Clearing out G5s?

[sushi]

if they are going to bring out the G5 1.8, 2 and 2.2, I wonder how many people will buy them.

Where do you get this?!

SJ promised 3Ghz systems. He wouldn't have done that if he didn't believe that there was a good chance of making it happen. Granted, there may have been some setbacks. But to introduce systems only 200Mhz faster would be unbelieveable.

IMHO, SJ is not going to put his ego on the line if he isn't almost 100% sure that what he predicts is going to happen.

Sushi

 

[sushi]

Yeah, he/she is probably under a new guise... probably a "regular" poster by now...

Hey Sushi your a "regular" poster...

No offence, just having a bad week (and its only Monday)...

Where the $^&# is my PB...

Dang, I'm busted. I thought that I could change my ID and blend...

...guess not!

Reference your powerbook. We gave it to someone else who needed one more.

BTW, all of the above is BS.

aswitcher, hope your week gets better!

Sushi

 

[sushi]

Yeah, I am pretty stunned that so many of you give Apple absolutely no credit at all.

Amazing isn't it?

It is as if these folks think that Apple employees have just sat around and one nothing for the past year. In addition, this lack of action has permiated over to IBM and they are the same way. Unbelieveable!

Patiently awaiting the good news from Apple.

Sushi

 

[fpnc]

Okay, here is another possibility. Since it appears that the PPC970 processor design allows for bus multipliers of 2, 3, 4, and 6 maybe we can assume that the 970fx implements a 3X multiplier. This raises the possibility that Apple could do a simple processor replacement (drop in) to produce the following 970fx-based products:

2.4 GHz with 3 x 800 MHz FSB.
2.7 GHz with 3 x 900 MHz FSB.
3.0 GHz with 3 x 1000 MHz FSB.

Now compare this with the current 970 lineup.

1.6 GHz with 2 x 800 MHz FSB.
1.8 GHz with 2 x 900 MHz FSB.
2.0 GHz with 2 x 1000 MHz FSB.

The symmetry seems frighteningly logical.

I don't think 3 GHz is likely to be available at WWDC, but maybe (hope, hope) a 2.7 GHz 970fx is a possibility. After all, that's only slightly above the rumored yields at 2.6 GHz.

So, here are my very optimistic predictions for availability at WWDC:

2.4 GHz single processor (replaces current 1.6 GHz).
2.4 GHz dual processor (replaces current dual 1.8 GHz).
2.7 GHz dual processor (replaces current dual 2.0 GHz).

Then, perhaps towards the end of summer we'll see the 3.0 GHz and the configurations would be changed to:

2.4 GHz single processor.
2.7 GHz dual processor.
3.0 GHz dual processor.

Given this, I think Steve Jobs' pronouncements about the availability of 3 GHz system before the end of summer 2004 makes perfect sense. He could also point out that the least expensive 2.4 GHz single-processor system has achieved a 50% clock increase in only 12 months (partially fulfilling last year's WWDC promise).

Frankly, I don't think many people would complain about a dual 2.7 GHz system if it could ship during this July.

Update (edit): my bad, I was playing loose with the terminology again. Technically it would probably be more correct to say bus multiplier X FSB rather than bus multiplier X system bus. So, I've changed to the former.

 

[aafuss1]

Folks, as much as we love the Cube and wish new G5 cubes were here, this topic is not about the cubes AND it's highly unlikely anything cube related will ever happen again.

Agreed, but if say the Apple wanted say a G5 mini (unlikely).

 

[Phinius]

Multiplier

Okay, here is another possibility. Since it appears that the PPC970 processor design allows for bus multipliers of 2, 3, 4, and 6 maybe we can assume that the 970fx implements a 3X multiplier. This raises the possibility that Apple could do a simple processor replacement (drop in) to produce the following 970fx-based products:

2.4 GHz with 3 x 800 MHz system bus.
2.7 GHz with 3 x 900 MHz system bus.
3.0 GHz with 3 x 1000 MHz system bus.

Now compare this with the current 970 lineup.

1.6 GHz with 2 x 800 MHz system bus.
1.8 GHz with 2 x 900 MHz system bus.
2.0 GHz with 2 x 1000 MHz system bus.

Apple uses a 2X multipler for the G5 bus. That means the cpu is running at 2X the speed of the memory bus. So if it's 3X, then the bus is running at 1/3 the speed of the cpu.

