New PowerMac Model?

[Freg3000]

They won't say a word about missing the mark because the mark is "end of summer" 2004, e.g. around Sept 21, 2004 +/- a few days.

Steve revised his "within a year" statement from June 2003 in Sept of 2003, as has been beaten to death numerous times.

Whether or not your should be allowed to revise those types of projections is another story completely, but think about this for a moment. What do you think is more publicly known?
A) Steve's bold prediction of 3 Ghz within 12 months announced on on stage during the WWDC keynote.....or
B) Steve's adjustment to his claim at......Paris Expo? I don't even know. And I am on this site everyday!

Fact is, most people probably think that the mark still is this year's WWDC, but if Steve has no 3 Ghz Power Macs to give us, I doubt he'll say a word, leaving many people confused.

Although, he probably will claim that Apple did meet "the mark" if they get to 3 Ghz by September.

Nonetheless, I am sure that we both can agree that a 3 Ghz Power Mac sure sound goods.....whenever it comes!

 

[Wyrm]

this is purely speculation and should be taken with a grain of salt, but since the current PM is 7,2 and this "new model" is 8,1 it may be proof that this new model is in fact using the 975 chip as previously rumored since they jumped from 7 to 8. I have no idea what I'm talking about though so if i'm way off base, slap me silly....but gently...

I hope you're right, but I think you're wrong.

The 970 was a well known processor before it was even announced that Apple was going to use it (IBM let out all sorts of juicy info).
The 970fx is also a known entity, being the 90nm shrink of the 130nm 970.
The 975,980 etc is not known... no details... only what can be infered from the Power5 or <gasp> rumors.

We had Bernie Meyerson (CTO IBM Microelectronics, aka the guy who knows the PPC numbers) say scaling is dead for the 90nm node (in Prague)EETimes Link .

So IBM has some *issues* at Fishkill with 90nm: Again, known.

If we take Bernie's comment of 70% innovation required for the 90nm node, we can guess that the 970 required significant re-work, and would probably sap engineering resources from any other manufacturing project at 90nm (a guess, but the economic principle of IBM having finite resources is not debated)

Now, this doesn't rule out that the Power5 lite could be a 130nm part... (the Power5 is), or IBM has a skunkworks 90nm project for the 97x/98x. These would be surprises, although a 130nm Power5 Lite would probably not hit 3Ghz unless it increases the pipeline length like the Prescott, has double pumped ALUs, or employs some other sort of slight of hand.

The 970 is ~120 mm2 which, for example, is a little more than half the surface area of an AMD Opteron, so there's room to grow even at 130nm. Of course if it has changed significantly then IBM has to re-verify the processor... which then it might be ready for next year's WWDC if they started last year... and IBM hasn't released any sampling information, or any volume production information for a 97x series.

I think the only thing IBM has that could hit 3Ghz, and is in production, would be the 970fx.

And who knows, for Steve Jobs to eat his words, the Summer has to end...
(And Steve lives in California... it's always Summer there right?)

-Wyrm

 

[aswitcher]

They won't say a word about missing the mark because the mark is "end of summer" 2004, e.g. around Sept 21, 2004 +/- a few days.

Steve revised his "within a year" statement from June 2003 in Sept of 2003, as has been beaten to death numerous times.

End of Australian Summer is Feb 28 2005

Thats probably the earliest we will get it

 

[Wyrm]

What is your explanation for this roadmap clearly stating the POWER4 (Code GP) and then showing the PPC 970 (GP-UL). Now explain how we have the POWER5 (Codename GR) and it's "97x" (GR-UL) is slated for 2H 2004 at 2.5Ghz+.

I don't see how you get 2H 2004 for the 97x GR-UL from the slide, since by the same logic the GP-UL would have been released in the beginning of 2003.


I hope the GR-UL rears it's head at WWDC, but I think we are going to get the GP-UL FX (maybe a '+' in IBM notation?)

