Back to the topic. <a href="http://news.zdnet.co.uk/story/0,,t269-s2131244,00.html"; target="_blank">ZDNet</a> says: [quote] IBM says a later PowerPC 970 will reach 2.5GHz using a 90-nanometre manufacturing process.<hr></blockquote>
I don't know whom to believe. One possibility is that the 1.8?2.5 GHz stuff we discuss was a wrong piece of info and that's why IBM pulled it back. The other is that it's all true and IBM pulled it because they wanted to make a surprise. Like Santa, you know.
In the first case ZDNet provided the most accurate info, which I doubt; in the latter they just reposted outdated quotes.
There is no reason that the PPC 970 could not scale higher in Mhz than an athlon.
The ppc 970 is up to fourteen stage vs twelve for the AThlon and seven for the G4. IBM is also a top chip maker. 2,5 ghz on SOI 0,13 do not seems impossible.
<strong>I don't know whom to believe. One possibility is that the 1.8?2.5 GHz stuff we discuss was a wrong piece of info and that's why IBM pulled it back. The other is that it's all true and IBM pulled it because they wanted to make a surprise. Like Santa, you know.
In the first case ZDNet provided the most accurate info, which I doubt; in the latter they just reposted outdated quotes.</strong><hr></blockquote>
apple likes options and a steady Mhz increase is one of them. so i think they asked ibm to pull the press release. so apple can introduce a 1.8Ghz ppc970 (maybe mp) as highend machine with a 1.6Ghz and 1.4Ghz as middle and low end powermacs. so the real fast chips can be introduced in a super powermac at mwsf2004 or so, as a "oh, one more thing" santa surprice.
not IBM likes surprices: steve jobs does. so the reason to pull a press release with such an impact is because one of you're main customers asks you to. one of the reason is because it would hurt apple sales.
and don't give me the "ibm is to big to listen to apple" shit, because apple is much to important for the entire computer industry to ignore. yes, size does mather... and apple has a very big one in its pants (everybody in the industry thinks so: if apple shows it's stuff everybody pais attention :eek: )
So you're saying that IBM won't introduce their new servers with the 2.5 GHz Power PC 970 just to appease Apple? I don't think so.
Let's say that the 2.5 GHz processors are ready to go in July and they put the 1.8 GHz 970 in the new machines as the 'Fastest' machine.
If IBM launches their server with dual 2.5 GHz 970s, then we know Apple is holding back. That's not going to make customers happy. We want the power and speed, and we want it two years ago. For Apple to hold back now is not acceptable. We'll take the 2.5 GHz Power PC 970 for a year if we need to; we just want it ASAP.
Not only that, but we're expecting big things from this 970, including the ability to increase clock speed on a regular basis, epsecially with the drop from .13 to .009. I hope that IBM puts out the new chips so fast that it's hard for Apple to keep up.
I hop that IBM put out the new chips so fast that Apple will have the opportunity to put the 970 not only in the PowerBook, but the 970 in the iMac fairly quickly.
Imagine this for a top end line up a year from now:
Power Mac G5
Dual Processor 2.5 GHz Power PC 970
PowerBook G5
1.8 GHz Power PC 970
iMac
1.4 GHz Power PC 970
iBook
1.0 GHz Power PC G4
Why can't this be a realistic outlook? Apple needs this type of speed for their machines.
<strong>So you're saying that IBM won't introduce their new servers with the 2.5 GHz Power PC 970 just to appease Apple? I don't think so.
.</strong><hr></blockquote>
me neither and i didn't say that, i said they pulled the press release because apple asked them to. todays 1.42Ghz (oops have to edit my last post) DP customers will be mad as hell when apple introduces a 2.5Ghz highend and a 1.8Ghz low end as next pm at mwny2003 or so.
.... <img src="embarrassed.gif" border="0"> , okay now i think i understand what you mean...
or they introduce single processor configurations then they can go on.
but nevertheless i think they pulled the press release because of apple.
Well that's kind of closing the barn door after the horse has escaped. They already let the cat out of the bag.