The 2GHz 970 uses a bus that runs at 500MHz, but it is double pumped to reach 1,000MHz. But due to command and address overhead it's throughput is effectively the equivilant of a 800MHz bus.

When the 970 design reaches 3GHz, a 2X bus multiplier would bring the bus frequency to 1,500MHz. But, again, due to command and address overhead, that comes with it's design, it's effective speed would be perhaps as much as 1,300MHz. To match up with that bus speed there is 667MHz DDR2 memory that has already started to be produced and with dual-channel memory that would be 1,334MHz. Because this memory is going to be produced in much lower quantities at first it will be significantly more expense, so a 1.5GHz G5 bus is probably more a matter of whether Apple chooses to use 667MHz DDR2 memory this early in it's production.

 

[fpnc]

...When the 970 design reaches 3GHz, a 2X bus multiplier would bring the bus frequency to 1,500MHz. But, again, due to command and address overhead, that comes with it's design, it's effective speed would be perhaps as much as 1,300MHz. To match up with that bus speed there is 667MHz DDR2 memory that has already started to be produced and with dual-channel memory that would be 1,334MHz. Because this memory is going to be produced in much lower quantities at first it will be significantly more expense, so a 1.5GHz G5 bus is probably more a matter of whether Apple chooses to use 667MHz DDR2 memory this early in it's production.

That's fine, but won't Apple's custom system controller also need to work at those speeds? So, it's more than just the CPU and memory, you also need a system controller that can run on that faster bus.

...that comes with it's [sic] design, it's [sic] effective speed would be perhaps as much as 1,300MHz...Apple chooses to use 667MHz DDR2 memory this early in it's [sic] production.

I guess if we can complain about math then we can complain about grammar. That should be "its" not "it's." When you write "it's" you are saying "it is" or "it has" while the possessive form of it is simply "its" (no apostrophe). Then again, that's in American English and maybe it is different in "real" English (i.e. British). Or, perhaps English is your second language in which case I congratulate you greatly for being so literate in another language. I'll admit that my grammar isn't always the best (I make writing errors too), but it was a little "jarring" to see this mistake so many times in a row.

 

[Palador]

I HIGHLY doubt Apple will stay with the 2x multiplier unless slower than expected updates are released topping at 2.4GHz or so. It just wouldnt make sense for them because even now the RAM cannot keep up with the bus... let alone the rest of the system which is even slower.

I think the fastest we'll see at WWDC is a 2.4. I could be wrong, but all evidence points to it. Except for Job's statements a year ago that is...

Its ironic that the laptops are handicapped by their bus, but the PowerMacs have more that they can handle!

 

[thatwendigo]

On to partial responses, since this thread has nearly gotten away from me. First off, I'd like to thanks those who understood my need to be away for most of the weekend. I'm the brother of a future jarhead, and I wanted to see him before he went off to Parris Island and then to his post when he graduated, since he's elected to go active duty intelligence.

I support his want to do something with his life, and though I think this is the most recklessly dangerous thing he's ever done, I was there to see him off this morning and spent as many hours of the past two weeks as I could with him, no matter what I might think of current policy decisions, there will always be a need for grunts to do the work on the ground, and he's willing to put his neck on the line.

An iMac (headless) offers a cheap G5, much cheaper than any curent tower. i dont even care what it looks like, the fact is selling a base of a iMac could be cut down to 800+$, which would bring alot more costomers into Apples ballpark. Considering if they dont have the money to spend on a beautiful LCD display, they can use their current CRT monitors and be haoppy. IT would distinctly separate the low end computers, mid-level consumers, and professional buyers with all 3 levels. Offering all types of pricing options, along with bringing their monitor proces down due to the separate iMac sales instead of the packaged current deals.

There is absolutely no reason to believe that a headless iMac would be at all cheaper than the current generation if Apple moved to the 970 instead of the G4. On the contrary, there's a whole host of new technologies that weren't in the case before that would have to be integrated, new motherboards to be designed and fabbed, and other economic factors that are typically left out of this kind of comment. Be wary of just jumping onto any bandwagon in the mac camp - my own included - and always do your research about the components involved and the other issues.

