May 18th? G5 Updates?

[DaveClarkOne]

I really hope Apple does update the G5, and quickly. It would be great to get a discount on the Dual 2.0 GHz, and be able to take advantage of the Brilliant Savings promotion, which alone is worth it to get an 'older' G5.

The 23" HD Cinema display rocks; and, I still, for the love of G*d do not understand why folks are so uptight about Apple revising the current display schemes to 'match' the G5s' aluminum motif.

I've said it before and I'll say it again, matching is out, combining dissimilar products are in for the fashion world, and in general. The current clear plastic displays DO match the aluminum G5's. Anyone who says they are an eyesore is on crack.

If I catch your meaning, you have a preference for the existing cinema displays? Actually, I agree. A very modest refresh of the displays would be ok, but don't underestimate Apple's ability to move the bar higher than where the current displays set it (very high).

 

[ddtlm]

thatwendigo:

2003 Revenues:
IBM $89,131,000,000
Intel $30,100,000,000

2003 Net Income:
IBM $7,584,000,000
Intel $3,100,000,000

Who doesn't have what?

Consider the money that IBM takes in from processors and clearly Intel is larger. It might even be the case that IBM is loosing money on their CPU division. On the other hand, despite Itanium, despite Prescott, despite P4's in general, despite AMD's best efforts, Intel is raking in profits.

Ah, right... So the most successful PC chipmaker shouldn't have developed the Centrino, because the Penitum4-Mobile could fill in the laptop market?

Intel can afford multiple, specialized processor lines. IBM can't.

That makes absolutely no sense. Technology advances, things get faster, and you need to design around them. It wouldn't be another FSB, so much as the one that would supplant the one that's choking the G4 as it stands. If you rid yourself of a chip in favor of another, then you merely replace its needs with the newer ones.

Going to G5's without this 750vx involved saves Apple one FSB to design for. Presumably the G5-compatible chipset for a PBook is much like one for an iMac.

 

[Borg3of5]

You are correct, sir!

If I catch your meaning, you have a preference for the existing cinema displays? Actually, I agree. A very modest refresh of the displays would be ok, but don't underestimate Apple's ability to move the bar higher than where the current displays set it (very high).

I do like the current displays. They go well together with the aluminum of the G5's design. I don't doubt, however, that Apple will 'wow' us with something revolutionary with displays. I have a friend in graphic design, who says he'd love to have a 30" display. I think the 23" is humongous already.

The only thing I hope is that even though the Brilliant Savings promo ends the day before WWDC, when the new G5's will be announced, the $500 price cut will hold for future purchases of the 23" display. I'd love to get a dual 2.0-2.4, 8x DVD-/+R drive, or Dual-Layer burner, with the current 23" display. I know that's just wishful thinking.

We'll see. Next February I'll be able to visit the SF Apple Store. The bastards only opened it a few weeks after my visit. I about fainted when I went walked past the the Virgin Megastore off Market and Stockton, and saw the white Apple logo. Woo-hoo!!!

 

[thatwendigo]

Consider the money that IBM takes in from processors and clearly Intel is larger.

No, not really. IBM as a company is enormous and has some of the mosst prodigious resources in the world. They file for more patents per year than any other company, and as I recall, also hold the single highest number. Saying that you should only consider IBM's chip section is like saying that you should only compare Microsoft's Office department to any competitor. There's nothing to keep them from rolling money in off off other sources if they decides to.

How do you think HP and Sony sell their computers relatively cheaply?

I might even be the case that IBM is loosing money on their CPU division. On the other hand, despite Itanium, despite Prescott, despite P4's in general, despite AMD's best efforts, Intel is raking in profits.

And AMD is gaining, so is IBM, and if the stories are true, they've got Freescale nipping at their heels in the embedded market while also preparing a chip that could very well swing into the laptop market.

Intel/Microsoft have made a lot of people uhappy over the years. Some of them are now working together, and the fruits of that labor are still in the making.

Intel can afford multiple, specialized processor lines. IBM can't.

Based on what? IBM has far more money to throw around than Intel does.

Going to G5's without this 750vx involved saves Apple one FSB to design for. Presumably the G5-compatible chipset for a PBook is much like one for an iMac.

What in the world would make you say that? The iMac has more than a few luxuries that the PowerBook doesn't, not the least of which is a steady, constant power supply that doesn't drain out, and a larger, more open form factor to pack the parts into.

