IBM Intros Power5 Server

[MacRumors]

IBM announced (http://home.businesswire.com/portal/site/google/index.jsp?ndmViewId=news_view&newsId=20040503005393&newsLang=en) the first in a line of servers powered by the Power5 processor today.

The new server, the eServer i5, is aimed at small and mid-sized businesses.

The eServer i5 systems are powered by IBM's next-generation POWER5 microprocessor, the most advanced 64-bit chip in the world. POWER5 features an impressive 276 million transistors per processor, and is manufactured with IBM's 0.13-micron copper wiring and SOI (Silicon-on-Insulator) technologies. In addition to providing communications acceleration and chip multiprocessing, POWER5 offers simultaneous multithreading (SMT), which transforms a single processor into two processors, essentially allowing the chip to run two applications at the same time and reducing the time it requires to complete a task.

The new server/Power5 will become available starting on June 11, 2004.

Early rumors had hinted that future PowerPC chips may be derived from the Power5 processor.

[joelypolly]

Good for apple if thereis one

[P-Worm]

Sounds pretty impressive to me, but we'll see how thing play out in the real world.

IBM is a great company and I'm glad Apple has a good relationship with them.

P-Worm

[Veldek]

IBM is a great company and I'm glad Apple has a good relationship with them.

I just hope they don't become to Motorola-like. The latest news didn't sound too good. Let's hope these new processor will evolve (soon) to a desktop version for Apple. Imagine 3.5GHz processors with SMT...

Edit: Does anyone wonder why these chips are manufactured by 0.13-micron and SOI-technology and not 0.09 and SSOI?

[wdlove]

It's impressive. Now for this to be used in our beloved Apple Products. The IBM VP said last year with the G5 release, "We are already working on the next generation chips." I wonder if this was what he meant?

[wrldwzrd89]

I just hope they don't become to Motorola-like. The latest news didn't sound too good. Let's hope these new processor will evolve (soon) to a desktop version for Apple. Imagine 3.5GHz processors with SMT...

Edit: Does anyone wonder why these chips are manufactured by 0.13-micron and SOI-technology and not 0.09 and SSOI?
It's good to see the Power5! I think that SMT will make it into the derivatives Apple uses in their machines. In regards to your other question, there are two reasons: Firstly, IBM's 90nm/SSOI manufacturing process is not very mature yet and needs to have some kinks worked out of it. Secondly, doing the same thing with the Power5 that was done with the Power4 saves development costs, at least initially - plus it leaves room for the Power5 to improve once 90nm manufacturing becomes stable.

[idkew]

Imagine 3.5GHz processors with SMT...


i would imagine that this would be left out, in order to make the chip smaller, less complex and less expensive... i believe they did a similar thing with the power4/g5

[Mac-Xpert]

Well this processor is fabed at 130 nm. So this news probably doesn't say anything about the state of the 970fx. I do still hope we will see the 3.0 Ghz G5's in June, but with recent news I won't be surprised if the rev-B's would go any higher than 2.5 Ghz.

[D*I*S_Frontman]

WaHoo!!! G5 PowerBook is on the way!!!

(just thought I'd say it--you know SOMEONE had to!)

[wrldwzrd89]

i would imagine that this would be left out, in order to make the chip smaller, less complex and less expensive... i believe they did a similar thing with the power4/g5
No - the Power4/PowerPC G5 (970) did not have SMT support at release. IBM may have planned earlier to include it, but decided to hold off until the Power5 to use SMT.

[adamfilip]

Im sure when IBm makes the G5 replacement processor it will have SMT.
and i bet it will come next summer.

My Prediction for Summer of 2005 at WWDC

Dual 4.5 Ghz G5 Pro with SMT (equal to 4 virtual processors)

[wrldwzrd89]

Im sure when IBm makes the G5 replacement processor it will have SMT.
and i bet it will come next summer.

My Prediction for Summer of 2005 at WWDC

Dual 4.5 Ghz G5 Pro with SMT (equal to 4 virtual processors)
That would be a huge step forward for the Mac world if it happened. Do you think Apple will dub IBM's Power6 the G6 and introduce it in 2006, for added effect?

[adamfilip]

well in Order to keep pace with intel and AMD.

they need too

The P4 will be at 4ghz by year end. and already has a form of SMT
and a better version of SMT i believe is expected year end.

so Apple/IBM has no choice

[Veldek]

Did Apple (or someone else) tell us when they went to IBM, so that we can calculate how long it took IBM to make the 970? Then we had an estimation how long it takes until we see a Desktop version of this chip.

[Veldek]

That would be a huge step forward for the Mac world if it happened. Do you think Apple will dub IBM's Power6 the G6 and introduce it in 2006, for added effect?

Well, it's the Power5, so we might even see it next year... ;)

[pjkelnhofer]

This can only be good news for Apple. If the Power5 becomes a widely used chip the cost per chip will probably go down. I have always believed that one of the reasons of Mac's high costs is the high cost of a CPU that very few people are using.

One thing though, I read the entire press release, but could not find any mention of how fast the Power5 chips used in the eServer i5 actually runs.
Any ideas?

Also, could some one refresh me on the difference between the chip that is currently the "G5" and the Power5 chip in the IBM server. I thought that one is a derivative of the other, but I am not really sure.

[Veldek]

Also, could some one refresh me on the difference between the chip that is currently the "G5" and the Power5 chip in the IBM server. I thought that one is a derivative of the other, but I am not really sure.

The G5 (or 970) is a derivative of the Power4 AFAIK.

[idkew]

No - the Power4/PowerPC G5 (970) did not have SMT support at release. IBM may have planned earlier to include it, but decided to hold off until the Power5 to use SMT.

sorry, maybe i was thinking of the dual cores then.?.

[wrldwzrd89]

Well, it's the Power5, so we might even see it next year... ;)
I think Apple might just call the Power5 an enhanced G5 and save the G6 moniker for the Power6. It would be nice to see a G6 next year, but it doesn't fit with Apple tradition. Look at what Apple did with the 603/603e/604/604e (all considered "G2") and Motorola's 74XX series (all of them are called "G4" by Apple) for examples.