I know a lot, and try to find out what I don't know, but I'm fallible and so is everyone else here.

Oppossed to the cuurent huge price gap between a decently loaded AMD computer and Apples low end computer.

That "decently loaded" computer often needs $300 or more of software to equal the power, stability, security, and options that OS X possesses right out of the box. Believe me, I've done a huge amount of price comparison and competition searching, and most of the "cheap" machines are that way for a reason - integrated graphics and shared RAM, lower-grade RAM, small HDs, and other cut corners.

Show me one of those machines you're talking about, and give me a link to it. I'll find out if it's one of the boxes you get from a company that grazes the bottom line, uses their consumer electronics to prop up selling at a loss or no profit, or one of the other tricks that makes Apple look "bad" in comparison.

You and I understand the specific details of engineering it and how difficult it is. Mom and pop just want it because it looks cool, and it beats the hell out of the comparable pentium M, and if Apple never pulls through on this front, then they will eventually loose.

Then Apple is doomed to death, if your statement is true. No one company can fight the whole computer industry on its own, and that includes Microsoft.

P.S. .........Irish carbombs frikkin rock!!!!!!!

I'm more of a Crown and Coke, Margarita, and Antichrist drinker, myself. Never had an Antichrist? Oh, well you just take a double shot of 151 and add a finger of tabasco, then sprinkle cayenne pepper on top.

Trust me... You drink your first one and it becomes obvious where the name came from.

According to Apple's data the 800MHz G4 flat-panel iMac consumes 130 watts max and produces a maximum thermal output of 445 BTU/hr. This was the Jan. 2002 model with the 15" LCD.

Now let's compare that to the single 970fx-based Xserve G5 running at 2.0 GHz. According to Apple's data this configuration requires 170 watts max and produces a maximum thermal output of 580 BTU/hr. This is for what they call a minimum configuration -- Single 2.0 GHz G5 Processor, single 80 GB Apple Drive Module, 512 MB ECC RAM (2x256 MB DIMMs), no PCI cards, no video. Adding a PCI video card apparently increases the power consumption by 10 watts (according to Apple).

You're leaving out something really, really important about the xServe. Those two holes on the front are channels for the basically constantly-on fans that are blowing air through the system without any regard for noise or power consumption. It's a continuously blown cooling system, and one that's famous for being far louder than any previous Apple computer, because the xServes are about reliability and stability, not quiet operation.

By contrast, the iMac is all about being unobtrusive in footprint and near-silent in operation. The same solution wouldn't work for it.

I think in the more traditional sense it is really what is called the system bus and AMD calls the Athlon 64's HyperTransport link the system bus. And certainly, the 1GHz bus that runs between the G5 and the Power Mac's system controller is both the system bus AND the FSB.

...

Interestingly, this article then goes on to discuss the various differences in the methods that Intel, AMD, and IBM (G5) are pursuing to break what they termed the "Frontside Bus Bottleneck." Hint, although there are similarities it's apparent that IBM and AMD are not pursuing the same track.

Here is the link:

http://www.technewsworld.com/story/31594.html

Well, I'd like to say in my defense that I think I might have been unclear in what I was attempting to say, so I'll try to clarify it a bit more. What I refer to as the "FSB" in the G5 tower is the same as what your definition from geek.com says that one is, because the FSB on the PowerMac is a carrier for memory controller and I/O systems at once, which the pure AMD HyperTransport bus is not. This is one reason that I see it being likely that the next major revision of Apple's chips will have on-die memory controllers, because it is one of the few places that it loses on bandwidth at the moment.

To quote from the article:

Because Apple's bus is bidirectional, the G5 processor -- developed by IBM -- can send and receive information continuously, unlike most other processor interfaces that can carry data in only one direction at a time.[*correction] The bidirectional technique -- which also will be employed by AMD's 64-bit processors -- lets data move in opposite directions simultaneously with no other demands on the data stream and no wait time while the processor and system controller compete for use of the bus.

In Apple's G5 configuration, the data streams integrate clock signals along with the data, allowing the frontside bus to work at speeds of up to 1 GHz for a total of 8 Gbps of aggregate bandwidth.