Apple is far better off with a low power chip in the portables, unless you'd like to see PowerBooks that weigh ten pounds.

 

[ddtlm]

thatwendigo:

No, not really. IBM as a company is enormous and has some of the mosst prodigious resources in the world. ... Saying that you should only consider IBM's chip section is like saying that you should only compare Microsoft's Office department to any competitor. There's nothing to keep them from rolling money in off off other sources if they decides to.

In the real world, companies operate to make money. This means that IBM won't be dumping cash into their CPU division, at the expense of their other divisions, just to wage a silly war so that Apple can win. Which is really what everyone seems to think IBM should be doing. In this "real world", if IBM doesn't see a way that designing a 750vx is going to make them a profit one way or another, they won't do it. What someone here needs to do is establish some way that this 750vx has a snowball's chance in hell of turning a profit, despite being heavily redesigned (R&D cost), despite its new FSB and the logic board and system controller changes that requires, despite going into low-margin computers and embedded devices, and despite going directly into a market that Moto is supplying with a reasonably compeditive product lineup. People talk about how IBM could do this, and could do that, but they don't step back and consider what actually makes sense. Someone needs to make a case why the 750vx makes sense.

Based on what? IBM has far more money to throw around than Intel does.

Consider how good it is for IBM that people aren't up there "throwing around" money to wage strange personal wars.

What in the world would make you say that? The iMac has more than a few luxuries that the PowerBook doesn't, not the least of which is a steady, constant power supply that doesn't drain out, and a larger, more open form factor to pack the parts into.

Sheesh, you act like a little system controller chip is some kind of massive obsticle to packaging. There is nothing keeping Apple from making a system controller suitable for a PBook and then putting the same exact thing in an iMac. Most of their requirements are the same: one FSB interface, one DDR RAM interface, one AGP interface, lots of integrated periphrials, small packaging requirement.

Apple is far better off with a low power chip in the portables, unless you'd like to see PowerBooks that weigh ten pounds.

Uh, the system controller is not all that hot.

 

[thatwendigo]

In the real world, companies operate to make money. This means that IBM won't be dumping cash into their CPU division, at the expense of their other divisions, just to wage a silly war so that Apple can win.

It's not just between Apple and Microsoft at this point, and you misunderstood me if you at all got that impressions. Micrososft and AMD, at the very least, are interested in taking some serious shots at the Wintel hegemony, and Apple certainly wouldn't mind soaking up some the fallout. In fact, one might say that certain characteristics of Apple's recent strategy look like preparations for opening shots across the bow:

1) Open source development channels, where the core of the OS is freely available on both platforms, minus certain proprietary technologies.
2) Replacement of Microsoft's key products in several areas - OS, Safari, Mail, etc.
3) Aligning with companies that are working LotD (linux on the desktop).

IBM now has all the parts necessary to give a serious run at PowerPC LotD, if they can just get a distribution that would be workable for others. They have a competitive chip, the manufacturing wherewithall to do the assembly. All that's lingering on the backburner is an OS that could be thrown at the corporate world a little more easily than current incarnations of Linux.

Darwin with a graphical layer, perhaps?

In this "real world", if IBM doesn't see a way that designing a 750vx is going to make them a profit one way or another, they won't do it. What someone here needs to do is establish some way that this 750vx has a snowball's chance in hell of turning a profit, despite being heavily redesigned (R&D cost), despite its new FSB and the logic board and system controller changes that requires, despite going into low-margin computers and embedded devices, and despite going directly into a market that Moto is supplying with a reasonably compeditive product lineup.

According to MacPhisto, Apple was footing the R&D bill. As such, IBM loses nothing in the process, and gains new IP with the chip design and the process, not to mention the option to use the processor for whatever they decide to throw it in.

How is this a losing situation, again?

People talk about how IBM could do this, and could do that, but they don't step back and consider what actually makes sense. Someone needs to make a case why the 750vx makes sense.

It lets IBM fab their own laptop chip to be used when they take a shot at Intel/Microsoft. That's all the argument that needs to be made.

Consider how good it is for IBM that people aren't up there "throwing around" money to wage strange personal wars.

It's neither strange, nor personal, though there is some degree of the latter. Microsoft screwed IBM on OS/2 about fifteen years ago, and the company has been steadily churning along since then as a mostly hardware and enterprise company. I think they're just about ready to drop a bomb in the next few years, though, one that could very well shake the MS foundations pretty hard.