-Wyrm

 

[nuckinfutz]

I don't see how you get 2H 2004 for the 97x GR-UL from the slide, since by the same logic the GP-UL would have been released in the beginning of 2003.


I hope the GR-UL rears it's head at WWDC, but I think we are going to get the GP-UL FX (maybe a '+' in IBM notation?)

-Wyrm

Yes I'm beginning to think WWDC is not going to see the fabled 3Ghz announcement. They still have until mid Sept to announce that machine and keep Steves promise. The diagram is definately sloppy. According to this it looks like the GR-UL could come at "any" time in 2004 just as it shows the GP-UL could have come at "any" time in 2003.

If a 3Ghz isn't announced I won't be heartbroken. I'll look to see the performance of the top Powermac and then we can see how much another few hundred megahertz will get us.

 

[Macmaniac]

I Love how AppleInsider chimes in to claim that the new powermacs will be coming in June, its like throwing more gasoline onto an already burning forrest fire of rumors.

 

[Wyrm]

Yes I'm beginning to think WWDC is not going to see the fabled 3Ghz announcement. They still have until mid Sept to announce that machine and keep Steves promise. The diagram is definately sloppy. According to this it looks like the GR-UL could come at "any" time in 2004 just as it shows the GP-UL could have come at "any" time in 2003.

If a 3Ghz isn't announced I won't be heartbroken. I'll look to see the performance of the top Powermac and then we can see how much another few hundred megahertz will get us.

Presentation slides are intentionally unclear.
For marketing reasons... ("What was that? We didn't get that CPU out in the timeframe that was on the slide? Oh, you mean that slide.. well <sound of chuckling> I think it's an optical illusion, yeah... no no not this year... sampling, yeah right, *sampling*...")

Yeah, 3Ghz would be very impressive (but so would a dual 2.5Ghz for that matter), and a 970fx might be able to hit that, so you still may be surprised. I just don't think we are going to see the Power5 UL one (so no SMT, etc).

-Wyrm

 

[Wonder Boy]

Given Apple's recent comments, I think Tiger in October is not what will happen.

i agree thats why i said october(ish).

i figure jaguar was in august->panther in october->tiger in december or january's expo.

~ 2 months apart

 

[mountainmac]

New Imac

According to what I've read so far the 8,1 indicates a new Imac. This may have been mentioned but all of the even numbers have been Imac updates and odd numbers have been power mac updates

2,x - IMac
4,x - IMac
6,x - IMac
8,1 - ????

3,x - Powermac
5,x - Powermac
7,x - Powermac

This was shown on one of the sites listed on anothe site like macbytes or something like that with a link to applemuseum.

Just some more fuel for the fire.

 

[Stike]

According to what I've read so far the 8,1 indicates a new Imac. This may have been mentioned but all of the even numbers have been Imac updates and odd numbers have been power mac updates

2,x - IMac
4,x - IMac
6,x - IMac
8,1 - ????

3,x - Powermac
5,x - Powermac
7,x - Powermac

This was shown on one of the sites listed on anothe site like macbytes or something like that with a link to applemuseum.

Just some more fuel for the fire.

I also noticed that. My money is on a new iMac. G5? Hopefully! The performance is currently a letdown, and the iMac is currently the worst performing model (saleswise AND CPU-wise)...

 

[Sun Baked]

According to what I've read so far the 8,1 indicates a new Imac. This may have been mentioned but all of the even numbers have been Imac updates and odd numbers have been power mac updates

2,x - IMac
4,x - IMac
6,x - IMac
8,1 - ????

3,x - Powermac
5,x - Powermac
7,x - Powermac

This was shown on one of the sites listed on anothe site like macbytes or something like that with a link to applemuseum.

Just some more fuel for the fire.

Don't look at it as a iMac vs. PowerMac thing -- look a bit deeper to chipsets used in these things.

There is a good chance there is a new chipset coming. Good chance it is a consumer machine, might not be.