Also, in the G5's dual-processor configuration, each processor is connected to its own dedicated 64-bit bidirectional bus, allowing each CPU to send and receive data without competing with the other processor for bandwidth. According to Apple, this technique allows larger chunks of data to be processed per clock cycle.​

Well, since Arstech specifically asked about the available bus multipliers when they interviewed IBM's Peter Sandon (chief architect on the PPC970) I think that probably indicates that at least some people think that bus multipliers other than 2X are potentially useful. When I went back to this article I found that Peter Sandon said that the "processor design" supports bus multipliers of 2, 3, 4, and 6. However, I think it was unclear whether the PPC970 actually implemented those multipliers (i.e. the design could support them, but it may not be implemented in the 970).

There really is no benefit to running the bus at a higher multiplier, that I'm aware of, because it means that you're cramming more traffic for a faster processor through a more narrow pipe. I read the link to the Ars article, and when you compare the commentary about multipliers to the parts about AltiVec and other aspects of the design, it looks pretty minor.

Sure, someone might be interested in a higher-multiplier 970, but I doubt that it's going to be Apple, especially not since the bus on AMD systems is now reaching to 2.4ghz. For references, see our good friend AidenShaw's post somewhere above mine.

--

Once again, I'm slogging through. Phinius, you're next.

 

[clr900]

I would be disappointed if the lowest model does not have dual processors. I probably am going to buy the middle model regardless of what they release but if the bottom model is fast enough and has dual processors, this would probably be fast enough for me and many other people. However, if Apple only makes it a single processor, they are pretty much forcing people to buy the next up only because of the much better value of two processors for not that much more money. I also think that they need to maintain dual processors throughout the entire power mac line to differentiate between the consumer (headless?) g5 imac which would have only one if they decide to release this.

 

[thatwendigo]

Motorola's own data sheet on the 7447A shows that at 1.42GHz it typically uses 21 watts and has a maximum usage of 30 watts. So, if the 7447A is theoretically pumped up 1/3 more in frequency to 2GHz, then it should consume a maximum of about 40 watts, which is about 18% less than the maximum you list for the 2GHz 970FX.

This is only without the DFS feature enabled, which cuts the heat down to 12.5 watts, which is roughly the same as the 11-12 watts that i've been claiming for some time. While I can't find my original source for that heat figure, I did plug the necessary numbers into the product release PDF for the 7447A and you end up with a power slewing feature that averages out power consumption to 12.5 watts typical and around 25 max, as opposed to the 24.5 watts typical and 49 watts maximum of the 970FX 2.0ghz. However, I doubt that we'll see a 7447A 2.0ghz, but rather an e600 at either 16-18 watts for a single or 25-28 watts for a dual-core chip, and both of those with a 400mhz bus, on-die DDR controller, and 128-bit dual-precision SIMD.

The main justifications for making the 970FX was probably for a speed increase and to reduce costs the die size was greatly reduced by leaving the L2 cache the same size as the 970. Could it be that Apple intends to eventually use the 970FX as the lowend 970 in the consumer Macs, in much the same way as Intel reduces the L2 cache for the Celeron line. That way Apple can still justify the price differences in the consumer and pro models and yet still have them both running at the same frequencies.

Actually, I think you might not be all that far off with this speculation, though I think that the product line will have a bit more differentiation than that. I fully expect that the 970FX will be a consumer chip in the future, or that some iteration (970GX, maybe, to follow in the 750's naming pattern?) will go on to be the replacement for the lowend desktop chip. However, I also fully expect to see Apple and IBM exploit the Power5 architecture as soon as Cupertino can get the massively more powerful cores into their products. I've said for some time now that the 970 felt like a rush job, and the article quoted from Ars in my above post admits such a bit obliquely, through what Sarndon said about the AltiVec implementation.

So, what is the maximum watts do you believe that a 3GHz 970FX will use? I'd venture to guess it will be close to the same as the 2GHz 970.

I'm thinking 50-60 watts, actually, and would be surprised by anything more than that.

Two 7447A processors, both running at 1.42GHz, could use a maximum of 60 watts (30 watts X2 = 60). Which is more than the maximum power use of a 2GHz 970FX. Motorola has already stated that a dual-core G4 (probably made on a 90-nm process) will typically use 25 watts at 1.5GHz.