[wrldwzrd89]

sorry, maybe i was thinking of the dual cores then.?.
Yes - you were thinking about the dual cores. These were present in some Power4 models, but did not make it to the PowerPC 970.

[adamfilip]

The G5 is based on the Power 4 Architecture , 970

its basicly the kid brother of the Power 4. much smaller die. but runs at higher clock speed.

same will happen with the Power 5

[Billy_ca]

Who on earth could rate this as negative?!!!

IBM is really the king of the hill as far as top end server processors go...they are leaving HP, Sun, Intel and the others in the dust! The others still haven't found a match for the POWER4 let alone the POWER5.

Only AMD is looking good at the lower end of the server spectrum with the Opteron. But the Opteron isn't in the same class of processors as the POWER5.

This is really good news for Apple. IBM is really motivated to keep pushing the envelope with its process technology for the POWER line...and Apple will stand to benefit with the PowerPC 9x0 line!

Hurrrray!

[Veldek]

I think Apple might just call the Power5 an enhanced G5 and save the G6 moniker for the Power6. It would be nice to see a G6 next year, but it doesn't fit with Apple tradition. Look at what Apple did with the 603/603e/604/604e (all considered "G2") and Motorola's 74XX series (all of them are called "G4" by Apple) for examples.

Ah, I see what you mean, but I think the Power6 is still more than a year off, so I doubt we will see a Power6 derivative in 2006. Of course, I like to be surprised... :)

[JFreak]

always remember that there are severe differencies with server class hardware and workstation class hardware, the reliability being the most important. it is actually ok to allow workstation to make calculation mistakes and have some amount of downtime, but in server hardware both of the two are severe mistakes.

server class hardware always has mature technology, not the latest and greates. so that's why no 90nm and no ssoi.

ibm technology looks very good. i'm really looking forward what apple can bring out within next six months. something's up, that's certain ;)

[wrldwzrd89]

always remember that there are severe differencies with server class hardware and workstation class hardware, the reliability being the most important. it is actually ok to allow workstation to make calculation mistakes and have some amount of downtime, but in server hardware both of the two are severe mistakes.

server class hardware always has mature technology, not the latest and greates. so that's why no 90nm and no ssoi.

ibm technology looks very good. i'm really looking forward what apple can bring out within next six months. something's up, that's certain ;)
You are right, but don't forget about the problems IBM's been having getting reasonable yields from their 90nm manufacturing process. These problems surely were a factor in making the decision to go 130nm initially with the Power5, no?

[IrishGold]

IBM is really the king of the hill as far as top end server processors go...they are leaving HP, Sun, Intel and the others in the dust! The others still haven't found a match for the POWER4 let alone the POWER5.


Hehe, no, but they are close.

[neilw]

You are right, but don't forget about the problems IBM's been having getting reasonable yields from their 90nm manufacturing process. These problems surely were a factor in making the decision to go 130nm initially with the Power5, no?

The decision to go with 130 nm was undoubtedly made long before the problems with 90 nm surfaced.

A similar discussion came up when the 970 came out, about how it was smaller and faster than the Power4, because the high-end server chips are manufactured in fairly conservative fashion, in order to be absolutely bulletproof.

Unfortunately, I suspect that we will hear little or nothing official about the purported PPC975 until Apple releases it in a product. Should keep the rumor mills running nicely over the next 6 months or so.

[JFreak]

You are right, but don't forget about the problems IBM's been having getting reasonable yields from their 90nm manufacturing process. These problems surely were a factor in making the decision to go 130nm initially with the Power5, no?

if i remember correctly, the power4 was 180nm so it's only logical to produce the next version with 130nm technology. i would have been very surprised if they would have chosen 90nm for power5.

to repeat myself: ibm needs to have rock-solid technology behind the power-series. it's a benefit for ibm to be able to get money (by selling the next step for apple) from the research they must do anyway (because they need to be always at least one step ahead in r&d than in powerX chip production) - and that's the #1 reason ibm can sell G5 chips for apple at competitive prices.

so...

if ibm didn't have the powerX series, apple would not have a chance at getting G5 chips from ibm at current prices.

[wrldwzrd89]

The decision to go with 130 nm was undoubtedly made long before the problems with 90 nm surfaced.

A similar discussion came up when the 970 came out, about how it was smaller and faster than the Power4, because the high-end server chips are manufactured in fairly conservative fashion, in order to be absolutely bulletproof.

Unfortunately, I suspect that we will hear little or nothing official about the purported PPC975 until Apple releases it in a product. Should keep the rumor mills running nicely over the next 6 months or so.
I'm certainly looking forward to whatever the rumor mill has in store regarding new PowerPC processors. Any reason why you specify 6 months? My guess is that 6 months is the usual time between upgrades for Apple (and that would be the logical time to introduce the PPC 975) - am I correct?

[minstryoffunk]

I just hope they don't become to Motorola-like. The latest news didn't sound too good. Let's hope these new processor will evolve (soon) to a desktop version for Apple. Imagine 3.5GHz processors with SMT...

Edit: Does anyone wonder why these chips are manufactured by 0.13-micron and SOI-technology and not 0.09 and SSOI?

i may be mistaken (and probably am), but if memory serves, the problems that IBM has been having with the 970fx is because of the migration to 90nano and ssoi at the same time.

my guess is that by using the process that they've already got figured out, they can move units as soon as possible and then probably update with a 90nano ssoi version when all the kinks get worked out.

[thatwendigo]

well in Order to keep pace with intel and AMD.

they need too

The P4 will be at 4ghz by year end. and already has a form of SMT
and a better version of SMT i believe is expected year end.

so Apple/IBM has no choice

The current Pentium 4 "Prescott" cores aren't going above 3.4ghz until they solve that 100w power draw. When their Centrino Dothans hit in about a week, unless the shipping target misses, then they'll have a chip that draws about 35w at peak performance for 2.0ghz, scales down to 600mhz and around 8-12w at idle, and performs roughly on par with a 3.55ghz P4.

When that happens... How long do you think the P4 will have on the desktop? Someone will get the bright idea to create a standard ATX board for the Centrino, and then the desktop modders will be running something just as performance-capable but at a third of the heat, while not needing to spend as much as, say, as P4EE 3.4 ($1,049) or an Athlon FX-53 ($804).

