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.
IBM 64-Bit PowerPC Details
- Thread starter MacRumors
- Start date
- Sort by reaction score
gotohamish
macrumors 65816
Is it not possible that all this talk of the Power4 NEXT YEAR isn't actually true. I'm (hoping and) thinking that it's a case of a deal between IBM and Apple.
Think about it, look how long it took all the upgrade chips to appear for speeds over 550 G4 - when Apple had had them for well over a year. This makes sense, Apple won't want too many 'upgraders' as they want people to buy new machines.
Well, no one can purchase Power4 Lites until a year from now, except Apple who is already stocking up on Power4's now, so that perhaps the next PMs will be using them (no OS9 boot), and a year from now they'll be available to OTHERS like upgrade companies etc.
God, I'm being positive this morining. Here's to it!
Think about it, look how long it took all the upgrade chips to appear for speeds over 550 G4 - when Apple had had them for well over a year. This makes sense, Apple won't want too many 'upgraders' as they want people to buy new machines.
Well, no one can purchase Power4 Lites until a year from now, except Apple who is already stocking up on Power4's now, so that perhaps the next PMs will be using them (no OS9 boot), and a year from now they'll be available to OTHERS like upgrade companies etc.
God, I'm being positive this morining. Here's to it!
jayscheuerle
macrumors 68020
Weener show...
What are you people doing that current machines do not provide enough horsepower for (other than running OSX)? ;-)
Are you running renderfarms? The reason that megahertz doesn't matter is that a 450mHz G4 is fast enough for 95% of the applications out there (excluding games perhaps) if you have enough RAM.
Seems to me that most people just want to open up a can of WHUPASS on some PC weenies....
What are you people doing that current machines do not provide enough horsepower for (other than running OSX)? ;-)
Are you running renderfarms? The reason that megahertz doesn't matter is that a 450mHz G4 is fast enough for 95% of the applications out there (excluding games perhaps) if you have enough RAM.
Seems to me that most people just want to open up a can of WHUPASS on some PC weenies....
jayscheuerle
macrumors 68020
Falleron,
Maybe so... I'm not up on the real tech details but it sure looks good. Oh and remember this CPU will be moving from the current 130nm process to the 90nm process in a pretty quick time frame... From what I know about that kinda move... Very little to be honest but one thing I do know is smaller IS better. =)
- More speed
- Less heat
Even if the 1st round of 970s don't make it the 2nd round sure seems like a lock! Somehow I don't think the same thing can be said for a 4Ghz Intel cpu. 😀
Dave
Wow. That's amazing.
How often can Intel say, "we'll give you a massive perfomance boost with less heat and power consumption?"
Next gen Apple laptops are going to be amazing!
ps. where are those stats from?
Falleron
macrumors 68000
The article is from :
http://maccentral.macworld.com/news/0210/16.ibm.php
The bit about power consumption is near the bottom of the article.
http://maccentral.macworld.com/news/0210/16.ibm.php
The bit about power consumption is near the bottom of the article.
dongmin
macrumors 68000
sounds like MWSF '04
From the MacCentral article:
[Peter] Sandon [senior PowerPC architect at IBM] said that the SIMD unit on the PowerPC 970 uses the same set of 162 instructions that Motorola's AltiVec uses. This means that applications optimized for AltiVec will perform normally on the PowerPC 970. Halfhill added that all applications optimized for AltiVec on the Macintosh should see performance boosts on the PowerPC 970, if it indeed is used by Apple.
"From my analysis," Halfhill said, "it (IBM's SIMD) is virtually compatible (with AltiVec). Why would IBM add this if it wasn't for Apple?"
I couldn't have agreed more. The facts are out there. And everyone seems certain this processor is slated for Macs. Apple might as well come out and say it.
Sandon said the IBM lab has PowerPC 970 samples today. He expects the PowerPC 970 to sample to customers by the second quarter of 2003, with the processor expected to enter production in the second half of 2003.
What's the typical lag between when samples go to customers and when the chips make their way into actual machines that are shipping? Six months? Nine months? My guess is the new machines will debut at MWSF 2004. Unless of course, IBM is on an accelerated timetable with Apple, but I doubt it.
From the MacCentral article:
[Peter] Sandon [senior PowerPC architect at IBM] said that the SIMD unit on the PowerPC 970 uses the same set of 162 instructions that Motorola's AltiVec uses. This means that applications optimized for AltiVec will perform normally on the PowerPC 970. Halfhill added that all applications optimized for AltiVec on the Macintosh should see performance boosts on the PowerPC 970, if it indeed is used by Apple.