Default 7447A processors can reach 30 watts, but the power-slewing that's on-chip in the A-revision chip allows it to be functionally 12.5 watts typical, which pus the dual-chip typical at 25 watts and moves the max to 50 watts for two chips. This contrasts to 24.5 watts typical and 59 watts maximum for the 2.0ghz 970FX, which has a far higher overall BTU output based on the Apple documents I cited some way back.

PC makers have been putting 3GHz desktop Pentium 4 chips in notebook computers for some time now and I don't imagine that there are a lot of noisy high-speed blowers cooling them. Putting a <2GHz 970FX processor in a small form factor desktop box would not be next to impossible judging from that.

PC makers have been getting an hour of battery life and needing constantly-on fans to cool Pentium 4 desktop chips in 9-12 pound "laptops," actually. If you want a windmill on your desktop, then you could advocate the idea of cramming a 50 watt max chip into a laptop (hey, it's half the P4, right?), but I'll take a miss.

 

[fpnc]

...I'm the brother of a future jarhead, and I wanted to see him before he went off to Parris Island and then to his post when he graduated, since he's elected to go active duty intelligence.

I wish the best for your brother. Both my brother and I are military vets. I can't say that my time in the Air Force was always great, but I've never had any serious regrets about the time I spent in the service.

...You're leaving out something really, really important about the xServe. Those two holes on the front are channels for the basically constantly-on fans that are blowing air through the system without any regard for noise or power consumption. It's a continuously blown cooling system, and one that's famous for being far louder than any previous Apple computer, because the xServes are about reliability and stability, not quiet operation.

By contrast, the iMac is all about being unobtrusive in footprint and near-silent in operation. The same solution wouldn't work for it.

I'm having a problem understanding your logic here. If you look at the BTU/hr values you can see that the 800MHz iMac produces almost as much heat as the minimum configuration, single-processor, 2GHz G5 xServe. That's my point, a previous generation iMac produced almost as much heat as a G5 xServe. Any yet, that iMac didn't have a "continuously blown cooling system...famous for being far louder than any previous Apple computer."

In fact, your arguments help prove my point. The xServe does have multiple, high-speed fans and it's very noisy but that's required because of its extra beefy power supply, support for PCI slots, and its three drive enclosures (not to mention that it can be equipped with dual processors). If you strip out all of that xServe stuff that isn't really required for an iMac design you suddenly have a G5-based system that requires even less power and generates even less heat.

So, the BTU/hr values were previously:

800MHz iMac -- 445 BTU/hr.
2GHz G5 xServe -- 580 BUT/hr.

Thus the xServe produces (580-445/445) = 30% more BTU/hr.

That's fairly significant but probably manageable when you consider that you could reduce that difference even further by throwing out many of those xServe extras that wouldn't be required in the iMac.

Then there is the fact that the minimum configuration G5 xServe only consumes 40 watts more (170w versus 130w) than the 800MHz iMac. Start throwing out some of the xServe's parts as I mentioned above and I think this difference could be reduced significantly. But then, there are few parts that would have to be added to create an iMac design (one example: graphics support, kind of important I guess). It may be worth noting, however, that the current 20" LCD iMac comes with a 190w power supply.

And then there is always the possibility of reducing the clock speed of the iMac's G5 (i.e. you could ship lower than 2GHz).

Also, let's look at how much heat a maximum configuration of the G5 xServe produces (Dual 2.0 GHz G5 Processors, three 250 GB Apple Drive Modules, 8 GB ECC RAM (8x1 GB DIMMs), no PCI cards)

Dual 2GHz xServe ("max" configuration) -- 990 BTU/hr.

That's more than twice the value of the 800MHz iMac and now you begin to see why the xServe needs all those fans. And this value doesn't even include any PCI cards. Put a couple of power hungry PCI cards in the xServe and you'll need those fans even more.

Finally, I think we should stop trying to compare the FSB speed potential on the G5 Power Mac with the Athlon 64. Saying that the Power Mac should be able to run its FSB at speeds similar to the Athlon 64 is a big assumption (IMO). A better comparison, I think, would be to the Pentium 4 architecture which is currently running at 800MHz (with, I guess, 1066MHz coming in the next chip-set generation).

If you want, you can now begin the debate on why the Pentium 4 can't run its FSB as fast as the Athlon 64. I mean, can't those Intel engineers do anything right?