Now they think they can pull the wool over our eyes? (Okay, enough with the stupid sayings.)
It's like Apple getting Toshiba to pull the PR on the 40 GB iPod HD press release. It's too late- we know the drive is available and now we're expecting new iPods. If we don't get them, customers will be unhappy.
The same is true with the 970. Perhaps Apple does want to surprise us, but IBM is going to release their 970 servers at some point, and Mac buyers aren't going to be happy to see the 970s running much faster in those servers. It will mean that Apple could have shipped them but didn't.
How embarassing would it be if you could run Mac OS X 20% faster in Mac on Linux on a non-Apple machine? It wouldn't be good, that's for sure.
To whomever commented that Nintendo was already committed to 2004:
It appears then that one of two things has happened. 1.) Nintendo has decided to change business models only AFTER losing a ton of money, or 2.) What did you expect them to say?
As for DSP's, not happening. QE is Raycer's contribution. No one is going to use PPC reference boards when they're an even worse value that Apple's typically overpriced hardware. Also, spending this money to put such silicon on board doen't get you much when CPU performance would eclipse it in a matter of months. Then again, a certain baseline of A/V performance would then always be there (on any mac) regardless of what CPU performance did or didn't do. It could make a better value out of your mac. But this still has nothing to do with raycer. QE is raycer, these were video guys, QE is a unique application of the GPU that seems like it is about 24 months ahead of anything by Windows guys. Money well spent, but it's spent now. They bought talent too, those guys are working on other things because you'll never unseat the ability of ATI and nVidia, or even VideoLogic and Matrox, to deliver more performance every six months-year. You spend the money custom designing hardware and the GPU makers soon make your efforts irrelevant -- that's why there is no raycer "chip" -- the rayer guys are just making the steady supply of otherr people's chips do more work for the mac than they would otherwise do, a good strategy.
<strong>How embarassing would it be if you could run Mac OS X 20% faster in Mac on Linux on a non-Apple machine? It wouldn't be good, that's for sure. </strong><hr></blockquote>
We know Jobs knows it. Jobs also knows we know he knows it.
[quote]To whomever commented that Nintendo was already committed to 2004:
It appears then that one of two things has happened. 1.) Nintendo has decided to change business models only AFTER losing a ton of money, or 2.) What did you expect them to say?<hr></blockquote>
This is fairly off topic, but who says that Nintendo is losing money? Up until a few months ago, Nintendo was the only company making a profit on their consoles. Sony is now also making a profit on their hardware.
Seriously, Nintendo doesn't go on the model of losing money on hardware and since they are the biggest seller of games on their platform, they are making tons of money. Don't forget that the GBA controls the handheld gaming market.
They are also the largest company in Japan. If they want to stay in the console business, they won't have a problem.
Now back to your regularly scheduled Apple discussion.
Well, I only mentioned it because Nintendo was offered up as evidence of PPC development. Just as you say, Nintendo doesn't like losing money on consoles, but the next generation basically demands that the console maker lose quite a bit of money on the consoles. I didn't say they're losing money, I said they will most likely lose a ton of money on the next generation (if they aren't very careful).
M$ is really spending a lot of cash on the xBox, of the three, I expected the xBox to die a lot sooner, but it's hanging in there and slowly winning franchises (or buying them.)
Sony has a lot of money and alternate cash flows, as does M$. I dunno if nintendo can sweat enought details to go it alone one more time. The safest route would be to take the gameboy and their gaming franchise and run for the platform agnostic hills.
Well that's good and all Matsu, but remember that the Infinium Labs (a new comer to the market) and Sammy/Sega are planning new consoles. Supposedly Sega has been able to contain the power of the DC on a single chip and they are always working on new arcade boards. Who knows what will happen in 2005?