Then there's savings on cooling, form factor, and so on...

The competition is not with the Pentium 4, which is hitting all kinds of walls because of inneficient core design and wasteful use of resources. It's with AMD's very PowerPC-like Athlon cores, and with the much more efficient Penitum-M line.

As for there being no choice... You're wrong.

IBM is rumored to be working on a chip for the XBox at the moment, one that is pretty well in line with some of their other designs, which will end up as a three-core single chip processor. Sharing 1MB of L2 cache and on-die memory control, this monster will act like six processors at once (due to SMT providing virtual "cores" on each real one) as far as an aware OS is concerned. In other words, whatever number of threads your software can assign to a processor, it now has six of them to do its spooling through.

Freescale is working on a dual-core, double-precision AltiVec processor as well (starting as en e600, then transferring to the e700 line). While not a triple-core SMT machine, it still acts as two processors on one chip at just slightly higher heat than the current 74xx generation. For applications that make use of floating point math, the AltiVec implementation should prove a godsend, because it will allow that complicated dual-precision math to be done twice as fast.

Clockspeed is not the be-all, end-all, as you would think mac users would realize by now. The G5 and Athlon XP and 64 are all lower clocked, as is the Opteron. Going to claim that they can't keep up, though?

I thought not.

[minstryoffunk]

Well this processor is fabed at 130 nm. So this news probably doesn't say anything about the state of the 970fx. I do still hope we will see the 3.0 Ghz G5's in June, but with recent news I won't be surprised if the rev-B's would go any higher than 2.5 Ghz.

i think this shows that theres a chance that when the next g5's come, they'll be 130nano power5 derivatives.

that seems like that would make the most sense for both apple and IBM short term, and that way it buys IBM time to smooth out 90nano.

[wrldwzrd89]

The current Pentium 4 "Prescott" cores aren't going above 3.4ghz until they solve that 100w power draw. When their Centrino Dothans hit in about a week, unless the shipping target misses, then they'll have a chip that draws about 35w at peak performance for 2.0ghz, scales down to 600mhz and around 8-12w at idle, and performs roughly on par with a 3.55ghz P4.

When that happens... How long do you think the P4 will have on the desktop? Someone will get the bright idea to create a standard ATX board for the Centrino, and then the desktop modders will be running something just as performance-capable but at a third of the heat, while not needing to spend as much as, say, as P4EE 3.4 ($1,049) or an Athlon FX-53 ($804).

Then there's savings on cooling, form factor, and so on...

The competition is not with the Pentium 4, which is hitting all kinds of walls because of inneficient core design and wasteful use of resources. It's with AMD's very PowerPC-like Athlon cores, and with the much more efficient Penitum-M line.

As for there being no choice... You're wrong.

IBM is rumored to be working on a chip for the XBox at the moment, one that is pretty well in line with some of their other designs, which will end up as a three-core single chip processor. Sharing 1MB of L2 cache and on-die memory control, this monster will act like six processors at once (due to SMT providing virtual "cores" on each real one) as far as an aware OS is concerned. In other words, whatever number of threads your software can assign to a processor, it now has six of them to do its spooling through.

Freescale is working on a dual-core, double-precision AltiVec processor as well (starting as en e600, then transferring to the e700 line). While not a triple-core SMT machine, it still acts as two processors on one chip at just slightly higher heat than the current 74xx generation. For applications that make use of floating point math, the AltiVec implementation should prove a godsend, because it will allow that complicated dual-precision math to be done twice as fast.

Clockspeed is not the be-all, end-all, as you would think mac users would realize by now. The G5 and Athlon XP and 64 are all lower clocked, as is the Opteron. Going to claim that they can't keep up, though?

I thought not.
Well, that scenario would be interesting. The Pentium-M, modified for desktops, competing with the Athlon line by AMD and the POWER/PowerPC line by IBM/Apple/Motorola/Freescale? I would never had suspected that would occur. If it does occur, though, it could jump-start Intel's processor development again (which is stalling right now) - perhaps egging on IBM to follow suit? Who knows how it would work out - but it sure makes the ballgame more interesting.

[JoeMacDaddy]

Folks, Big Blue is coming back and Apple has tied their wagon to their shooting star. Big Blue is set to give Intel, Microsoft, Sun, Dell & HP the Big Blue middle finger. ;) If you look at what they have rolled into the unit, they control everything that goes into the unit. They pay nothing to Intel, MS or others, except for comodity parts. Big Blue is Open Source because it means taking back the entreprise and desktop from Microsoft. It has taken them 20 years but they are coming back. Apple stands to profit because their enemies and Big Blue's are one in the same. Also, they compliment each other. If you buy a new G5/G6 Macintosh, you are buying and IBM machine inside.

[areyouwishing]

Well, that scenario would be interesting. The Pentium-M, modified for desktops, competing with the Athlon line by AMD and the POWER/PowerPC line by IBM/Apple/Motorola/Freescale? I would never had suspected that would occur. If it does occur, though, it could jump-start Intel's processor development again (which is stalling right now) - perhaps egging on IBM to follow suit? Who knows how it would work out - but it sure makes the ballgame more interesting.

This scenario is FACT, its not a rumor kids. 2006/2007 will see the end of the Pentium 4 line as we know it today.

http://news.com.com/2100-1006-5201338.html
By 2006, a Pentium M derivative called Jonah is expected to become the mainstay for the desktop line.

and...

http://news.com.com/2100-1006-5181256.html?tag=nl
The convergence of the two chip families through Merom means that desktop performance won't likely slow down, but the acceleration of megahertz will. Some notebook chips may creep into the desktop line with Jonah, a predecessor of Merom, due in 2006. A new chip-numbering plan that de-emphasizes megahertz will help Intel get around any thorny marketing issues.