 

[Phinius]

Your partially right

This is only without the DFS feature enabled, which cuts the heat down to 12.5 watts, which is roughly the same as the 11-12 watts that i've been claiming for some time. While I can't find my original source for that heat figure, I did plug the necessary numbers into the product release PDF for the 7447A and you end up with a power slewing feature that averages out power consumption to 12.5 watts typical and around 25 max, as opposed to the 24.5 watts typical and 49 watts maximum of the 970FX 2.0ghz.

You are right that the power saving feature of the 7447A will reduce the typical watt usage below what Motorola mentions. However, this feature works by changing the frequency of the processor dynamically. So, if the processor is pushed to its full potential then the maximum power use that Motorola lists should be correct at 30 watts for a 1.42GHz 7447A. Afterall, the processor would not have a change in frequency if it is constantly pushed to maximum performance.

Actually, I think you might not be all that far off with this speculation, though I think that the product line will have a bit more differentiation than that. I fully expect that the 970FX will be a consumer chip in the future, or that some iteration (970GX, maybe, to follow in the 750's naming pattern?) will go on to be the replacement for the lowend desktop chip. However, I also fully expect to see Apple and IBM exploit the Power5 architecture as soon as Cupertino can get the massively more powerful cores into their products. I've said for some time now that the 970 felt like a rush job, and the article quoted from Ars in my above post admits such a bit obliquely, through what Sarndon said about the AltiVec implementation.

The Power5 derived 970 processor would have enough performance advantage over the 970FX that Apple could easily move the 970FX to the consumer line and keep the bigger die sized/more expensive, higher power use chip for the pro line. IBM claims the Power5 is getting an average of 40% performance improvement by adding dual-threading. I would assume that a 970 implementation of that would also have a significant performance boost. In which case there would be enough G5 processor performance differentiation between the pro and consumer line, even at the same frequencies. Plus, it is likely that the Power5 derived G5 would get a larger L2 cache than the 970FX, in which case there could be a 10% boost in speed just from that.

Default 7447A processors can reach 30 watts, but the power-slewing that's on-chip in the A-revision chip allows it to be functionally 12.5 watts typical, which pus the dual-chip typical at 25 watts and moves the max to 50 watts for two chips. This contrasts to 24.5 watts typical and 59 watts maximum for the 2.0ghz 970FX, which has a far higher overall BTU output based on the Apple documents I cited some way back.

Again, your right about the typical watt usage going down with the chips ability to change frequency dynamically. But, if the processor is going full bore the processor will be running at maximum frequency and therefore the maximum watt usage that Motorola lists would still be correct. Or are you stating that due to the implentation of this power saving feature the processor could not go at listed top frequency for an extended period of time?

 

[dekator]

Whatever the actual machines will be...

their codenames are: 8,1; 7,3 (current G5 is 7,2) and there will be a Neo2.
Take a look here (10.3.4): /System/Library/Extensions/AppleMacRISC4PE.kext/
Contents/info.plist

7,3 is probably a PM G5 revision of some sort, 8,1 a new iMac ?
Neo2 might be what, the Xserve ?

Compare http://www.macbidouille.com/niouzcontenu.php?date=2004-06-01#8674 (in French) and
http://macnytt.se/news/index.lasso?news_id=7503&part=shownews.inc&s= (in Swedish)

 

[AidenShaw]

There's a market....

PC makers have been getting an hour of battery life and needing constantly-on fans to cool Pentium 4 desktop chips in 9-12 pound "laptops," actually.

If you want a windmill on your desktop, then you could advocate the idea of cramming a 50 watt max chip into a laptop (hey, it's half the P4, right?), but I'll take a miss.

PC makers have also been putting Pentium M chips into 2 1/2 lb systems with 6.5 hrs of battery life (http://www.toshibadirect.com/td/b2c/cmod.to?seg=ENT&coid=-26371&sel=0&rcid=-26367&ccid=1291021).

It's all about choice. Those 12 pound monsters are selling well, because some people like them.

I'd wager that the big laptops are more likely competing with the iMac than with PowerBooks. While the iMac makes a statement on your desk, many people prefer to be able to make the computer disappear when not in use. Therefore, they'll buy a big laptop that's about as powerful as a tower - yet can be closed and put in a drawer or on a shelf when not in use.