<strong>Imagine this for a top end line up a year from now:
Power Mac G5
Dual Processor 2.5 GHz Power PC 970
PowerBook G5
1.8 GHz Power PC 970
iMac
1.4 GHz Power PC 970
iBook
1.0 GHz Power PC G4
Why can't this be a realistic outlook? Apple needs this type of speed for their machines.</strong><hr></blockquote>
i think your quit realistic about this, but, according to your own comments about my previous post, aren't these specs a little pessimistic? especially the highend tower looks a little weak, if, as you imply, the 2.5Ghz 970 will be introduced at mwny2003 or fall 2003. it almost looks like it stagnates like a G4. besides that: how will this processor act in a powerbook (powerconsumption/heatproduction), or will we see a G4 7557 at 1.8Ghz instead and is a 1Ghz ibook not a little underpowered (a clockspeed increase of 20% in 1 year) or will we indeed see a steady increase of clockspeed during the next 1 year with your expected specs as a result?
<strong>As for DSP's, not happening. QE is Raycer's contribution. No one is going to use PPC reference boards when they're an even worse value that Apple's typically overpriced hardware. Also, spending this money to put such silicon on board doen't get you much when CPU performance would eclipse it in a matter of months. Then again, a certain baseline of A/V performance would then always be there (on any mac) regardless of what CPU performance did or didn't do. It could make a better value out of your mac. But this still has nothing to do with raycer. QE is raycer, these were video guys, QE is a unique application of the GPU that seems like it is about 24 months ahead of anything by Windows guys. Money well spent, but it's spent now. They bought talent too, those guys are working on other things because you'll never unseat the ability of ATI and nVidia, or even VideoLogic and Matrox, to deliver more performance every six months-year. You spend the money custom designing hardware and the GPU makers soon make your efforts irrelevant -- that's why there is no raycer "chip" -- the rayer guys are just making the steady supply of otherr people's chips do more work for the mac than they would otherwise do, a good strategy.</strong><hr></blockquote>
I find it unlikely that the Racyer guys had anything at all to do with QE unless a couple of them just happened to be OpenGL experts and were transfered to the OpenGL team (entirely possible). The use of the GPU as a compositing engine isn't revolutionary -- many video games have been doing it since GPUs appears on the scene. What is revolutionary about it is that Apple finally put in the effort to make it happen at a system level.
You're right, VSPs in the chipset probably won't have anything to do with Quartz Extreme. QE is the compositing engine and the existing system uses the GPU and works fine. Much more likely, in addition to the other uses I listed, would be for the hypothetical VSPs to be used to accelerate Quartz the rasterizing engine. This piece of software operates to build the images in main memory before they are fed to the QE compositor. The GPUs are designed to rasterize triangles while Quartz primarily uses curves, thus GPU acceleration isn't likely any time soon and would be of questionable efficiency if it was done. A seperate VSP with full memory bandwidth to the system RAM, on the other hand, could possibly do a decent job of it. Whether or not any Raycer people were involved is irrelevent.
Apple has done the DSP thing before, and that didn't work out very well. That was also before they could integrate it into the chipset, and before VSPs came along and became as well understood as they are today. This would give Apple a new way to distinguish their hardware from the PC world and would certainly help in manipulating multimedia.
For IBM the 970 is an alternativie to the Xeon. If the Xeon spec marks are about the same as the P4 3GHz, IBM will need 1.8 GHz to get parity and more than that for having an advantage. So for IBM to compete with their own dual Xeon blade servers they will need dual 970 at a high speed.
For Apple the internal competition is the lame dual G4 and for that even one 1.8 or a dual 1 GHz 970 is enough to be much faster . For Appe to compete with a single P4 3 GHz they will need either really fast 970 or midrange duals.
While ramping up IBM could keep the 970 above 1.8 GHz to use in their servers that sell in relative low numbers compared to personal computers. Apple could get the stuff at and below 1.8 GHz a lineup like SP 1.5, DP 1.5, DP 1.8 would induce a feeding frency among the Mac buyers. Even the lowend would be faster than the current top of the line and the top end 3 times as fast as the current dual 1.42 G4 and also much faster than a SP P4 3 GHz. <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
So I think there is a way to keep both IBM and Apples customers happy
The notion that Apple should hold back speed to not annoy people that have G4s is stupidity! We G4 owners should know better:
1984 68000
1985
1986
1887 68020
1988 68030
1989
1990
1991 68040
1992
1993
1994 601
1995 604
1996
1997 G3
1998
1999 G4
2000
2001
2002
2003????