[wrldwzrd89]

This scenario is FACT, its not a rumor kids. 2006/2007 will see the end of the Pentium 4 line as we know it today.

http://news.com.com/2100-1006-5201338.html


and...

http://news.com.com/2100-1006-5181256.html?tag=nl
It was inevitable - Intel have finally admitted that the Pentium 4, in its current form, is a dead architecture. It consumes WAY too much power - just look at what happened when Intel tried to shrink the Pentium 4 to a 90nm process - the power consumption dropped 10W (110W -> 100W). Hardly impressive, now is it? 100W is still far too high. The Pentium M, though, is a whole different story. Its power consumption rates are down there with the likes of the PowerPC G3 - in some cases, BETTER than the G3.

[sinisterdesign]

Folks, Big Blue is coming back and Apple has tied their wagon to their shooting star. Big Blue is set to give Intel, Microsoft, Sun, Dell & HP the Big Blue middle finger. ;)

i agree, it seemed like we didn't hear much from IBM for YEARS in the 90's. a new Stinkpad here & there, but nothing major. all the buzz was around the Wintel cartel. i'm stoked to see IBM come blazing back w/ what seems to me (processor ignorant) an excellent line of processors coming down the pipe.

i'm glad that Apple got tired of watching Moto flounder around trying to crank out a new processor once in a blue moon & found someone that wants to LEAD the way.

hmmmm....if Windows + Intel = Wintel, what do we call the Apple/IBM partnership? "IB-Apple"? "AppleBuM"? "Big Blue Apple"? i'm stumped...

[IrishGold]

i agree, it seemed like we didn't hear much from IBM for YEARS in the 90's.

You weren't listening very well, hehe.

They had their times.

[pjpoon]

If my memory serves me correctly, Apple came in late in the development cycle of the Power 4 when they had IBM add in the vector processing and develop the G5. I believe that there was a rumor/news story/comment somewhere around that time that Apple was developing the desktop equivalent of the Power 5 simultaneously with IBM. If that is so, perhaps the June 11 debut of IBM's new servers will be closely followed by the announcement of the Apple Power 5 equivalent at the WWDC a couple weeks later.

[thatwendigo]

It was inevitable - Intel have finally admitted that the Pentium 4, in its current form, is a dead architecture. It consumes WAY too much power - just look at what happened when Intel tried to shrink the Pentium 4 to a 90nm process - the power consumption dropped 10W (110W -> 100W). Hardly impressive, now is it? 100W is still far too high. The Pentium M, though, is a whole different story. Its power consumption rates are down there with the likes of the PowerPC G3 - in some cases, BETTER than the G3.

The Centrino 1.7ghz runs at 30-35w at peak, and 25-27w average consumption, but drops to around 8-12w when it clocks down to 600mhz at idle. When compared to P4 desktop machines, it tends to perform favorably when lined up with 2.8-30ghz P4s. If you extrapolate this to the jump from 1.7 to 2.0ghz, you see why Intel is going to be using the Centrino and its derivatives on the desktop.

It's hardly rocket science:
P4 3.4ghz @ 100w
P-M 2.0ghz @ 30-35w

I wonder which I would choose, were I to be buying a PC...

[thatwendigo]

If my memory serves me correctly, Apple came in late in the development cycle of the Power 4 when they had IBM add in the vector processing and develop the G5. I believe that there was a rumor/news story/comment somewhere around that time that Apple was developing the desktop equivalent of the Power 5 simultaneously with IBM. If that is so, perhaps the June 11 debut of IBM's new servers will be closely followed by the announcement of the Apple Power 5 equivalent at the WWDC a couple weeks later.

I've already suggest this in another thread, when the Power5 announcement was being discussed but hadn't happened yet. Essentially, I believe that we'll see PowerPC 980s at WWDC that scale to at least 3.0ghz.

I think that Steve's got something up his sleeve, or rather more than one thing. The VX is waiting in the wings in case Freescale doesn't deliver on the e600 promises, the 970 is about to become a low-end processor that will be further revised to take care of power consumption issues, and the 980 will become the new high-end.

The main question is one of the features that will remain when the 980 is revealed. Will it be SMT? Will it be multi-core? What clock will it top out at on the first revision? Is it pin-compatible with the current motherboards? If not, has Apple already taken that into account and had a board waiting?

Most importantly, can I order and have one shipped the day of the conference? :D

[gerardrj]

Edit: Does anyone wonder why these chips are manufactured by 0.13-micron and SOI-technology and not 0.09 and SSOI?

A guess, but:

.13 SOI is a proven process, it's a safe place to start with a new design. Starting a brand new chip on a very new process would cause too mane debugging issues (is it the process, or the design, or the interaction of the two).
Once IBM gets the .09 process stable with high yields, then they will start to debut new chips with it.

Also, ince these are server chips and will likely be in server room environments the demands are different: get performance at any cost. You don't need to worry about heat or noise since the box is not sitting at a desk.

[rikers_mailbox]

WaHoo!!! G5 PowerBook is on the way!!!

(just thought I'd say it--you know SOMEONE had to!)

Don't be that guy.

[Soire]

What are the odds that this is all just really optimistic specualtion?

Isn't it still conceivable that at WWDC they simply introduce a revised PM running on a 970 or 970fx and clocked at 2.4 or 2.6?

Don't get me wrong, I'm waiting to buy a rev. B as soon as they debut, and want it to be ridiculously fast, but I get the feeling all this talk is setting me up for a big letdown. I just don't want to expect that a 3Ghz Power5 derivative is coming out in 6 weeks when we'll be getting something far from it. Please don't toy with my hopes!! ;)

[IrishGold]

Please don't toy with my hopes!! ;)


Im willing to bet nothing extremely great will come out of WWDC.

It will be another mini iPod event.

[the future]

I've already suggest this in another thread, when the Power5 announcement was being discussed but hadn't happened yet. Essentially, I believe that we'll see PowerPC 980s at WWDC that scale to at least 3.0ghz.

I sincerely hope so, too, and think that it's quite possible. One minor point, though: from all I've read on the Mac web the Power5 derivative will be called PPC 975, not 980.

[thatwendigo]

What are the odds that this is all just really optimistic specualtion?

Nobody but certain employees of Apple and IBM know.

Isn't it still conceivable that at WWDC they simply introduce a revised PM running on a 970 or 970fx and clocked at 2.4 or 2.6?