 

[aswitcher]

PC makers have also been putting Pentium M chips into 2 1/2 lb systems with 6.5 hrs of battery life (http://www.toshibadirect.com/td/b2c/cmod.to?seg=ENT&coid=-26371&sel=0&rcid=-26367&ccid=1291021).

It's all about choice. Those 12 pound monsters are selling well, because some people like them.

I'd wager that the big laptops are more likely competing with the iMac than with PowerBooks. While the iMac makes a statement on your desk, many people prefer to be able to make the computer disappear when not in use. Therefore, they'll buy a big laptop that's about as powerful as a tower - yet can be closed and put in a drawer or on a shelf when not in use.

There are some nice PC laptops appearing over the last year. Plenty of competition for the 12s and 15s but I still think Apple are well ahead on their sleek 17...

 

[Zaty]

No intermediate model

So, here are my very optimistic predictions for availability at WWDC:

2.4 GHz single processor (replaces current 1.6 GHz).
2.4 GHz dual processor (replaces current dual 1.8 GHz).
2.7 GHz dual processor (replaces current dual 2.0 GHz).

Then, perhaps towards the end of summer we'll see the 3.0 GHz and the configurations would be changed to:

2.4 GHz single processor.
2.7 GHz dual processor.
3.0 GHz dual processor.

I don't think Apple will do what you suggest because if they introduce a dual 2.4 as an intermediate offering, it will either hurt sales of the single 2.4 GHz (if there is only a small difference in price) or if the former is too expensive in comparison with the dual 2.7, no one will buy a dual 2.4, so if they belive they can deliver a 3 GHz in quantities by the end of summer, they will announce it at WWDC as well, available in September (or late August).

 

[sushi]

I'd wager that the big laptops are more likely competing with the iMac than with PowerBooks. While the iMac makes a statement on your desk, many people prefer to be able to make the computer disappear when not in use. Therefore, they'll buy a big laptop that's about as powerful as a tower - yet can be closed and put in a drawer or on a shelf when not in use.

Good point.

I would suspect that in many cases it is a space issue. Many folks I know use their computer on their kitchen or Den table. Being able to hide it, or move it conveniently is a nice option to have.

Also, personally while I love new hardware and software, most folks simply do not need it. For the majority who do e-mail, web browsing, word processing, some spreadsheet work, and some other minor apps, a G3/G4 is way plenty power. Same on the PC side. You can do all of these with a PIII/P4 Celeron, or low end Athlon based system. That is why you see laptops becoming more and more popular. No longer is their a big power hit when using a laptop vice a tower/desktop system like there used to be.

Sushi

 

[AidenShaw]

since this is an *opinion* board...

but I still think Apple are well ahead on their sleek 17...

For me, I put the 17" in the same bucket as the 12 pounders - too big and awkward for my needs.

I like the 14" screens in the 4 to 5 pound range - I don't want a behemoth like the 17", and the 2 to 3 pounders sacrifice too much for size.

 

[manu chao]

their codenames are: 8,1; 7,3 (current G5 is 7,2) and there will be a Neo2.
Take a look here (10.3.4): /System/Library/Extensions/AppleMacRISC4PE.kext/
Contents/info.plist

7,3 is probably a PM G5 revision of some sort, 8,1 a new iMac ?
Neo2 might be what, the Xserve ?

Compare http://www.macbidouille.com/niouzcontenu.php?date=2004-06-01#8674 (in French) and
http://macnytt.se/news/index.lasso?news_id=7503&part=shownews.inc&s= (in Swedish)

The XServe is "Rack Mac 1.x" for the G4 modell and "Rack Mac 3.1" for the G5 modell. Since 7.3 was already in 10.3.3, I guess that would have been the 970FX update that we have never got. For sure we could still get it.

8.1 could be a 975 based Powermac.
Or 7.3 could as well be used for a 975 update, and 8.1 is a G5 iMac or a new headless Mac. NEO could refer to a new processor, i.e. the 975.

 

[mrsebastian]

just gimmie my [bleepin] updated g5 already apple!

 

[Mac Dummy]

I don't care anymore what they release, as long as its $100 across the line cheaper and with better graphics cards across the line! Come on Apple I want to buy my Mac gaming machine, update!