Sheesh, you act like a little system controller chip is some kind of massive obsticle to packaging. There is nothing keeping Apple from making a system controller suitable for a PBook and then putting the same exact thing in an iMac.

Cost? The fact that the ASIC in G5 towers has a heatpipe on it, on top of the nine fans? That it's as hot as the G4 that's currently in the laptops? That a "heavily redesigned (R&D cost), ... new FSB and ... logic board and system controller" are all things that will jump the heat profile like crazy and most likely require even more engineering than the VX?

Nah.

Most of their requirements are the same: one FSB interface, one DDR RAM interface, one AGP interface, lots of integrated periphrials, small packaging requirement.

Yes, and one is low-power, low-heat and scales even lower. The other, however, is still having to be worked on just to get the processor down to what the whole processing budget for the laptop is. A 1.6 G5 hardly does better than the 1.5 G4 at the moment, and yet it's quite a bit hotter and uses much hotter components for FSB, memory, and processing.

Uh, the system controller is not all that hot.

Let me reiterate: It has it's own heatpitpe.

You know what my eMac uses? One heatpipe. My processor, GPU, and Superdrive all cool off of a single fan and one pipe, which is what Apple had to stick on the ASIC for the G5.

 

[oingoboingo]

Cost? The fact that the ASIC in G5 towers has a heatpipe on it, on top of the nine fans?

I've been taking a look at this G5 pull-apart:

http://homepage.mac.com/dabaer/PhotoAlbum8.html

trying to figure out how the ASIC cooling system might actually work. The G5 ASICs are mounted on the back-side of the system board (ie: on the opposite side to the CPUs, PCI slots, RAM slots etc). From examing the two attached images (the back and front sides of the G5 system board), it looks like there is a major chip located just back from, and mid-way between the two banks of RAM slots, which is covered by a heatpipe assembly. There is another chip further down and back, just forward of the AGP and PCI slots which is covered by a passive heatsink. And then directly above the AGP slot and just behind one of the CPU connectors, there is another chip, with a 9-blocked passive heatsink attached. From cross-checking the front and back photos, I'm not sure there is anything major mounted under the large flat part of the heatpipe assembly, located right between the two RAM banks. There is some kind of ferrite core on the front side of the board and some patch wires running *through* the board, so maybe this is something to do with power regulation.

I don't really see any passage of the heatpipe conductors through the board itself to the CPU heatsinks located on the front side of the board, but I could very well be wrong there. When I shine a torch into the front of my own 1.6GHz G5 system, it looks like the distance from the 'back side' (ie: the side of the case which does not open) of the case to the system board is about the thickness of my index finger (ie: roughly 15mm). It appears that the two sets of fins on the heatpipe are in direct, or very close contact with the actual aluminium casing. From feeling my way around the case, the warmest parts of the case certainly correspond to the approximate locations of the heatpipe fins.

That said, the back-side of the case is barely warm to the touch. I don't have any type of contact heat probe here, so I can't be more accurate than that, but 'tepid' might be a good adjective to describe it. Most of the area of the case forward of the Apple logo is almost cool to the touch (the ambient temperature is around 20 degrees celsius).

OK...so my theory is that the G5 ASICs (or whatever the chips are on the rear-side of the motherboard) are cooled by passively dissipating their heat through the actual G5 aluminium case itself, rather than any type of active fan cooling, or by heat-pipe transferring heat through the motherboard over to the CPU heatsinks. Given the very moderate temperature of the back-side of the G5 case (the system has been up and running for about 4 hours this morning), I would also hazard a guess that the amount of heat produced is not excessive.

Any opinions on my amateur analysis?

 

[thatwendigo]

I've been taking a look at this G5 pull-apart:

http://homepage.mac.com/dabaer/PhotoAlbum8.html

Any opinions on my amateur analysis?

Actually, most of what you've said seems pretty reasonable and consistent with the pictures of that dual 2.0 that's been pulled apart. However, with that being said, I don't exactly have a G5 tower here in front of me that I can test, though I can probably go to the local shop and ask them if they'd open to case and let me look at the board without touching it.

Hmmm.

We need some kind of wattage spec on the ASIC to really get a serious commentary on what it's going to be dissipating, but that's the kind of thing I find it unlikely will ever be published on the web. I know that the things I can find about my systems (MPC7450 under the hood) list the processor as 14-17w at peak, and the SuperDrive is something I'm not sure I'll be able to get a heat figure on. I do know that the two are sandwiching the pipe, though. So, just for the sake of argument, let's say that the pipe is moving 30w of heat to a single exhaust fan.

