Quad PowerMacs and the 970MP?

[daveL]

Eastend:

Sure I care. I certainly don't know if it is normal for IBM's Power sales to drop, and since Kagetenshi sounded pretty confident about what he was saying it should be easy for him to elaborate.

Any stats that concentrate on year-over-year % improvement have to be investigated. If I sold 3 units of X last year, but I succeed in selling 5 units this year, which is a 60% increase, does it really matter? I still only sold 5 units. I guess if I was selling aircraft carriers, it would be significant, but if I'm selling $1000 cpus, it isn't. Before you jump all over me, this is called hyperbole, which is an accepted method of making a point.

As an exercise for the reader: Can you tell me when the first Itanium was available for sale? Having answered this query, tell me how dynamic you think Itanium has been as a product; how well has it been received by the marketplace?

 

[ddtlm]

daveL:

Any stats that concentrate on year-over-year % improvement have to be investigated.

Another way to look at it is: in 2004 Itanium system sales were 23% of Power system sales (based on the figures provided by JCheng).

As an exercise for the reader: Can you tell me when the first Itanium was available for sale? Having answered this query, tell me how dynamic you think Itanium has been as a product; how well has it been received by the marketplace?

Clearly Itanium is taking some time to get started, as one would expect of a new ISA that doesn't offer massive performance increases. However $1.7B USD is nothing to sneeze at and your attempt to dismiss it as "unpopular" (or whatever for are aiming for) is misguided. It seems Intel has the money, the talent, and the plan to see this through.

 

[~loserman~]

As an exercise for the reader: Can you tell me when the first Itanium was available for sale? Having answered this query, tell me how dynamic you think Itanium has been as a product; how well has it been received by the marketplace?

I will agree that the Itanium 1 was not well received. The itanium 2 has been though. Intel is selling a lot of them. They are also releasing dual cores this year with massive cache amounts 9 to 12 Meg. The Itanium is really starting to pick up steam now. 13 of the TOP 100 systems are using Itanium 2. Compare that to 2 years ago when there were zero.

 

[~loserman~]

However $1.7B USD is nothing to sneeze at and your attempt to dismiss it as "unpopular" (or whatever for are aiming for) is misguided. It seems Intel has the money, the talent, and the plan to see this through.

Can you imagine what these guys would be saying if Apple was selling $1.7B in Xserve sales?

 

[Cochrane]

Very curious they pulled the software....

Although it doesn't really help or harm the rumor mill - its just showing that Apple knows they slipped up somehow.

Now if they put up another new version that has just 2 cpu options that might give reason to think that we won't see the quads anytime soon.

D

Not really, they could just release the version with the 4 CPUs again once they start selling such PowerMacs.

By the way, isn't it strange that there are multiple references to the 970, the 970FX and the 970MP, but none to the 970GX (I checked)? To me this looks like the rumors about Apple using a 970GX which is a single 970MP core might be wrong.

On the other hand, there was a reference to the 970GX on a page from IBM...

Added: Apple Stores are down. Any relation?

 

[adamjay]

this thing would have to be priced higher than the current offerings.
while the current offerings are aging and it IS time for an update, it would be bad business for Apple to come out with a QUAD core 2.5ghz G5 for $2999 when people are paying $2999 for a dual G5 2.5ghz 1 month before.
the riots that would follow... not pretty.

So, one of three things will happen.

1) we don't see quad cores but rather just the speedbump we've become accustomed to. (2.7ghz to 3ghz)

2) we see a quad core 2.5ghz G5, and it costs $3699 or more.

3) we see quad core G5's at slower clock speeds than the current offerings.

and those scenarios are listed in order of likeliness to happen - in my opinion.

 

[maya]

Finally LightWave 3D can finally live up to its name on the Mac.

Now to max everything out when its released, so I can finally render my 3D work in RT.

Bring the systems on Steve.

 

[FFTT]

The get the feeling that Apple will first offer a single dual core
somewhere near the 3 Ghz threashold for $2999.

They could still surprise us with a Power5 to achieve that goal.

A QuadraMac might follow later for closer to $3999.

If it includes everything the high end user wants,
then that price would not be terribly unreasonable.

This would still be $1000 lower than the current Alienware ALX


IF however, IBM is able to manufacture these chips
in large enough quantities, then the prices may be lower.

The most important issue to me is that any new motherboard
is capable of supporting a variety of options allowing the user
to customize their systems as needed.