Certainly it's conceivable, and just as likely as anything else. Without solid information on the process, this is all speculation.

Don't get me wrong, I'm waiting to buy a rev. B as soon as they debut, and want it to be ridiculously fast, but I get the feeling all this talk is setting me up for a big letdown. I just don't want to expect that a 3Ghz Power5 derivative is coming out in 6 weeks when we'll be getting something far from it. Please don't toy with my hopes!! ;)

A bit of advice:

Use rumors sites for what they're good at, in ideal circumstances, and only listen to what the general trends are. Never get your hopes pinned on what some people on a messageboard say. It's a lot harder to have your expectations broken that way.

This is all just combining information that's out there with a healthy dose of critical thinking.

[tace]

The decision to use the 130nm for this chip versus the 90nm was a smart business decision. Since the 90nm line at Fishskill was already busy pumping out the latest and the greatest, why not bring new life to your old plant/line that is 130nm capable by introducing this new architecture.

I imagine this way they had more time to perfect the Power5 with SMT while the other line was busy with making G5s. As someone has already mentioned once the 90nm plant is upto speed it could be upgraded to produce these new Power5s. Then IBM can retool the older line to produce whatever is the next generation of chips after 90nm.

[mgargan1]

Well, please don't think of me as a jerk when i say this, but it's not 130nm, it's .13nm. BIG (well, technically small)... the other thing is, what's to keep apple from using the POWER5, and use it for large mainframes, kinda like IBM sells? Also, i remeber reading that the 2.0GHz 970fx has a 550MHz Frontside bus, while the 970 has a 500MHz frontside bus. However, if that were true, the specs on the xserve website would show an 1100MHz Frontside bus. Oh well...

[Hattig]

If that is so, perhaps the June 11 debut of IBM's new servers will be closely followed by the announcement of the Apple Power 5 equivalent at the WWDC a couple weeks later.

I wouldn't say "weeks", maybe a few months.

This could get Apple out of the 3Ghz PPC 970FX issue. IBM is having problems with the 90nm shrink and all that, and 3GHz clocks might not be possible. This is an industry wide problem as well. IBM About 90nm (http://www.siliconstrategies.com/article/showArticle.jhtml?articleId=19400132)

However a 2GHz-ish 130nm POWER5 derivative could be claimed by Jobs to be "3GHz equivalent". Firstly the core is more efficient, so it could perform 20% better per clock (2.4GHz equivalent). Then SMT adds around 30% performance increase (according to IBM, IIRC), making it 3.1GHz equivalent. So even if things aren't that rosy, a 2.2GHz POWER5 derivative should be a good match for a 3GHz 970FX on multi-threaded applications.

[Hattig]

Well, please don't think of me as a jerk when i say this, but it's not 130nm, it's .13nm. BIG (well, technically small)... the other thing is, what's to keep apple from using the POWER5, and use it for large mainframes, kinda like IBM sells? Also, i remeber reading that the 2.0GHz 970fx has a 550MHz Frontside bus, while the 970 has a 500MHz frontside bus. However, if that were true, the specs on the xserve website would show an 1100MHz Frontside bus. Oh well...

The 970 / 970FX interconnect to the chipset runs at 1/4 the clock speed with DDR.

So a 2.2 GHz processor has a 550MHz interconnect ("Elastic I/O") performing 1100 MTps (mega-transfers per second, although most people just say "effective MHz").

[Bulgroz]

Well, please don't think of me as a jerk when i say this, but it's not 130nm, it's .13nm.

No it's 130nm (or .13 micron (µm))

[thatwendigo]

No it's 130nm (or .13 micron (µm))

Man, if only it were .13nm... We could have, what, 80ghz processors then? ;)

[howard]

hmmmm....if Windows + Intel = Wintel, what do we call the Apple/IBM partnership? "IB-Apple"? "AppleBuM"? "Big Blue Apple"? i'm stumped...



this is for all you people who hate it when people spell mac with capitol letters

IBMAC

[mgargan1]

yea, you're right... as i was re reading what I wrote, i realized i was wrong, and that I am a jerk for sayin that...

[adamfilip]

Well, please don't think of me as a jerk when i say this, but it's not 130nm, it's .13nm. BIG (well, technically small)... the other thing is, what's to keep apple from using the POWER5, and use it for large mainframes, kinda like IBM sells? Also, i remeber reading that the 2.0GHz 970fx has a 550MHz Frontside bus, while the 970 has a 500MHz frontside bus. However, if that were true, the specs on the xserve website would show an 1100MHz Frontside bus. Oh well...

What keeps apple from using it is.. IBM!.. I dont think IBM will sell Apple Power5 Processors to Build a competing product. IBM makes too much money on monster servers that it doesnt need another competitor in that space.

IBM sells apple 970's to fill a niche. area with apple workstations.

[Piker]

Only on a rumor board can a group of people take one little piece of news like this and run with it years out into the future. We all read this news about a new IBM server and just started dreaming, and somehow the string evolved and we started talking about whether or not the Power6 (which doesn't even exist) could make the transition to a G6 for a 2006 release date. I LOVE IT!

Right now, I'd be happy just to see a G5 go anything over 2.0 ghz. Or what would make me REALLY happy is if somebody could figure out a way to cram a G5 into a powerbook already. I'm dying here.

-Piker

[dontmatter]

Well, please don't think of me as a jerk when i say this, but it's not 130nm, it's .13nm. BIG (well, technically small)... the other thing is, what's to keep apple from using the POWER5, and use it for large mainframes, kinda like IBM sells? Also, i remeber reading that the 2.0GHz 970fx has a 550MHz Frontside bus, while the 970 has a 500MHz frontside bus. However, if that were true, the specs on the xserve website would show an 1100MHz Frontside bus. Oh well...

It would compete with IBM, and apple kinda wants to make IBM happy. Also, I would remind you that apple is still pretty darned new in everything but consumer computers.

[oingoboingo]

When that happens... How long do you think the P4 will have on the desktop? Someone will get the bright idea to create a standard ATX board for the Centrino, and then the desktop modders will be running something just as performance-capable but at a third of the heat, while not needing to spend as much as, say, as P4EE 3.4 ($1,049) or an Athlon FX-53 ($804).