In a PowerBook, you're not going to have the frame be nearly so useful for dissipation, especially not when you're resting it on a user's lap. The chip itself is going to be 25w (if not more) and the ASIC is going to be another 20-30w (from combining my figures with your guesswork). That's already a doubling of the heat budget in the current Powerbooks, not to mention a doubling of the battery draw.

Not pretty.

 

[ddtlm]

thatwendigo:

Cost? The fact that the ASIC in G5 towers has a heatpipe on it, on top of the nine fans?

Yeah it makes more heat than most chipsets, but as the Xserve clearly shows no heat pipes are needed to cool it, just a passive heatsink with some air movement, not unlike other chipsets in 1U rackmounts. The nine fans are for the entire system.

That it's as hot as the G4 that's currently in the laptops? That a "heavily redesigned (R&D cost), ... new FSB and ... logic board and system controller" are all things that will jump the heat profile like crazy and most likely require even more engineering than the VX?

The chip in PBooks and iMacs need not be clocked as high (slower FSB's, even 1/3 multipliers instead of 1/2), nor does it need to support two FSB's, nor two channels of memory. These "hot" components are therefore half the size and clocked lower, so obviously heat would come down a lot. More than they would like in a laptop probably, but still a whole heck of a lot less expensive than designing a mostly new processor and the system controller to go with it.

According to MacPhisto, Apple was footing the R&D bill. As such, IBM loses nothing in the process, and gains new IP with the chip design and the process, not to mention the option to use the processor for whatever they decide to throw it in. How is this a losing situation, again?

If Apple is footing the bill I find this no more likely, as I've outlined before. Instead of "why would IBM do it", the question I would ask is "why would Apple do it".

It's been explained by MacPhisto as protection against Moto screwing up, which is about the only justification because obviously a 750vx goes head to head with the 74x7 chips and the upcoming e600. But the 750vx was supposed to show in March, that's what, 6 months behind when Moto delivered? So if Moto slipped more than 6 months from their already slow schedule the 750vx would "save the day". (So... how much did Apple supposedly spend on this?)

But what about this doomsday Apple was saving themselves from possibly encountering? Consider the case where Moto dropped the ball on the 7457 (more than 6 months late) and where they had no 750vx under development. With no more difficulty that they would have designing new motherboards and chipsets for a 750vx, Apple could have been putting G5's into PBooks and iMacs starting as soon as the 970fx was available, Feb 2004 according to schedule I believe. They could leave iBooks with either 180nm G4's or with G3's, and eMacs with 180nm G4's. The G5's in some PBooks (like the 12") would be quite low clocked, but hey we're pretending the 7457 never showed so they'd be replacing slow G4's. So, not a terribly bad situation for Apple if they never funded a 750vx and Moto dropped the ball.

So lets summarize what you're claiming: Apple bankrolled the development of a 750vx that competes as directly as can be with processors Moto was developing, just to make sure that if Moto slipped by more than six months, they could madly scramble together some new chipsets and motherboards and save their hide... from using the 970fx and G3's. Oh and then Apple cut funding, so all that money was wasted. But IBM didn't mind because apparently they didn't have any other customers waiting for this chip, or designing products around its fancy new FSB. Whew, quite a clever plan!

It's neither strange, nor personal, though there is some degree of the latter. Microsoft screwed IBM on OS/2 about fifteen years ago

Try taking that to a shareholder meeting.

IBM now has all the parts necessary to give a serious run at PowerPC LotD

So you figure that PPC/Linux would be compelling enough to defeat x86/Linux and x86/Windows? The only way to do that is to undercut on price because its pretty damn easy to have a "fast enough" desktop system, and people will default to what they already have laying around. Ever look at IBM's desktop prices? Last time I checked Dell had them beat, and they would be all over x86/Linux the second Linux started looking like a desktop money maker. Them and about a zillion other companies, big and small. Your expecting IBM to take on the entire x86-box industry, and defeat them on price, while keeping two processor lines up to date. I'm not seeing the profit.

Oh but maybe IBM has a clever plan to sell expensive PPC/Linux desktops. Heh.

 

[ktlx]

So you figure that PPC/Linux would be compelling enough to defeat x86/Linux and x86/Windows? The only way to do that is to undercut on price because its pretty damn easy to have a "fast enough" desktop system, and people will default to what they already have laying around.