"From my analysis," Halfhill said, "it (IBM's SIMD) is virtually compatible (with AltiVec). Why would IBM add this if it wasn't for Apple?"
I couldn't have agreed more. The facts are out there. And everyone seems certain this processor is slated for Macs. Apple might as well come out and say it.
Sandon said the IBM lab has PowerPC 970 samples today. He expects the PowerPC 970 to sample to customers by the second quarter of 2003, with the processor expected to enter production in the second half of 2003.
What's the typical lag between when samples go to customers and when the chips make their way into actual machines that are shipping? Six months? Nine months? My guess is the new machines will debut at MWSF 2004. Unless of course, IBM is on an accelerated timetable with Apple, but I doubt it.
Originally posted by DaveGee:
Also remember this: Motorola claimed they would be manufacturing chips on the 90nm process 6 months before anyone else.
I don't know if this means any of these chips will be powerpcs for Apple desktops, but I remember Moto making the claim about the 90nm process.
Also remember this: Motorola claimed they would be manufacturing chips on the 90nm process 6 months before anyone else.
I don't know if this means any of these chips will be powerpcs for Apple desktops, but I remember Moto making the claim about the 90nm process.
Till you read this :
SPECint_base2000
=================
970 Pentium4 2.8Ghz
1212 Pentium4 3.5 Ghz
1385 Pentium4 4 Ghz
1628 Pentium4 4.7Ghz
804 Power4 1.3 GHz
=================
SPECfp_base
=================
938 Pentium4 2.8Ghz
1172 Pentium4 3.5 Ghz
1340 Pentium4 4 Ghz
1574 Pentium4 4.7Ghz
1202 Power4 1.3 GHz
=================
Estimates I made with a calculator
Of course in realworld terms a 1.8Ghz GPUL could be faster than a 4Ghz Pentium 4 in all areas. The memory throughput on the GPUL is half the clockspeed speed at 1.8Ghz (1800MHz with 900-MHz RAM) compared with almost a quarter of the throughput on a 4Ghz Pentium 4 (4000Mhz with 1066Mhz RAM). That should make it more efficient to start with, especially seeing as it's 8 way superscaler compared with 4 way on the Pentium 4 and G4. Infact given that in realworld tests the pentium 4 has an IPC of about 1.8 and a G4 has an IPC of 2.3, a GPUL could be something like 4.6 - 5 IPC in the realworld. Can't wait!
What if...
I'm a little green on this stuff, so bear with me.
If Apple got of its butt and actually brought up the rest of the system components around this new chip, especially the memory bus, couldn't it compete with SGI workstations on some levels.
I'm not quite sure why SGI machines are so powerful. Do they have some kind of proprietary system architecture. They are in the lower MHz ratings and nobody complains about them, yet they kick everyones butts.
In a nutshell, what would need to be done to get the SGI performance on a Mac, seeing that MHz really doesn't matter?
I'm a little green on this stuff, so bear with me.
If Apple got of its butt and actually brought up the rest of the system components around this new chip, especially the memory bus, couldn't it compete with SGI workstations on some levels.
I'm not quite sure why SGI machines are so powerful. Do they have some kind of proprietary system architecture. They are in the lower MHz ratings and nobody complains about them, yet they kick everyones butts.
In a nutshell, what would need to be done to get the SGI performance on a Mac, seeing that MHz really doesn't matter?
jadam
macrumors 6502a
no, its only SPECint_base and SPECfp_base.
these chips will rock with there 6.4gb/sec of bandwith and 900mhz bus!!
SGI uses Many Many CPUS in its systems.
and this is going to be one sweet processor, ohh yeah, they say there going ot sample 2ndq uarter of 2003 and systems will be released in 2nd half of 2003, well, apple does have MacWorld Boston, and remember the latest 1.25ghz powermacs came in the 2nd half of 2003 sooooooo!!!!
The main question is.
Do you even care what apple bring out in the meantime if these monsters cpus are on the horizon ?
Do you even care what apple bring out in the meantime if these monsters cpus are on the horizon ?
That's the reason why Apple is going to wait to announce they're going to use this chip until we hear Steve say "Oh, and one more thing...".
Apple doesn't want to kill all their sales in the time between now and when this chip is used by them. If a dualie G4 is what you need now, buy it. The best computer for anyone is the one they use efficiently and well, not the one they "might" be getting.
Now, that being said, I'm waiting to get an iMac until the situation with the 15" version settles itself out. 😉 😀
Re: Weener show...
Rendering takes time.... time is money. The faster the Mac the faster I can get paid. Waiting sucks.