Anyway Apple started mixing up the odd/even stuff used to keep pro/consumer portable lines apart.

It could happen with the desktops.

 

[eyeluvmyimac]

Update: According to Appleinsider, the new PowerMac should be released by the end of June.

<extreme sarcasm>wow good call apple insider, where did you ever come up with a date like that!??!? they must have someone working on the inside...</extreme sarcasm>


'ya think?

 

[Wyrm]

According to what I've read so far the 8,1 indicates a new Imac. This may have been mentioned but all of the even numbers have been Imac updates and odd numbers have been power mac updates

2,x - IMac
4,x - IMac
6,x - IMac
8,1 - ????

3,x - Powermac
5,x - Powermac
7,x - Powermac

Just some more fuel for the fire.

Yep, we don't know since they are both listed as "Power Mac" for the IMac, eMac, and the PowerMac... why, we could be looking at something totally new... or maybe someone was just putting in the number after 7?
The codename is "Image" which doesn't exactly bring the *image* of a supremely power computing platform, but rather the idea of a smart mirror or something . I thought the *French* rumors cited the codename as "Trinity" - YAMCN.

Even the CPU info has changed, whatever a "SMU_Neo2" is..

m'gosh, maybe it's the XBox2 that runs OS-X????

Only 26 days until WWDC... which will probably leave more questions unanswered (like ship dates)

-Wyrm

 

[diego]

Yep, we don't know since they are both listed as "Power Mac" for the IMac, eMac, and the PowerMac... why, we could be looking at something totally new... or maybe someone was just putting in the number after 7?
The codename is "Image" which doesn't exactly bring the *image* of a supremely power computing platform, but rather the idea of a smart mirror or something . I thought the *French* rumors cited the codename as "Trinity" - YAMCN.

Even the CPU info has changed, whatever a "SMU_Neo2" is..

m'gosh, maybe it's the XBox2 that runs OS-X????

Only 26 days until WWDC... which will probably leave more questions unanswered.

-Wyrm

Trinity was the codename for the Cube, Neo was the codename for the G5 (970), so this could mean 975

 

[tortoise]

Blue Gene isn't SMP

So why is IBM's Blue Gene based on the PPC 440? Is it really the PPC that's not SMP friendly, or the chipset?

Blue Gene isn't SMP. It is a fine-grained message passing architecture, kind of a modern remake of the old 1980s massively parallel supercomputers, with gobs of relatively simple processors. In a sense, it is basically a computing cluster that has been cut down to fit in a box, and with the same limitations on what it is useful for.

The poor multiprocessor scalability isn't intrinsic to the PPC, it was an explicit design choice. The Power series processors actually do have all the glue logic and memory hardware to scale extremely well in multiprocessor environments. These features were probably not included for two reasons. First, it was likely a cost saving measure by Apple and IBM, as vanilla SMP systems are much less costly to produce. Second, IBM may not want to cannibalize the sale of their Power systems. The additional features don't come cheap; with its huge cache, onboard memory management, and ccNUMA, the Opteron has a substantially larger die size than the PPC970 for the same fab process which translates into cost.

Apple may eventually sell a scalable version of the PPC processor, but they have a lot of work to do to get there, including coming up with a scalable operating system (OSX is hampered by a crusty old ABI that they have to support). If it ever shows up, it will show up in the XServe first, but I don't see that happening any time soon as a practical matter -- all these changes would take years if they are currently planned at all.

So in the meantime, Apple will be making a killer workstation and a decent low-end server box. Nothing wrong with that.

 

[bathysphere]

 

[thogs_cave]

We agree.

The poor multiprocessor scalability isn't intrinsic to the PPC, it was an explicit design choice. The Power series processors actually do have all the glue logic and memory hardware to scale extremely well in multiprocessor environments. These features were probably not included for two reasons. First, it was likely a cost saving measure by Apple and IBM, as vanilla SMP systems are much less costly to produce. Second, IBM may not want to cannibalize the sale of their Power systems. The additional features don't come cheap; with its huge cache, onboard memory management, and ccNUMA, the Opteron has a substantially larger die size than the PPC970 for the same fab process which translates into cost.
...snip...
So in the meantime, Apple will be making a killer workstation and a decent low-end server box. Nothing wrong with that.