Even then I cannot see that happening. Everyone I know who uses Linux on the desktop uses either a dual-boot configuration or something like VMware for those times they need to run a Windows application. Given the price/performance of an AMD Athlon XP and the performance of an AMD64, I just cannot see why anyone not stuck on a PowerPC platform would choose Linux/PPC.

 

[Kid Red]

OK, so since the threads been hi-jacked i figured I'd ask if any new info on the updates has become available?

 

[thatwendigo]

Yeah it makes more heat than most chipsets, but as the Xserve clearly shows no heat pipes are needed to cool it, just a passive heatsink with some air movement, not unlike other chipsets in 1U rackmounts. The nine fans are for the entire system.

Some air movement? The xServe is nearly constantly blown, active cooling. It was back when there were G4s under the hood, and it's not exactly gotten cooler with the addtion of G5s and the new subsystems. Also, the xServe has no hot GPU, which is getting to be a major factor in many systems.

Unlike the comparison that many people wish to make, the PowerBook is far more cramped and would be using much less available power and airflow. It's not all a proof that you could make a PowerBook when you compare something that's over twice as big, on a wall outlet or other steady power supply, and able to use outside cooling solutions (rack's often have secondary thermal systems).

The chip in PBooks and iMacs need not be clocked as high (slower FSB's, even 1/3 multipliers instead of 1/2), nor does it need to support two FSB's, nor two channels of memory. These "hot" components are therefore half the size and clocked lower, so obviously heat would come down a lot. More than they would like in a laptop probably, but still a whole heck of a lot less expensive than designing a mostly new processor and the system controller to go with it.

And also a whole heck of a lot less performance capable. In case you missed the discussion, I advise you to go look up barefeat's test of the single 1.5 G4 versus the single 1.6 G5 on Final Cut. Here's a hint... It very nearly gets beaten, in a desktop environment with all those things that you're talking about ditching.

So you think they ought to cut performance lower than the current systems, just to have a G5 under the hood with less battery life?

If Apple is footing the bill I find this no more likely, as I've outlined before. Instead of "why would IBM do it", the question I would ask is "why would Apple do it".

Simple. If Freescale doesn't deliver on the e600s or e700s, then Apple has a replacement processor ready to run as soon as IBM can tool a line for them. If they go to tape-out and hold there, with the design ready to go on wafer, they have a backup plan and damage control.

It's been explained by MacPhisto as protection against Moto screwing up, which is about the only justification because obviously a 750vx goes head to head with the 74x7 chips and the upcoming e600. But the 750vx was supposed to show in March, that's what, 6 months behind when Moto delivered? So if Moto slipped more than 6 months from their already slow schedule the 750vx would "save the day". (So... how much did Apple supposedly spend on this?)

Does that really matter (the cost), as long as they're looking out for their customer's interests? People have accused Apple of just taking whatever Motorola hands over, and this looks like a case of their hedging bets against a possible repeat of bad circumstances.

But what about this doomsday Apple was saving themselves from possibly encountering? Consider the case where Moto dropped the ball on the 7457 (more than 6 months late) and where they had no 750vx under development. With no more difficulty that they would have designing new motherboards and chipsets for a 750vx, Apple could have been putting G5's into PBooks and iMacs starting as soon as the 970fx was available, Feb 2004 according to schedule I believe.

Except that it still takes a redesign of logic boards, and a complete redo of their power and heat management scheme. It could even be that the G5 is impossibly at current heat levels and the form factor, and Apple's been known for their sleek laptops. They're not goin to kill that just to wedge in a processor that isn't clearly a better performer in those highly limited circumstances.

Also, it could very well be that the "dropped ball" is what convinced them they needed this backup plan.

They could leave iBooks with either 180nm G4's or with G3's, and eMacs with 180nm G4's. The G5's in some PBooks (like the 12") would be quite low clocked, but hey we're pretending the 7457 never showed so they'd be replacing slow G4's. So, not a terribly bad situation for Apple if they never funded a 750vx and Moto dropped the ball.

Except, you know, the whole G5 not outperforming the G4 thing when you cut the bus and get low enough clock. Even a full-bus, desktop G5 doesn't outrun the 7447A at 1.5ghz by all that much. If Freescale delivers a dual-core 2.0ghz e600 that Apple can use, or the VX debuts at 2.0ghz, and their total heat budget is around 20-30w, that's still far, far better than the 970fx.