Do note that before G4 there was a CPU there was new top of the line CPUs at least every 3 years sometime more often. The fact that there was no new CPU in 2001 was a bit odd and that there was no one in 2002 is actually a record.
With the G4 as the longest "top of the line" CPU in the Macintosh history combined the infamous 500 MHz barrier and scaling problem a new CPU vastly superior to the G4 can only come as a surprise to ignorant
I got a new dual 1.25 earlier this week and I will be happy if the next low end tower is several times as fast as this one
Comments
[ 03-06-2003: Message edited by: costique ]</p>
In the first case ZDNet provided the most accurate info, which I doubt; in the latter they just reposted outdated quotes.
The ppc 970 is up to fourteen stage vs twelve for the AThlon and seven for the G4. IBM is also a top chip maker. 2,5 ghz on SOI 0,13 do not seems impossible.
<strong>I don't know whom to believe. One possibility is that the 1.8?2.5 GHz stuff we discuss was a wrong piece of info and that's why IBM pulled it back. The other is that it's all true and IBM pulled it because they wanted to make a surprise. Like Santa, you know.
In the first case ZDNet provided the most accurate info, which I doubt; in the latter they just reposted outdated quotes.</strong><hr></blockquote>
apple likes options and a steady Mhz increase is one of them. so i think they asked ibm to pull the press release. so apple can introduce a 1.8Ghz ppc970 (maybe mp) as highend machine with a 1.6Ghz and 1.4Ghz as middle and low end powermacs. so the real fast chips can be introduced in a super powermac at mwsf2004 or so, as a "oh, one more thing" santa surprice.
not IBM likes surprices: steve jobs does. so the reason to pull a press release with such an impact is because one of you're main customers asks you to. one of the reason is because it would hurt apple sales.
and don't give me the "ibm is to big to listen to apple" shit, because apple is much to important for the entire computer industry to ignore. yes, size does mather... and apple has a very big one in its pants (everybody in the industry thinks so: if apple shows it's stuff everybody pais attention :eek: )
[ 03-06-2003: Message edited by: gar ]</p>
Let's say that the 2.5 GHz processors are ready to go in July and they put the 1.8 GHz 970 in the new machines as the 'Fastest' machine.
If IBM launches their server with dual 2.5 GHz 970s, then we know Apple is holding back. That's not going to make customers happy. We want the power and speed, and we want it two years ago. For Apple to hold back now is not acceptable. We'll take the 2.5 GHz Power PC 970 for a year if we need to; we just want it ASAP.
Not only that, but we're expecting big things from this 970, including the ability to increase clock speed on a regular basis, epsecially with the drop from .13 to .009. I hope that IBM puts out the new chips so fast that it's hard for Apple to keep up.
I hop that IBM put out the new chips so fast that Apple will have the opportunity to put the 970 not only in the PowerBook, but the 970 in the iMac fairly quickly.
Imagine this for a top end line up a year from now:
Power Mac G5
Dual Processor 2.5 GHz Power PC 970
PowerBook G5
1.8 GHz Power PC 970
iMac
1.4 GHz Power PC 970
iBook
1.0 GHz Power PC G4
Why can't this be a realistic outlook? Apple needs this type of speed for their machines.
[ 03-06-2003: Message edited by: Fran441 ]</p>
<strong>So you're saying that IBM won't introduce their new servers with the 2.5 GHz Power PC 970 just to appease Apple? I don't think so.
.</strong><hr></blockquote>
me neither and i didn't say that, i said they pulled the press release because apple asked them to. todays 1.42Ghz (oops have to edit my last post) DP customers will be mad as hell when apple introduces a 2.5Ghz highend and a 1.8Ghz low end as next pm at mwny2003 or so.
....
or they introduce single processor configurations then they can go on.
but nevertheless i think they pulled the press release because of apple.
