can't, that means only single CPUs for the new PowerMacs.
No one's mentioned this, so I'll raise the question. The rumor is stating that two new motherboards have gone out for bids, one for the XServe and one for the PowerMacs (that's new speculation I first noticed above). If this is the case, and since the original rumor stated that one of these boards was for a dual processor and one for a single processor configuration, doesn't that suggest that the dual is for the XServe and the single is for the PowerMac? Or, can the PowerMac use the XServe's motherboard in a dual CPU setup? If it can't, that means only single CPUs for the new PowerMacs.
all I have to say...
All I have to say is that i'm 19, just started my first REAL job (been helping on the farm the last year) and despite having other things that I could save the money for, I would VERY seriously consider going out and buying a Powermac with one of these...a powermac with two of these...wet dream. Heck, i'd prolly lack anything to compare it to since i'm typing this on a PowerBase 180 with a 350mhz g3 upgrade.
"Chooooocccccckkkkllaaaaate." *drool* -Homer Simpson
"Power Peeeeceeeee nine-seventy." *drool* -Mac users of the world
All I have to say is that i'm 19, just started my first REAL job (been helping on the farm the last year) and despite having other things that I could save the money for, I would VERY seriously consider going out and buying a Powermac with one of these...a powermac with two of these...wet dream. Heck, i'd prolly lack anything to compare it to since i'm typing this on a PowerBase 180 with a 350mhz g3 upgrade.
"Chooooocccccckkkkllaaaaate." *drool* -Homer Simpson
"Power Peeeeceeeee nine-seventy." *drool* -Mac users of the world
I dont know about a 3.06 P4 (Isnt that the highest end P4 available?) but I know my DP 867 feels quite a bit faster than the Athlon 1700 at work. This is doing basic office work like Microsoft Office, Web browsing, Email etc.
Oh, and of course we have all heard the Photoshop stuff a million times, but my G4 is so much faster than the Athlon in PS its just sad...I wanna pull my hair out using PS on that thing.
I dont really know what im getting at...and its offtopic..but Im very happy with the cash I plunked down on this machine
Oh, and of course we have all heard the Photoshop stuff a million times, but my G4 is so much faster than the Athlon in PS its just sad...I wanna pull my hair out using PS on that thing.
I dont really know what im getting at...and its offtopic..but Im very happy with the cash I plunked down on this machine
Is a 3 GHz P4 really faster than a dual 1.42 G4?
I was looking at Barefeats the other day and saw that he has a test of the dual 1.42GHz G4 on his site.
He included a comparison with a 3 GHz P4, which is a lot faster. Until you notice that this P4 actually has a Radeon 9700 Pro on board, while the Mac has a GeForce4 Ti. Clicking through on the link to Anandtech, I read that a 9700 gives a 30%-70% speed increase over the GeForce4. Then suddenly the speed difference is not that impressive anymore, it could well be caused by just the difference in gfx cards alone.
I was looking at Barefeats the other day and saw that he has a test of the dual 1.42GHz G4 on his site.
He included a comparison with a 3 GHz P4, which is a lot faster. Until you notice that this P4 actually has a Radeon 9700 Pro on board, while the Mac has a GeForce4 Ti. Clicking through on the link to Anandtech, I read that a 9700 gives a 30%-70% speed increase over the GeForce4. Then suddenly the speed difference is not that impressive anymore, it could well be caused by just the difference in gfx cards alone.
no
I assume you are talking about this article linked on barefeets
that links to
http://www.digitalvideoediting.com/2002/11_nov/reviews/cw_macvspciii2.htm
No the speed of the video card does not account for the difference, These are processor heavy benchmarks, ie composites/renders. They have nothing to do with how quickly pixels are being rendered and/or pushed around in the framebuffer. Whats more The P4 tested was nearly 100% faster in most of the benchs, not 30%.
Just give it up, the g4 is slow, end of story, nomatter how overclocked it is. Your dollar buys you ALOT more performance if you spend it on a pc, indeed if you read afew more of his articles it would seem obvious that the athlons are the smart money.
To continue to believe otherwise is to wantonly ignore the facts. I understand the want, heck I even sympathise, but it's fantasy, desperate fantasy.