So lets summarize what you're claiming: Apple bankrolled the development of a 750vx that competes as directly as can be with processors Moto was developing, just to make sure that if Moto slipped by more than six months, they could madly scramble together some new chipsets and motherboards and save their hide... from using the 970fx and G3's. Oh and then Apple cut funding, so all that money was wasted. But IBM didn't mind because apparently they didn't have any other customers waiting for this chip, or designing products around its fancy new FSB. Whew, quite a clever plan!

Mind not being quite so insulting?

The obvious point of the VX is to have the option to produce it if necessary, because Motorola does have a history of making promises they can't back up. Even if it isn't used right away, Apple could sell the rights to the chip or the IP genereated from it later on, or bring it into play if there are problems in the future. Direct competition is not suicide, as IBM has proved with the G5, and which you have stated already by talking about how the 970 will eventually be some low-power solution that can be used where the G4 is.

Why is the G5 not a stupid investment, then, since it's obviously attacking a market that Motorola has a lock on. Hell, why isn't the PowerPC a stupid investment for desktops? AMD and Intel have that all sewn up, after all!

So you figure that PPC/Linux would be compelling enough to defeat x86/Linux and x86/Windows?

Take a look at today's announcement from IBM, if you would. Aside from some commodity parts, the entire machine is provided by Big Blue. It's not at all beholden to Intel or Microsoft for anything.

As with all things, it will scale down eventually.

The only way to do that is to undercut on price because its pretty damn easy to have a "fast enough" desktop system, and people will default to what they already have laying around.

Incorrect. Undercutting is one way to do things, but offering a superior product is another. Chevy sells a lot of cars, but so do BMW and Porsche. The only reason I'm using the car analogy is on pricing and features, in this case, not performance, though that's gotten pretty comparable lately.

Ever look at IBM's desktop prices? Last time I checked Dell had them beat, and they would be all over x86/Linux the second Linux started looking like a desktop money maker.

That's buying parts from outside. There's no volume discount like doing your own manufacturing and providing your own processors.

Them and about a zillion other companies, big and small. Your expecting IBM to take on the entire x86-box industry, and defeat them on price, while keeping two processor lines up to date. I'm not seeing the profit.

I'm expecting IBM to start the revolution, with the help of AMD, Apple, and others who have a vested interest in Intel going down. Notice who's working together on HyperTransport, have we? How about the partners in PowerPC and the companies that are buying onto the platform (which IBM owns, incidentally, not Motorola)?

Here's a list for you:
Hypertransport - AMD, Apple, Cisco, Transmeta, Broadcom, Sun, nVidia, ATI, Agilent, VIA, PMC, TexasInstruments, NEC, IBM
PowerPC - Apple, AMD, Sony, Freescale, Nintendo

The PowerPC list isn't immediately as eyedrawing, nor is there an easy-to-find partner list like there is for Hypertransport. However, there is some serious clout behind both movements. Some might say worrying clout, if you're Intel or Microsoft.

Oh but maybe IBM has a clever plan to sell expensive PPC/Linux desktops. Heh.

Why's that so terrible an idea? Linux distributions are increasingly becoming commercial ventures for the companies that provide enterprise-grade support. It could even be argued that OS X is the forerunner of the movement, a graphical *nix that runs on PowerPC hardware and isn't vulnerable to all of the Windows headaches.

 

[windowsblowsass]

I never said there wouldn't be advantages to it, but you're still dead wrong if you think it will ever happen. there's still the physical space issue--apple will not put 2 CPUs on a logic board that small--there just isn't room, even with a small chip like that. another issue is that going back to a G3, even if you got better performance, is a marketing disaster.

IBM is working on making the 970s run cooler and faster. they're making headway, although they really screwed up on the latest batch. before long they will have a feasible chip for a laptop, and that is what apple will use.

you should know that people have always been clamoring for DP powerbooks. i have heard it over and over. there's a reason they're always page 2 type rumors.

no its not just becuse its g3 based doesnt men theyll call it a g3 a pentium m is a reworked p3 but is it called p3 no

 

[windowsblowsass]

"magic steve day."

greatest line in the history of the earth

 

[Palador]

How did this thread get so off topic?

Anywho, it is very obvious many people here have too much book knowledge and not enough hands on experience. The G5 is hot. Contrary to popular forum rumor about smaller processes and less watts... the new G5s will be even hotter. Think INSIDE the box on this one. What happens when I put more transistors, pathways, and whatnot into a smaller space at a higher speed? It gets hotter.