Then there's savings on cooling, form factor, and so on...

The competition is not with the Pentium 4, which is hitting all kinds of walls because of inneficient core design and wasteful use of resources. It's with AMD's very PowerPC-like Athlon cores, and with the much more efficient Penitum-M line.

Well said. People get caught up on the Pentium 4, but it's really a sideshow attraction compared to what's happening with the Pentium M, Athlon 64 and Opteron. 'Dothan' on the desktop will offer up some stiff competition to the G5 in more ways than just computational speed...and that's not even considering the current explosion of 64-bit Athlon 64 and Opteron desktops, workstations and servers. With a seemingly invigorated Motorola/Freescale and IBM's POWER5 being released soon, this is a pretty interesting time for CPU development. Certainly more interesting than waiting for Intel to crank another 100MHz and another 50W out of the Pentium 4 <yawn>

[eSnow]

Dothan' on the desktop will offer up some stiff competition to the G5 in more ways than just computational speed...

Well, yes and no.
The PentiumM lacks the SMT-cabilities of the P4 as well as the cache coherency protocol needed for dual- or quad-CPU boards. Furthermore, it has a much shallower pipeline than the P4 which is good on the one hand since it makes it more performant per Mhz, but on the other hand limits its clock frequency. While the P-M can match a P4 in some tasks, this is not true for all - there are some parallels to the comparison between 7447A and 970.

[aussiemac86]

Well, please don't think of me as a jerk when i say this, but it's not 130nm, it's .13nm. BIG (well, technically small)....


Actually thats not correct it is 130nm or 0.13microns, 1 micron=1000nm(nanometers)- i double checked this conversion from my university physics text book

And yeh pleasee dont think of me as a jerk either

[rdowns]

WaHoo!!! G5 PowerBook is on the way!!!

(just thought I'd say it--you know SOMEONE had to!)

I thought these inane, PB G5s are coming, posts in every thread couldn't get more annoying. I am not afraid to admit I was wrong.

[rdowns]

i agree, it seemed like we didn't hear much from IBM for YEARS in the 90's. a new Stinkpad here & there, but nothing major. all the buzz was around the Wintel cartel. i'm stoked to see IBM come blazing back w/ what seems to me (processor ignorant) an excellent line of processors coming down the pipe.

i'm glad that Apple got tired of watching Moto flounder around trying to crank out a new processor once in a blue moon & found someone that wants to LEAD the way.

hmmmm....if Windows + Intel = Wintel, what do we call the Apple/IBM partnership? "IB-Apple"? "AppleBuM"? "Big Blue Apple"? i'm stumped...

How about I've Become a Macintosh?

[johnnowak]

Imagine 3.5GHz processors with SMT...

I don't have to imagine it... I can go use my friend's PC.

[ffakr]

Well, yes and no.
The PentiumM lacks the SMT-cabilities of the P4 as well as the cache coherency protocol needed for dual- or quad-CPU boards. Furthermore, it has a much shallower pipeline than the P4 which is good on the one hand since it makes it more performant per Mhz, but on the other hand limits its clock frequency. While the P-M can match a P4 in some tasks, this is not true for all - there are some parallels to the comparison between 7447A and 970.
This is the post I was about to make. :D
Claiming that a PM is everybit the performer of a P4 is really stretching it. It's a good chip, though the PM's I've used have had noticable lag when they were in MHz slewing. It isn't as fast as a P4 though.
Dothan will be faster, and it will have a fairly massive 2MB L2 cache.. but there are limits where L2 becomes a rapidly diminishing return in terms of performance boost. It's likely that, to some extent, the large L2 is probably tacked on as filler for the relatively small PM core.. so it's fabbed at a manageable size. No one wants to try and hook up 500 or 700 leads to a 20mm^2 chip. The 970fx is a much more complex (wide) processor than the PM and it only comes in at 65mm^2 with 512K L2. It's pretty small.

Also, keep in mind that though Intel has been saying for years that they are serious about controlling waste heat, they've continued to develop cpus that follow the design mantra of the P4 to an ever greater extent. The Prescott is really like a P4 plus. It's less efficient with longer pipes. Once again, Intel decided to trade more top end speed for less efficiency. Tejas is next in line and it's expected to clock even FASTER.

I think it's safe to say that it will be a long time before Intel gets rid of its fascination with raw MHz.. regardless of what news.com.com says. Hey, didn't news.com.com recently proclaim that Apple had abandoned all support for versions of OS X prior to 10.3? Hmn.

[mgargan1]

Actually thats not correct it is 130nm or 0.13microns, 1 micron=1000nm(nanometers)- i double checked this conversion from my university physics text book

And yeh pleasee dont think of me as a jerk either

nah, no jerk here... actually, i caught myself, and I said that I was the jerk for making an ass out of myself... But thanks for keeping me in check.

[oingoboingo]

Well, yes and no.
The PentiumM lacks the SMT-cabilities of the P4 as well as the cache coherency protocol needed for dual- or quad-CPU boards. Furthermore, it has a much shallower pipeline than the P4 which is good on the one hand since it makes it more performant per Mhz, but on the other hand limits its clock frequency. While the P-M can match a P4 in some tasks, this is not true for all - there are some parallels to the comparison between 7447A and 970.

I agree. I'm not trying to say that the Dothan Pentium M will be the death of the Pentium 4, but it makes much more sense than the Pentium M in many circumstances (obviously notebooks, but also small form factor systems like Shuttles which are becoming very popular, and in other low-noise desktop systems). While the Dothan Pentium M may not be able to touch the Pentium 4 at the very high end, or for SMP systems, it is a undoubtedly a strong performer, without needing to rely on a cranked-up clock speed and all the undesirable heat, power and cooling issues which follow.

I'm just guessing that the Pentium 4 might step back from the podium as Intel's flagship general purpose 32 bit processor and become more of a niche player, with the desktop and notebook implementations of the cooler and less power hungry Dothan M becoming much more common. That's not even mentioning the competition that Intel's Pentium 4 and Pentium 4 Xeon is facing from AMD in the high-end desktop, workstation and entry-server market, courtesy of the Athlon 64 and Opteron.