Now they think they can pull the wool over our eyes? (Okay, enough with the stupid sayings.)
It's like Apple getting Toshiba to pull the PR on the 40 GB iPod HD press release. It's too late- we know the drive is available and now we're expecting new iPods. If we don't get them, customers will be unhappy.
The same is true with the 970. Perhaps Apple does want to surprise us, but IBM is going to release their 970 servers at some point, and Mac buyers aren't going to be happy to see the 970s running much faster in those servers. It will mean that Apple could have shipped them but didn't.
How embarassing would it be if you could run Mac OS X 20% faster in Mac on Linux on a non-Apple machine? It wouldn't be good, that's for sure.
It appears then that one of two things has happened. 1.) Nintendo has decided to change business models only AFTER losing a ton of money, or 2.) What did you expect them to say?
As for DSP's, not happening. QE is Raycer's contribution. No one is going to use PPC reference boards when they're an even worse value that Apple's typically overpriced hardware. Also, spending this money to put such silicon on board doen't get you much when CPU performance would eclipse it in a matter of months. Then again, a certain baseline of A/V performance would then always be there (on any mac) regardless of what CPU performance did or didn't do. It could make a better value out of your mac. But this still has nothing to do with raycer. QE is raycer, these were video guys, QE is a unique application of the GPU that seems like it is about 24 months ahead of anything by Windows guys. Money well spent, but it's spent now. They bought talent too, those guys are working on other things because you'll never unseat the ability of ATI and nVidia, or even VideoLogic and Matrox, to deliver more performance every six months-year. You spend the money custom designing hardware and the GPU makers soon make your efforts irrelevant -- that's why there is no raycer "chip" -- the rayer guys are just making the steady supply of otherr people's chips do more work for the mac than they would otherwise do, a good strategy.
<strong>How embarassing would it be if you could run Mac OS X 20% faster in Mac on Linux on a non-Apple machine? It wouldn't be good, that's for sure.
We know Jobs knows it. Jobs also knows we know he knows it.
It appears then that one of two things has happened. 1.) Nintendo has decided to change business models only AFTER losing a ton of money, or 2.) What did you expect them to say?<hr></blockquote>
This is fairly off topic, but who says that Nintendo is losing money? Up until a few months ago, Nintendo was the only company making a profit on their consoles. Sony is now also making a profit on their hardware.
Seriously, Nintendo doesn't go on the model of losing money on hardware and since they are the biggest seller of games on their platform, they are making tons of money. Don't forget that the GBA controls the handheld gaming market.
They are also the largest company in Japan. If they want to stay in the console business, they won't have a problem.
Now back to your regularly scheduled Apple discussion.
M$ is really spending a lot of cash on the xBox, of the three, I expected the xBox to die a lot sooner, but it's hanging in there and slowly winning franchises (or buying them.)
Sony has a lot of money and alternate cash flows, as does M$. I dunno if nintendo can sweat enought details to go it alone one more time. The safest route would be to take the gameboy and their gaming franchise and run for the platform agnostic hills.
<strong>Imagine this for a top end line up a year from now:
Power Mac G5
Dual Processor 2.5 GHz Power PC 970
PowerBook G5
1.8 GHz Power PC 970
iMac
1.4 GHz Power PC 970
iBook
1.0 GHz Power PC G4
Why can't this be a realistic outlook? Apple needs this type of speed for their machines.</strong><hr></blockquote>
i think your quit realistic about this, but, according to your own comments about my previous post, aren't these specs a little pessimistic? especially the highend tower looks a little weak, if, as you imply, the 2.5Ghz 970 will be introduced at mwny2003 or fall 2003. it almost looks like it stagnates like a G4. besides that: how will this processor act in a powerbook (powerconsumption/heatproduction), or will we see a G4 7557 at 1.8Ghz instead and is a 1Ghz ibook not a little underpowered (a clockspeed increase of 20% in 1 year) or will we indeed see a steady increase of clockspeed during the next 1 year with your expected specs as a result?