One thing that some people do not get is that heat is not something that should delay a new machine. If people are out there overclocking 3.4GHz PressHots running at over 100 watts... and if 3.4GHz desktop processors are already showing up in 8 pound notebooks... well... I think Apple and IBM could figure out SOMETHING. Granted it might not be as small and sexy as we all want it... but something could be done to get these things into a 17" Powerbook. And Im sure they could fit into a 3 foot tower.

Im surprised it wasnt posted here, but I think it was on cnet or slashdot that I read an article about how the 90nm process was getting insane amounts of pathway leakage for every manufacturer out there. Moore's law is about to stall out in the near future unless we move in a new radical direction.

 

[rog]

I bet they won't be. That will make me cry.

 

[Blackheart]

I bet they won't be. That will make me cry.

Geez, steve is making people cry now. Give us our magic steve day!

 

[awesomebase]

No way 3GHz is coming this summer...

I think the fact that there have been no intermediate updates in the January to March timeline almost guarantees that there will be no 3GHz models available for the G5 until January to March '05. It is not so much about the possibility of doing it, but rather the numbers. Looking at Apple's history, I haven't seen that big a speed bump (i.e. 50%) in their PowerMac lines. However going to 2.4GHz is more likely. What I think Apple has more problems with is being able to bring out speedier processors as a "regular" habit instead of making a big deal out of it at certain times of the year. They should be able to bring out the fastest processors they can for the PowerMacs and as they develop them faster and faster, use the current processors in single units for the iMac lines since by that time the numbers and yields will drive down the costs.
I don't think too many people care about how fast their machine is right now unless they are in very specific scientific arenas. What they want is good value. Apple should save the big events for major design changes and product introductions, not a 10% to 20% speed bump. Here is one suggestion... how about making those G5 machines about 33% smaller so that it approaches something close to what other machines take up in terms of space... now THAT would be cool at just about any speed.

 

[Frohickey]

Here is one suggestion... how about making those G5 machines about 33% smaller so that it approaches something close to what other machines take up in terms of space... now THAT would be cool at just about any speed.

To paraphrase Yoda...

Speed leads to heat... heat leads to airflow...airflow leads to more space.

If you shrink the G5, that would lead to less space, and a more restrictive airflow. That would then lead to faster fans, which lead to a louder machine.

So, do you want small and loud, or big and quiet.

 

[awesomebase]

To paraphrase Yoda...

Speed leads to heat... heat leads to airflow...airflow leads to more space.

If you shrink the G5, that would lead to less space, and a more restrictive airflow. That would then lead to faster fans, which lead to a louder machine.

So, do you want small and loud, or big and quiet.

Well, of course the laws of physics are true, but IBM is also going to a smaller transistor (90nm as opposed to 135nm) which will reduce not just the size of the core, but the power and heat being produced by it. Besides, if Apple continues this trend, yes, we will all have very fast machines, but we'll need a room just to hold the thing!
All I'm saying, is that I would be much more enthusiastic about the G5 being redesigned to be smaller than I would be about gaining another 400MHz. Having both would be even better! And besides... they BETTER figure out how to cool this thing quickly because PowerBook sales will only suffer the longer they go without incorporating a G5. They are doing well now, but, if we're still talking about getting a G5 into a PowerBook a year from now, I think sales at that point will be pretty slow... much like the G4 PowerMacs were in the several months before the G5 update came. Well, we can only hope that the physical designs will be good enough to hold the new G5s (hopefully in a smaller case) in both the PowerMacs and (hopefully as well) PowerBooks in about a year from now.

 

[jane doe]

????

Anyone know if this date is real?

 

[Nethawk]

Anyone know if this date is real?

It's not.
Like the other 456² rumors which came up the last 6 months.

 

[thatwendigo]

Well, of course the laws of physics are true, but IBM is also going to a smaller transistor (90nm as opposed to 135nm) which will reduce not just the size of the core, but the power and heat being produced by it. Besides, if Apple continues this trend, yes, we will all have very fast machines, but we'll need a room just to hold the thing!

In general you can have lower heat or higher clock. It generally doesn't happen that you get both, because the increased clock means more transistors in a smaller space, and thus, more heat output. This is why you can look at a first-generation 970 running at 2.0ghz and see it putting out 45-55w at peak, but the redesigned 970fx at 2.0ghz putting out a comparatively balmy 25-35w. Also, I know you're joking on the last part, but please... Serious workstations and server chassis on the PC side of things can be 10-bay monsters that make the G5 look puny.