 

[maxvamp]

Wouldn't it be nice if Apple brought back some of the niceties of the PowerMac cases G3 B&W -> PowerMac MDD ? Specifically, something like the mainboard mounted on a hinged case door?

One of my biggest pet peeves is reaching into a case where vision becomes limited, AND have limited expansion capabilities.

After playing with a 2.5 this weekend, I would rather see see a case upgrade than 4 procs. I found the 2.5 to be INSANELY FAST!!!

Just a thought...

Max.

 

[JCheng]

I, on the other hand, am going to call you stupid. No, no it doesn't. It says Itanium is growing, certainly, but a 1.5% contraction is well within any sort of normal variance.

~J

Okay, so basically, you trying to tell me that there isn't demand for Itanium, NOT because Itanium didn't see considerable growth (because you yourself concede that it saw overwhelming growth both quantitatively and proportionally) but rather because POWER and SPARC saw only *MINOR* declines in sales (and even then, you completely ignore the fact that the market as a whole saw considerable growth). I'm sorry thats just completely idiotic, the very fact that Itanium sales went up so significantly over last year PROVES that the demand has gone up and is going up. I mean seriously, are you a moron? Do you even know what demand is?

And just consider for a moment the mere POSSIBILITY that Itanium is taking sales (or, rather potential sales) from Power and Sparc, do you honestly think that there is no competition between the brands for potential customers? Where do you think all those Itanium server sales would have gone had Itanium not existed? You've obviously proven your idiocy with your lack of grasp for the obvious.

Any stats that concentrate on year-over-year % improvement have to be investigated. If I sold 3 units of X last year, but I succeed in selling 5 units this year, which is a 60% increase, does it really matter? I still only sold 5 units. I guess if I was selling aircraft carriers, it would be significant, but if I'm selling $1000 cpus, it isn't. Before you jump all over me, this is called hyperbole, which is an accepted method of making a point.

Dave, I don't know about you but I'm concerned about truth here, not some fallacious distortion of the truth that is of no value to the issue at hand but to uphold one's personal bias (and its really not even an issue anymore as I've already proven my position). Theres absolutely NOTHING insignificant about a $1.2+ billion increase in sales, I'm afraid that your sadly mistaken if you really think that you have a point with this.

 

[Kagetenshi]

Okay, so basically, you trying to tell me that there isn't demand for Itanium, NOT because Itanium didn't see considerable growth (because you yourself concede that it saw overwhelming growth both quantitatively and proportionally) but rather because POWER and SPARC saw only *MINOR* declines in sales (and even then, you completely ignore the fact that the market as a whole saw considerable growth). I'm sorry thats just completely idiotic, the very fact that Itanium sales went up so significantly over last year PROVES that the demand has gone up and is going up. I mean seriously, are you a moron? Do you even know what demand is?

No, I'm saying that growth in Itanium combined with a tiny contraction in Power provides absolutely no information on "where demand is heading". If this year we see similar numbers then we might have that information. If they continue the following year then yes, we have that information. But apparently mister oh-so-smart here can determine trends from a single event

~J

 

[jcrowe]

How does Itanium relate to Macs???

I will agree that the Itanium 1 was not well received. The itanium 2 has been though. Intel is selling a lot of them. They are also releasing dual cores this year with massive cache amounts 9 to 12 Meg. The Itanium is really starting to pick up steam now. 13 of the TOP 100 systems are using Itanium 2. Compare that to 2 years ago when there were zero.

Itanium must be viewed as a collosal failure from the standpoint of both Intel and Hewlett Packard who have dumped billions of dollars into its development. Further, HP axed its own, popular PA-RISC ISA and merged with Compaq, which axed Alpha. HP has scaled back its Itanium offerings to servers only. Intel has a record of this type of failure....they had a similar experience back in the 80s when I worked for them...IAPX432....a neat concept but a failure in the real world. IBM has recently backed away from Itanium. I'm not intending to dump on the quoted poster, but Itanium has not earned many kudos from anybody. Further, given that the future of the Mac does not include Itanium, I think it's difficult to justify discussion here, unless it's related to how XServe will compete against Itanium servers.

 

[daveL]

Itanium must be viewed as a collosal failure from the standpoint of both Intel and Hewlett Packard who have dumped billions of dollars into its development. Further, HP axed its own, popular PA-RISC ISA and merged with Compaq, which axed Alpha. HP has scaled back its Itanium offerings to servers only. Intel has a record of this type of failure....they had a similar experience back in the 80s when I worked for them...IAPX432....a neat concept but a failure in the real world. IBM has recently backed away from Itanium. I'm not intending to dump on the quoted poster, but Itanium has not earned many kudos from anybody. Further, given that the future of the Mac does not include Itanium, I think it's difficult to justify discussion here, unless it's related to how XServe will compete against Itanium servers.