I assume you are talking about this article linked on barefeets
that links to
http://www.digitalvideoediting.com/2002/11_nov/reviews/cw_macvspciii2.htm
No the speed of the video card does not account for the difference, These are processor heavy benchmarks, ie composites/renders. They have nothing to do with how quickly pixels are being rendered and/or pushed around in the framebuffer. Whats more The P4 tested was nearly 100% faster in most of the benchs, not 30%.
Just give it up, the g4 is slow, end of story, nomatter how overclocked it is. Your dollar buys you ALOT more performance if you spend it on a pc, indeed if you read afew more of his articles it would seem obvious that the athlons are the smart money.
To continue to believe otherwise is to wantonly ignore the facts. I understand the want, heck I even sympathise, but it's fantasy, desperate fantasy.
Apple motherboards
hugh, and anybody else interested,go to ARS Technica there are some great articles explaining in great detail and comparing to intel hardware.
nuckinfutz, nice clarification, thankyou.
I would concur and add that Apple wouldn't recieve finished motherboards, they would go an assembler, who would finish the job (mount in cases, add processors, ram, etc, and ship to dealers or end-users in the case of Apple online purchasers.) Apple just does development and marketing. They contract out all the manufacturing.
hugh, and anybody else interested,go to ARS Technica there are some great articles explaining in great detail and comparing to intel hardware.
nuckinfutz, nice clarification, thankyou.
I would concur and add that Apple wouldn't recieve finished motherboards, they would go an assembler, who would finish the job (mount in cases, add processors, ram, etc, and ship to dealers or end-users in the case of Apple online purchasers.) Apple just does development and marketing. They contract out all the manufacturing.
Actually I meant this article:
http://www.barefeats.com/pm07.html
See the comparison on page 2 ( http://www.barefeats.com/pm07b.html )
In the Q3 test the dual 1.42 scores 291 fps, and the P4 3Ghz scores 366 fps, but has a Radeon 9700 Pro.
Anyway, I read the articles at Ars and I played around with FCE on a dual 1.25GHz PM last Saturday. It rendered the effects pretty fast. But then again, my reference is a 800MHz iMac and a 500MHz P3
The main problem as I see it, is that although Apple sells decently fast computers (but not extremely fast) that are fast enough for most things, they are (still) pretty expensive. And I guess that people who pay that kind of money want to know that they bought one of the fastest computers available.
So, if Apple would lower the prices some more, then suddenly their offering would become much more interesting.
Another thing, although the 3GHz P4 is out now, I haven't seen it in use anywhere yet. Most businesses still use old hardware and with the current market conditions are not likely to buy new stuff soon. That means that Apple's dual 1.42GHz might be outperformed by a P4 3GHz, it will still be a lot faster than most other machines around.
http://www.barefeats.com/pm07.html
See the comparison on page 2 ( http://www.barefeats.com/pm07b.html )
In the Q3 test the dual 1.42 scores 291 fps, and the P4 3Ghz scores 366 fps, but has a Radeon 9700 Pro.
Anyway, I read the articles at Ars and I played around with FCE on a dual 1.25GHz PM last Saturday. It rendered the effects pretty fast. But then again, my reference is a 800MHz iMac and a 500MHz P3
The main problem as I see it, is that although Apple sells decently fast computers (but not extremely fast) that are fast enough for most things, they are (still) pretty expensive. And I guess that people who pay that kind of money want to know that they bought one of the fastest computers available.
So, if Apple would lower the prices some more, then suddenly their offering would become much more interesting.
Another thing, although the 3GHz P4 is out now, I haven't seen it in use anywhere yet. Most businesses still use old hardware and with the current market conditions are not likely to buy new stuff soon. That means that Apple's dual 1.42GHz might be outperformed by a P4 3GHz, it will still be a lot faster than most other machines around.
Re: Re: What are we talking...
1.3GigaBYTES per send and 7.2GigaBYTES per second. Don't understate the bandwidth by almost an order of magnitude!
0.13um G4 has 512KB L2 cache, and is probably around 80 to 90 mm^2 in size.
Also the part of the 970 that bundles up instructions has some severe limitations that means that it is very rare that all 5 instruction slots in a bundle are filled - the 5th has to be a branch instruction (no branch, no slot filled), and is the only place a branch can go (1 instruction, 1 branch = 2 slots filled out of the 5). There are other limitations as well.
I expect the architecture to have plenty of headroom for the 980 with SMT when it comes out in a couple of years.