<strong>As for DSP's, not happening. QE is Raycer's contribution. No one is going to use PPC reference boards when they're an even worse value that Apple's typically overpriced hardware. Also, spending this money to put such silicon on board doen't get you much when CPU performance would eclipse it in a matter of months. Then again, a certain baseline of A/V performance would then always be there (on any mac) regardless of what CPU performance did or didn't do. It could make a better value out of your mac. But this still has nothing to do with raycer. QE is raycer, these were video guys, QE is a unique application of the GPU that seems like it is about 24 months ahead of anything by Windows guys. Money well spent, but it's spent now. They bought talent too, those guys are working on other things because you'll never unseat the ability of ATI and nVidia, or even VideoLogic and Matrox, to deliver more performance every six months-year. You spend the money custom designing hardware and the GPU makers soon make your efforts irrelevant -- that's why there is no raycer "chip" -- the rayer guys are just making the steady supply of otherr people's chips do more work for the mac than they would otherwise do, a good strategy.</strong><hr></blockquote>
I find it unlikely that the Racyer guys had anything at all to do with QE unless a couple of them just happened to be OpenGL experts and were transfered to the OpenGL team (entirely possible). The use of the GPU as a compositing engine isn't revolutionary -- many video games have been doing it since GPUs appears on the scene. What is revolutionary about it is that Apple finally put in the effort to make it happen at a system level.
You're right, VSPs in the chipset probably won't have anything to do with Quartz Extreme. QE is the compositing engine and the existing system uses the GPU and works fine. Much more likely, in addition to the other uses I listed, would be for the hypothetical VSPs to be used to accelerate Quartz the rasterizing engine. This piece of software operates to build the images in main memory before they are fed to the QE compositor. The GPUs are designed to rasterize triangles while Quartz primarily uses curves, thus GPU acceleration isn't likely any time soon and would be of questionable efficiency if it was done. A seperate VSP with full memory bandwidth to the system RAM, on the other hand, could possibly do a decent job of it. Whether or not any Raycer people were involved is irrelevent.
Apple has done the DSP thing before, and that didn't work out very well. That was also before they could integrate it into the chipset, and before VSPs came along and became as well understood as they are today. This would give Apple a new way to distinguish their hardware from the PC world and would certainly help in manipulating multimedia.
For Apple the internal competition is the lame dual G4 and for that even one 1.8 or a dual 1 GHz 970 is enough to be much faster . For Appe to compete with a single P4 3 GHz they will need either really fast 970 or midrange duals.
While ramping up IBM could keep the 970 above 1.8 GHz to use in their servers that sell in relative low numbers compared to personal computers. Apple could get the stuff at and below 1.8 GHz a lineup like SP 1.5, DP 1.5, DP 1.8 would induce a feeding frency among the Mac buyers. Even the lowend would be faster than the current top of the line and the top end 3 times as fast as the current dual 1.42 G4 and also much faster than a SP P4 3 GHz. <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
So I think there is a way to keep both IBM and Apples customers happy
The notion that Apple should hold back speed to not annoy people that have G4s is stupidity! We G4 owners should know better:
1984 68000
1985
1986
1887 68020
1988 68030
1989
1990
1991 68040
1992
1993
1994 601
1995 604
1996
1997 G3
1998
1999 G4
2000
2001
2002
2003????
Do note that before G4 there was a CPU there was new top of the line CPUs at least every 3 years sometime more often. The fact that there was no new CPU in 2001 was a bit odd and that there was no one in 2002 is actually a record.
With the G4 as the longest "top of the line" CPU in the Macintosh history combined the infamous 500 MHz barrier and scaling problem a new CPU vastly superior to the G4 can only come as a surprise to ignorant
I got a new dual 1.25 earlier this week and I will be happy if the next low end tower is several times as fast as this one
[ 03-06-2003: Message edited by: DrBoar ]</p>
The market moves fast, and if you don't sell the fastest chip at your disposal today, it's only going to get hardeer to sell it tomorrow!
They can't hold anything back, they have to come with the fastest duals they can find AND LOWER PRICES.