All I'm saying, is that I would be much more enthusiastic about the G5 being redesigned to be smaller than I would be about gaining another 400MHz. Having both would be even better!

I can't possibly imagine why, unless you're one of the people who thinks that the G5 just needs, needs, needs to go into the PowerBook despite all the of problems with shoehorning the increased support fabric into the form factor. For a desktop processor, IBM and Apple are far better off shrinking the process and using the descreased heat to ramp the clock as high as it can possibly go, because they've already proven that the G5 processors in Apple's enclosure can be cooled quite nicely with a combined heat budget of nearly the same as a single P4 Prescott (i.e. around 100w total processor at peak). If you scale up to keep roughly the same heat but get 600, 800, or even 1000mhz more, then you're doing what really needs to be done in this case.

The 970 is a deeper processor than the MPC74xx series, and so it needs clock to ramp up.

And besides... they BETTER figure out how to cool this thing quickly because PowerBook sales will only suffer the longer they go without incorporating a G5.

Or, you know, they could put a processor intended for small form factors in there and completely bypass the issue of heat from processor, bus, and ASIC until the 980s are ready and, as some have rumored, much cooler and better optimized designs. Or, as we've been increasingly seeing, Apple could go with the VX that they're holding off to give Freescale a chance to come through on the e600/e700 cores.

I'm sorry, but unless the G5 is suddenly going to become massively cooler (like 10-15w at 2.0ghz), then we're better off with a low power processor like the e600 dual-core 2.0ghz. It would allow SMP optimizations, a 400mhz FSB as opposed to the very, very power hungry 800mhz-1ghz bus for the G5, and probably a cooler ASIC than the 970s require. To top it off, you get a power scaling optimized chip that draws 30w at peak to run two cores, not one.

 

[awesomebase]

In general you can have lower heat or higher clock. It generally doesn't happen that you get both, because the increased clock means more transistors in a smaller space, and thus, more heat output. This is why you can look at a first-generation 970 running at 2.0ghz and see it putting out 45-55w at peak, but the redesigned 970fx at 2.0ghz putting out a comparatively balmy 25-35w. Also, I know you're joking on the last part, but please... Serious workstations and server chassis on the PC side of things can be 10-bay monsters that make the G5 look puny.



I can't possibly imagine why, unless you're one of the people who thinks that the G5 just needs, needs, needs to go into the PowerBook despite all the of problems with shoehorning the increased support fabric into the form factor. For a desktop processor, IBM and Apple are far better off shrinking the process and using the descreased heat to ramp the clock as high as it can possibly go, because they've already proven that the G5 processors in Apple's enclosure can be cooled quite nicely with a combined heat budget of nearly the same as a single P4 Prescott (i.e. around 100w total processor at peak). If you scale up to keep roughly the same heat but get 600, 800, or even 1000mhz more, then you're doing what really needs to be done in this case.

The 970 is a deeper processor than the MPC74xx series, and so it needs clock to ramp up.



Or, you know, they could put a processor intended for small form factors in there and completely bypass the issue of heat from processor, bus, and ASIC until the 980s are ready and, as some have rumored, much cooler and better optimized designs. Or, as we've been increasingly seeing, Apple could go with the VX that they're holding off to give Freescale a chance to come through on the e600/e700 cores.

I'm sorry, but unless the G5 is suddenly going to become massively cooler (like 10-15w at 2.0ghz), then we're better off with a low power processor like the e600 dual-core 2.0ghz. It would allow SMP optimizations, a 400mhz FSB as opposed to the very, very power hungry 800mhz-1ghz bus for the G5, and probably a cooler ASIC than the 970s require. To top it off, you get a power scaling optimized chip that draws 30w at peak to run two cores, not one.

Thanks for the information! Yes, I do agree that getting lower power processors would be quite advantageous, especially where portables are concerned as they are becoming a larger part of the market as time goes on. And of course the G5 is not as big as those workstations with 10 drive bays (I certainly remember what those were like...)
Lets hope those come along sooner than later!

 

[numediaman]

Things seem to be awfully quiet here on the rumor front (G5 related, that is). Are we in a lull waiting until WWDC? Or later?

What kind of optical drive does everyone think will be in a new G5? And are there any other upgrades you expect?

(Even the French site hasn't been updated since the 27th of April.)