Like I said before, with competition from AMD, IBM, a (hopefully) re-focused Motorola, plus improved Pentium M chips from Intel themselves, the CPU market is a much more diverse and interesting place than when it was just a singular focus on Pentium 4s hot enough to fry an egg :)

[Mac-Xpert]

Hmm, reading the article on news.com (http://news.com.com/2100-1006-5201338.html) I think they are saying that the chip which will replace the P4 will be the "Jonah" and not the "Dothan". The Dothan might not yet be as powerful as the P4 in certain aspects, but who knows what this "Jonah" will be capable of.

I do think however that the Rev-B Powermacs will use the 970fx and not some Power5 derivative (975 / 980 or whatever). It seems unlikely to me that the Xserve will be the only machine to feature the 970fx. The transition from the Power5 core to the 975 / 980 on 90 nm process might take some time. And I think the 90 nm process is required to offer chips above 2.0 Ghz.

Maybe I’m not as positive as some in this thread, but I think we might only see a introduction of 2.5-2.6 Ghz as the top at the WWDC. With Steve simply ignoring the 3.0 Ghz promise he made last year.

[nek]

I do think however that the Rev-B Powermacs will use the 970fx and not some Power5 derivative (975 / 980 or whatever). It seems unlikely to me that the Xserve will be the only machine to feature the 970fx. The transition from the Power5 core to the 975 / 980 on 90 nm process might take some time. And I think the 90 nm process is required to offer chips above 2.0 Ghz.

Maybe I’m not as positive as some in this thread, but I think we might only see a introduction of 2.5-2.6 Ghz as the top at the WWDC. With Steve simply ignoring the 3.0 Ghz promise he made last year.

I think that there is a real possibility that a Power5 derivative was made concurrently, and therefore will be ready for introduction at WWDC. Moving the Powermac and Xserve to this new chip would allow the 970fx to be used in the iMac. The 970fx can likely reach 2.6 or 2.8GHz, so that would provide enough room to upgrade the the iMac and eMac a couple times over the next year or so.

And if Freescale can deliver on its plans for the e600 and e700 in the near future, then Apple won't need to try to put the 970fx in the PowerBook.

[thatwendigo]

Well, yes and no.
The PentiumM lacks the SMT-cabilities of the P4 as well as the cache coherency protocol needed for dual- or quad-CPU boards. Furthermore, it has a much shallower pipeline than the P4 which is good on the one hand since it makes it more performant per Mhz, but on the other hand limits its clock frequency. While the P-M can match a P4 in some tasks, this is not true for all - there are some parallels to the comparison between 7447A and 970.

For desktop use, SMT is currently poorly implemented, if at all. The reason that most home users by anything like an Athlon FX or a P4 EE is to game on, and the performance just isn't showing up the way that the chipmakers have hyped it. This is probably due to coding not handling the extensions and taking advantage of the chip, but that's still not going to do much good when a Dothan revision a year from now starts to beat the P4 at most home-user tasks.

Professionals with big cases and no need to avoid noise can still use Xeon and Athlon FXs, but I don't think that there's going to be a huge market for them once anyone with half a brain starts realizing what they can do for home design with a decent Pentium-M design.

Interestingly, Intel's roadmap shows some indication that this is in the works, and that it will happen by 2006.

Dothan will be faster, and it will have a fairly massive 2MB L2 cache.. but there are limits where L2 becomes a rapidly diminishing return in terms of performance boost. It's likely that, to some extent, the large L2 is probably tacked on as filler for the relatively small PM core.. so it's fabbed at a manageable size. No one wants to try and hook up 500 or 700 leads to a 20mm^2 chip. The 970fx is a much more complex (wide) processor than the PM and it only comes in at 65mm^2 with 512K L2. It's pretty small.

Well... It's the old P6-generation core with some new stuff bolted on. What did you expect? ;)

Honestly, if you've got the processor budget to add cache on the chip without really screwing things on cost, I don't see a reason not to do it. Less fetching means better performance, in general. Yes, it hits a wall after a while, and this is one reason the P4EE just isn't as much of a killer as Intel would like.

Also, keep in mind that though Intel has been saying for years that they are serious about controlling waste heat, they've continued to develop cpus that follow the design mantra of the P4 to an ever greater extent. The Prescott is really like a P4 plus. It's less efficient with longer pipes. Once again, Intel decided to trade more top end speed for less efficiency. Tejas is next in line and it's expected to clock even FASTER.

They can't just drop their old design philosophy, not after it's been entrenched the way it has by their massively successful focus on the consumer tendency to think that numbers are good, and big numbers are better. It's going to take time, money, and effort to correct that perception before they can attempt the move to the Pentium-M descendants on the desktop around, as noted above, 2006.

The variant in question is a derivative of Jonah, the Dothan's successor, with 2MB L2, a 90nm process, and dual-cores. No figures released on speculated clockspeed. In the second half of 2006, Intel is slated to release Merom, and will be both mobile and desktop chip (in the form of Conroe), with multiple cores, 65nm process, 4MB of cache, and a target heat of 45w. Dothan is roughly on schedule, but Jonah and Merom are falling off schedule.

Power5 from IBM was salted for a June release. It's out now. The Power5+ is expected in 2005, and slated to clock up to 3.0ghz in the server version. To extrapolate from the Power4 to 970 clock: Power4 1.3ghz to 970 2.0ghz (54% increase), so Power5 3.0ghz to 980 4.6ghz. That's assuming that they're not being concurrently designed, in which case we might even be able to expect an even greater leap in performance because the lower end chip isn't a bastardization so much as a carefully created alternative.

All processor information from endian (http://endian.net/).

I'm just guessing that the Pentium 4 might step back from the podium as Intel's flagship general purpose 32 bit processor and become more of a niche player, with the desktop and notebook implementations of the cooler and less power hungry Dothan M becoming much more common. That's not even mentioning the competition that Intel's Pentium 4 and Pentium 4 Xeon is facing from AMD in the high-end desktop, workstation and entry-server market, courtesy of the Athlon 64 and Opteron.