Well, at least somebody agrees with me. I was actually on a project that attempted to use the iAPX432. It had context switching latencies measured in milliseconds; not real good.

 

[JCheng]

No, I'm saying that growth in Itanium combined with a tiny contraction in Power provides absolutely no information on "where demand is heading". If this year we see similar numbers then we might have that information. If they continue the following year then yes, we have that information. But apparently mister oh-so-smart here can determine trends from a single event

~J

Fallacious logic does nothing to help your baseless argument Kagetenshi. How can you define a year as “a single event” when clearly a year is in itself, a span of time that defines trends and demand. If you were for one moment to leave your imaginative world of wishful thinking and face reality, you'd realize that demand is driven by the current situation of the market and not some idiotic linear graph of revenues over x number of years. By your “logic”, a product that has been selling enormously well this year (thus reflecting on heightened demand over the past), might suddenly plummet the next year, just because it’s "the next year” (and no, you haven’t provided a single shred of evidence to support your position as of yet).

Clearly, you understand nothing about Itanium’s market situation since even if we WERE to believe your absurd assertion, we’d still come to the conclusion that demand for Itanium is growing for the mere reason that Itanium sales has been steadily increasing over at least the past three years.

 

[daveL]

Fallacious logic does nothing to help your baseless argument Kagetenshi. How can you define a year as “a single event” when clearly a year is in itself, a span of time that defines trends and demand. If you were for one moment to leave your imaginative world of wishful thinking and face reality, you'd realize that demand is driven by the current situation of the market and not some idiotic linear graph of revenues over x number of years. By your “logic”, a product that has been selling enormously well this year (thus reflecting on heightened demand over the past), might suddenly plummet the next year, just because it’s "the next year” (and no, you haven’t provided a single shred of evidence to support your position as of yet).

Clearly, you understand nothing about Itanium’s market situation since even if we WERE to believe your absurd assertion, we’d still come to the conclusion that demand for Itanium is growing for the mere reason that Itanium sales has been steadily increasing over at least the past three years.

I really think you need to take a chill pill. If you want to think that Itanium is the greatest thing since sliced bread, great.

 

[Kagetenshi]

Fallacious logic does nothing to help your baseless argument Kagetenshi. How can you define a year as “a single event” when clearly a year is in itself, a span of time that defines trends and demand. If you were for one moment to leave your imaginative world of wishful thinking and face reality, you'd realize that demand is driven by the current situation of the market and not some idiotic linear graph of revenues over x number of years. By your “logic”, a product that has been selling enormously well this year (thus reflecting on heightened demand over the past), might suddenly plummet the next year, just because it’s "the next year” (and no, you haven’t provided a single shred of evidence to support your position as of yet).

Clearly, you understand nothing about Itanium’s market situation since even if we WERE to believe your absurd assertion, we’d still come to the conclusion that demand for Itanium is growing for the mere reason that Itanium sales has been steadily increasing over at least the past three years.

*Thumbs up* Way to straw man. You haven't provided any evidence to support your position either there. You quoted one year's sales numbers. That means Itanium did well last year. That isn't the "direction demand is heading". A product that has been selling enormously well this year might indeed plummet the next year, just because the people who want them just bought them the previous year. I'm not saying that that is happening, but extrapolating from a single year is one of the most mindlessly stupid things you can do.

~J

 

[ddtlm]

Hey what do you all say we try to end this sidetrack. Presumably everyone can agree that:

1. Calling each other stupid is a bad idea.
2. 2004 was a pretty good year for Itanium.
3. Only time will tell where it goes from here.