1.3GigaBYTES per send and 7.2GigaBYTES per second. Don't understate the bandwidth by almost an order of magnitude!
0.13um G4 has 512KB L2 cache, and is probably around 80 to 90 mm^2 in size.
Also the part of the 970 that bundles up instructions has some severe limitations that means that it is very rare that all 5 instruction slots in a bundle are filled - the 5th has to be a branch instruction (no branch, no slot filled), and is the only place a branch can go (1 instruction, 1 branch = 2 slots filled out of the 5). There are other limitations as well.
I expect the architecture to have plenty of headroom for the 980 with SMT when it comes out in a couple of years.
Thanks Hattig. Normally I'm a stickler for getting that correct but I was fast and didn't check my work.
Yes I purposely did not mention the 7457 @.13um because it's not due to ship until Q3 possibly Q4. Once can only imagine how much heat the 1.4Ghz .18um G4's are producing. I'd guess about 45-50watts per processor!! Have you seen the heatsink on them!!!
Yes I agree Hattig. The PPC 970 looks to be IBM's initial foray. The fun starts with how well the handle the Power5. I can't wait to read about the SMT results.
Good question. I'll look to see if there's any info in any pdfs.
Re: Two Motherboard Prototypes
I would say it's just as likely that the dual processor is for high end desktop systems, whereas the single processor is for the Xserves.
I would say it's just as likely that the dual processor is for high end desktop systems, whereas the single processor is for the Xserves.
I have to disagree w/this statement. I split my time between macrumors and Anandtech.com and most of the PC geeks over there would rather run OS X but they see Macs as being underpowered for the price. Something I've noticed in the Mac vs. PC debates over there is that PC people keeping pointing to hardware specs and Mac people keep pointing to the software bundle. I've come to realize that when most people shop for computers they only look at the hardware, and not the software they are getting. This makes sense 'cause the bundled software that comes w/most PCs is either crap or a limited version of an expensive program. But when you buy a Mac you get, IMO, a very nice software bundle and can do almost anything you want right out of the box.
/rant
Lethal
Ah, not OT at all -- in fact, a keen observation. Presumably 10.3 will be the minimum OS for 970-based Macs. I will take this as a friendly amendment to my original prediction: Steve will announce the 970-based Macs and 10.3 in July and they both will ship in the late summer or early fall.
It's not just that, it's the flexability of the PC. If you're a hardware geek (Anand audience) you probably have more than one PC and swap stuff between them. You keep costs low by slowly upgrading, and you know how to assemble a faster better PC than a Dell for the same price.
It's a tough change for someone like that to accept Apple's market and price manipulation. What I have to wait for ever for the next chip, there's no roadmap, and I have to change my whole computer to upgrade? Forget that!!!
The problem with the 970 that everyone is waiting for is that it's going to be marketed in a computer by Apple. Do you really think you'll get one in an iMac? Forget it - these will be shipped in the $3000+ macs first, and most of the folks lusting for one on this board will have to wait a year until they get down to sensible prices.
Germanium
There was a yield problems in the late 1980's with some 5ns stuff made with germanium. The yield was about 5% while the performance was excellent. IBM has done a lot if they can achieve acceptable production yields using germanium.
There was a yield problems in the late 1980's with some 5ns stuff made with germanium. The yield was about 5% while the performance was excellent. IBM has done a lot if they can achieve acceptable production yields using germanium.
>Steve will announce the 970-based Macs and 10.3 in July
No he won't. It has already been annouced that Steve isn't going to be giving a Keynote at MacWorld in July, Phil Schiller will be. That's the best indication that the new Powermacs will be post-MacWorld.
No he won't. It has already been annouced that Steve isn't going to be giving a Keynote at MacWorld in July, Phil Schiller will be. That's the best indication that the new Powermacs will be post-MacWorld.
Q3 may be slightly over optimistic.
Motorola Press Release
Let's see what exactly is Motorola's track record meeting expectations
NO WAY THIS YEAR... WHY ??
Because Apple won't be at this year's Macworld Boston, so there will definately be no annoucements there. And Steve named 2003 the Year of the Laptop.
So it's January 2004.
Marc K.
Because Apple won't be at this year's Macworld Boston, so there will definately be no annoucements there. And Steve named 2003 the Year of the Laptop.
So it's January 2004.
Marc K.