I don't disagree with you at all. I was responding more to the individual who made the sweeping claim that SMP was not good for much, with the "50%, 33%, 25%" logic.

The G5 is good for what it does. And the overhead I see on the Dual 2.0 I have is about 15-ish % (sometimes more, sometimes less), which seems much better than 50%...

It is a killer UNIX workstation. It works quite well, despite netinfo. As for servers, Xserves are cool, but if I'm really pushing it, I want a Sun running Solaris. No reason why the two can't live together.

 

[JFreak]

Huh? Using your own logic and numbers, a quad CPU system would realize ~208% speed increase over a single CPU system (i.e. 100% + 50% + 33% + 25%).

bad math.

1.00 * 1.50 * 1.33 * 1.22 = 2.49 --> original plus 149%

 

[JFreak]

it's not quite that inefficiant as that http://www.barefeats.com/xserve2.html

i was just making an example, not putting hard facts on the table. but thank you, very nice link.

 

[JFreak]

~100% speed increase.

almost and exactly are two different things. but you're right, in the real world some 95% increase can very well be considered to have "doubled" the cpu power.

i was just stating an example, not facts, but i'm glad people write good and efficient software that put theories to shame

 

[JFreak]

I think Virginia Tech showed the world that Mac's scale better.

thank god for that actually, it is the underlying unix that performs well.

 

[Wyrm]

thank god for that actually, it is the underlying unix that performs well.

It clusters well.

Just think, you could spend 30-50 times more and get one of these:
The FASTEST MACHINE IN THE WORLD!! HA HA HA HA

<sorry, too tempting to show a flagrant waste of tax money>

I helped pay for that I... er I'm still helping to pay for that... along with 60 odd million other people in this country (oooh and to think they could have just bought 10,000 XServes)...

-Wyrm

 

[Unclezeppy]

iMac G5 has been in production for almost a month, 1.8 and 2.0GHz models. They still looked the same (cam phone pic) but might have white egg shell that can change color, each user can have their own color when they log in or you can match the color with you home decor (just my guess based on their patents, what would be cool and couple of other things).

So there, too bad, I live close to the factory but can't buy them until US market gets them.

 

[klaus]

iMac G5 has been in production for almost a month, 1.8 and 2.0GHz models. They still looked the same (cam phone pic) but might have white egg shell that can change color, each user can have their own color when they log in or you can match the color with you home decor (just my guess based on their patents, what would be cool and couple of other things).

So there, too bad, I live close to the factory but can't buy them until US market gets them.

sure, and if we don't believe this story, you'll tell us another one

lame attempt, you could've at least shown us that cam pic you were talking about

 

[rdowns]

anyone think that this means the end of the power mac line.
there have been no revisions to the g5 in months…
if the imac g5 is released this would bring the imac to the same level as the power mac but at a lower price point (even more so if apple keeps the all-in-one desing)
what I think we may see… a stand alone imac with a bit more expandability (meaning more ram, maybe a few PCI slots) with a G5 processor.
No Power Mac updates.

Crazy talk! End of the PM? Pass the joint dude.

I have no doubt that the PMs will get a speed bump and think Steve will meet his 3 GHz goal even if it is only an announcement with shipping in 60-90 days.

Let's say the iMac gets the G5 (please!!!!) and is on par, speed wise, with the PM. There is still a lot of other items to diffrentiate the systems such as

Single vs. dual processors
RAM expansion
Drive bays
Graphics cards

I don't see the iMac changing in terms of expandability; RAM is it. While I'd love to see a small form tower, Apple won't do it. They'd be afraid of canibalizing PM sales.