 

[blitzkrieg79]

Who gives a damn about Itanium when all the Mac addicts are interested in the next PowerPC technology... Anyway as I did some reasearch last nite it seems I dont know anything about any upgrades like the most of you lol Anyway, all I can say is, APPLE I got the money ready so please positively surprise me as I am waiting for the Power5 derivative of 97x (or whatever u will call it) processor... From what I understand the current 970 can't be clocked any higher than 2.8GHZ so I am also guessing that the current wait is all about the rumored Power5 derivative... Anyway, there are too new many features in Tiger such as Core Video and such that I think the next PM update will be rather more significant than just a "mere" speedbump... Maybe an on-chip memory controller, maybe pci-express, maybe ddr2, maybe dual cores... I am more than certain I dont know what it will really be but I know that we will possibly see something rather big, its just toooooo quiet right now and if Apple wanted to do just a speedbump or a 512kb cache increase I am more than sure they would have done it in January, as much as I dont like Steve Jobs lately, I know that the year of video speech wasn't just for the hell of it... in my view, in order to produce/edit any kind of hd video you definately need more than even dual 2.5ghz powermacs... So I think we might see a totally new processor architecture as well as a totally new case design and such... its been almost two years plus I know Jobs wants to redeem himself for last year... Even if they cant get past thru 3ghz they can always introduce a new processor design that might clock similarly but it will be much more efficient...

 

[wdlove]

Hey what do you all say we try to end this sidetrack. Presumably everyone can agree that:

1. Calling each other stupid is a bad idea.
2. 2004 was a pretty good year for Itanium.
3. Only time will tell where it goes from here.

Each persons opinion should be respected. Personal attacks are unacceptable. Each persons guess is just as good as another.

 

[Lacero]

Itaniums use too much power and Intel took way too long developing it. IBM's Power5 architecture is where SMP and dual-core technologies are heading.

 

[d.perel]

Just Imagine the heat that a quad processor unit is going to create. It will be insanely fast though.

Kornyboy

You wouldn't need central heating if you put it near an Air vent

 

[Surreal]

You wouldn't need central heating if you put it near an Air vent

but what to do in summer?


and does anyone know whether an underclocke MP wouold be a good choice for a powerbook?

i know, i know, the powerbooks are "fine" as they are...but i really am waiting for the next mobo redesign. i am going to use the computer for school, audio work, and some occasional video work.

i mean...what would the wattage and heat be like in a 1.75ghz 970MP?

 

[isgoed]

The get the feeling that Apple will first offer a single dual core
somewhere near the 3 Ghz threashold for $2999.

They could still surprise us with a Power5 to achieve that goal.

A QuadraMac might follow later for closer to $3999.

If it includes everything the high end user wants,
then that price would not be terribly unreasonable.

This would still be $1000 lower than the current Alienware ALX


IF however, IBM is able to manufacture these chips
in large enough quantities, then the prices may be lower.

The most important issue to me is that any new motherboard
is capable of supporting a variety of options allowing the user
to customize their systems as needed.

I think that the powermacs don't necessary will get an price increase. I did some googling and found:

Proc -=-=-=-=-=-=- Die Size -=- Transistors
Pentium Prescot -=- 112 mm2 -=- 125 million
Athlon 64 -=-=-=-=- 193 mm2 -=- 106million
Cell -=-=-=-=-=-=- 221 mm2 -=- 234 million
Itanium Montecito -=- ? -=- 1.7 billion
970FX -=-=-=-=-=- 66 mm2 -=- 52 million
512kb Cache -=-=-=- ? -=- ~55 Million?

As you can easily see, you can double the core of a 970FX and still have less transistors than any other modern processor.
Of course are the costs also dependent on processor development, yields and fabricagion facilities, but those will likely be lower than thos of the original PPC970.

Of interest is that an engineering sample of the Athlon 64 surfaced today. It features two processors on 2.4GHZ!!!!, If this is any indication for how well IBM's chips clock, we might expect pretty high clocked 970's.

 

[relimw]

ClimbingTheLog:


Almost all the threads running on your machine at any one time are actually sleeping (waiting for user input, and other events). I'm not the first one to say this, but more cores won't make them sleep any faster.

Uh, obviously you're underutilizing your Mac then
My dual 500 is not only running half a dozen server tasks (web, email, etc) but also is setup for xgrid uses, and I normally have at least 8 user/gui programs running at once.

For example, from 'top' as I write this:
Processes: 61 total, 11 running, 50 sleeping... 160 threads 12:45:34
Load Avg: 2.65, 2.50, 2.36 CPU usage: 88.3% user, 11.7% sys, 0.0% idle
SharedLibs: num = 130, resident = 36.9M code, 3.31M data, 10.7M LinkEdit
MemRegions: num = 9656, resident = 519M + 12.6M private, 166M shared
PhysMem: 130M wired, 643M active, 297M inactive, 1.05G used, 207M free
VM: 5.11G + 86.1M 77768(0) pageins, 13727(0) pageouts

 

[ddtlm]

relimw:

Heh, sounds like you might want a quad.