Am I the only one who noticed that, this past quarter, AMD outsold Intel? (http://www.forbes.com/enterprisetech/2004/05/03/cx_ld_0503amd.html?partner=yahoo&referrer=)

Like I said before, with competition from AMD, IBM, a (hopefully) re-focused Motorola, plus improved Pentium M chips from Intel themselves, the CPU market is a much more diverse and interesting place than when it was just a singular focus on Pentium 4s hot enough to fry an egg :)

Amen.

I think that there is a real possibility that a Power5 derivative was made concurrently, and therefore will be ready for introduction at WWDC. Moving the Powermac and Xserve to this new chip would allow the 970fx to be used in the iMac. The 970fx can likely reach 2.6 or 2.8GHz, so that would provide enough room to upgrade the the iMac and eMac a couple times over the next year or so.

I'm in agreement on everything but one factor... Let the iMac die! Do something new, Apple. Come on, Ives and Jobs. I know you have to have something up your collective sleeves that you could really wow the punters with, if only you'd do it.

Surprise me!

And if Freescale can deliver on its plans for the e600 and e700 in the near future, then Apple won't need to try to put the 970fx in the PowerBook.

Once again, amen.

[hacksaw]

Am I the only one who noticed that, this past quarter, AMD outsold Intel? (http://www.forbes.com/enterprisetech/2004/05/03/cx_ld_0503amd.html?partner=yahoo&referrer=)

Your conclusions about that article is flat out wrong. Don't fall victim media misrepresentations. :eek:


http://www.overclockers.com/tips00576/


"What 52% Really Is"
Ed Stroligo - 5/4/04


--------------------------------------------------------------------------------


Some people have written me mentioning a report saying that AMD outsold Intel for a week.

Here's the actual article with all the details, BTW.

What it actually says is that AMD desktop computers outsold Intel desktop computers in the U.S. retail market for a week.

That means sales in computer stores and places like Best Buy. It doesn't include mail-order or online sales or corporate purchases, and obviously doesn't include the rest of the world.

What the report doesn't say is that U.S. retail sales has always been a strong point for AMD. Having a marketshare of 40-ish% in this particular market is pretty normal for them, so 52%, while good, is hardly earth-shattering news.

The article linked above goes on to point out that the overall worldwide marketshare (that means everything) of AMD vs. Intel for the first quarter:

AMD: 14.9%
Intel: 83.6%

14.9% is historically on the low side for AMD under normal circumstances. They did worse when they had to curtail production a while back due to excessive inventory, but during the TBird era, marketshare was up around 22-23%.

Then again, folks like Hector Ruiz have been saying that getting a decent price now takes priority over sheer marketshare at AMD, so one would expect a somewhat lower marketshare figure than during the Jerry Sanders era.

Sorry if I burst any of your bubbles, folks, but that's all it is, a bubble.

Ed

[thatwendigo]

Your conclusions about that article is flat out wrong. Don't fall victim media misrepresentations. :eek:

Modification of my original question, then, since I apparently naively trusted Forbes to represent the story halfway fairly...

Am I the only one to notice that AMD outsold Intel in the retail space, where consumers are directly given a chance to touch the machine and use them, rather than just starin at numbers on an online orderform?

[DMann]

How about I've Become a Macintosh?
Now, that's good ;)

[DMann]

The decision to go with 130 nm was undoubtedly made long before the problems with 90 nm surfaced.

A similar discussion came up when the 970 came out, about how it was smaller and faster than the Power4, because the high-end server chips are manufactured in fairly conservative fashion, in order to be absolutely bulletproof.

Unfortunately, I suspect that we will hear little or nothing official about the purported PPC975 until Apple releases it in a product. Should keep the rumor mills running nicely over the next 6 months or so.

Where does one view the roadmap for the PPC 975?

[ffakr]

Where does one view the roadmap for the PPC 975?

There is no roadmap. As far as official info goes, the 975 doesn't exist.
Even more troubling to those of us waiting, or rather hoping for a 975 is the fact that the code number '975' came from Mac OS Rumors.. a site which doesn't have a great track record. Don't get me wrong, MOSR is near and dear to my heart but they often don't seem to understand what they are reporting on.. not to mention that they've been up for something like 6+ years and they've spent about a month off the web every year due to DNS problems (they've been off for at least the last month).

So.. remember that the "975" for all intents and purposes is supported by roughly zero reliable evidence outside MOSR.
I do strongly believe that there is a Power5 based chip in development and I belive that, as the largest by far buyer, Apple is making IBM keep a lid on it. I just can't point you to any info on it that is even remotely reliable.

[ltgator333]

I just hope they don't become to Motorola-like. The latest news didn't sound too good. Let's hope these new processor will evolve (soon) to a desktop version for Apple. Imagine 3.5GHz processors with SMT...

Edit: Does anyone wonder why these chips are manufactured by 0.13-micron and SOI-technology and not 0.09 and SSOI?



I totally agree with this comment ... if IBM doesn't stop dragging their feet, Freescale (formerly Moto..) does have a decent fabrication facility now and might gain some ground and make apple look bad for choosing IBM over them......
dual 3.5Ghz PPC 970fx's would be a killer system as it is, even without the SMT and enhancements of being based off the POWER5..
AND- IBM is making that chip on their 130nm process because it's a proven, rock solid process, as where they're still getting the kinks out of the 90nm process. I beleive if you read ars-technica's review of POWER 5's predecessor, POWER 4, they outline how IBM is very concerned about their POWER series server chips being physically indistructable, to the point of the physical build of the chip is unnecessarily "beefy", kind of like an armoured truck or something. So they will use the proven and totally stable 130nm process over the 90nm probably for a while.

[ffakr]

Man, if only it were .13nm... We could have, what, 80ghz processors then? ;)

actually, we'd probably have issues squeezing the electrons through.
.13 nm is 1.3 x 10^-10 Meters while a diameter of an electron is thought to be somewhere in the area of 5 x 10^-15.
that's plenty big enough for an electron, but like charges do repell each other. I'm sure we'll see plenty of crazy phenomenon before we get near that small (and god knows what we'll need to use to make chips like that).