Yeah if Apple does keep the dual 1.8, I wish they would install an ATI card in it instead of the Nvidia, it might give a little better graphics performance (in terms of framerates). I just got done reading the May issue of Macworld, they did a test on the dual 1.8 and at least in the Unreal Tournament 2004 gaming test the dual 1.8 tied the 1.6 ( both scored around 40.7)in that test with the dual 2.0 miles ahead, so to speak. It scored like 69 something. I wouldn't mind buying the middleweight machine if the graphics card could keep up with the high end model. Because in every other test the dual 1.8 came in just a few points behind the dual 2.0. Like on the Photoshop test the dual 2.0 scored 204, and the dual 1.8 scored 199, not much performance difference for the difference in price.

 

[Mac Dummy]

That's not a bad idea. At least they were quiet!

How many of you think the new G5s will be liquid-cooled?

I don't think we'll see liquid cooling anytime soon, but that would be cool
It would be a first for a mainstream computer manufacturer. The only other people building liquid cooled computer are gaming PC manufacturers AlienWare and maybe Falcon Northwest.

 

[thatwendigo]

I wish the best for your brother. Both my brother and I are military vets. I can't say that my time in the Air Force was always great, but I've never had any serious regrets about the time I spent in the service.

Yeah, thanks.

Let's just say that I'm going to be pushing my parents to buy him some Interceptor 3, since I don't trust the government to be issuing it and I'd like to have him back alive at the end of his tour. Marines are sent into dangerous places a lot, and I'm fully expecting he'll be with them.

I'm having a problem understanding your logic here. If you look at the BTU/hr values you can see that the 800MHz iMac produces almost as much heat as the minimum configuration, single-processor, 2GHz G5 xServe. That's my point, a previous generation iMac produced almost as much heat as a G5 xServe. Any yet, that iMac didn't have a "continuously blown cooling system...famous for being far louder than any previous Apple computer."

The iMac has a graphics card, which the xServe does not. If you look at the models that I listed and you're referencing, one of the biggest differences in heat between the xServe and the PowerMac G5 is the inclusion of a modern GPU. Seriously... Look at the difference in the heat of single processor PowerMacs and the dual-processor xServe:

PowerMac 1.6ghz single 970 - Has the same drive (80GB SATA), lower clockspeed, lower bus speed (800mhz vs. 1ghz), less RAM, SuperDrive instead of CD-ROM or Combo (if you believe a drive adds 180W maximum heat...), and only one fewer PCI slot. The difference is 40W less typical usage, but 180W greater maximum.

Do graphics cards like the GeForce FX 5200 really consume 180W more? Those figures are for systems without anything but stock parts, which means no cards drawing from the slots.

In fact, your arguments help prove my point. The xServe does have multiple, high-speed fans and it's very noisy but that's required because of its extra beefy power supply, support for PCI slots, and its three drive enclosures (not to mention that it can be equipped with dual processors). If you strip out all of that xServe stuff that isn't really required for an iMac design you suddenly have a G5-based system that requires even less power and generates even less heat.

The single G5 1.6 tower has a power draw of 420 watts at peak, and that's without dual processors or anything fancy added on. To quote the footnotes of the page about the PowerMacs, "Numbers assume no power draw from external peripherals, including monitors with an Apple Display Connector, USB, or FireWire devices." That's a pretty beefy number for something with no external devices or power for the monitor, especially since it's not even top of the line. As you can see from the idle power of the dualies, the G5 seems to sleep at a draw of about 20 watts per processor, as compared the G4's sleep power of 4-6 watts in the most current versions.

It's really not as cool as some people want you to believe.

So, the BTU/hr values were previously:

800MHz iMac -- 445 BTU/hr.
2GHz G5 xServe -- 580 BUT/hr.

That's fairly significant but probably manageable when you consider that you could reduce that difference even further by throwing out many of those xServe extras that wouldn't be required in the iMac.

About the only "extra" that shows up in the Tech Specs page is the Gigabit ethernet and the extra drivebay.

It may be worth noting, however, that the current 20" LCD iMac comes with a 190w power supply.

Current G5s come with 600 watt power supplies.

Dual 2GHz xServe ("max" configuration) -- 990 BTU/hr.

Single 1.6ghz G5 PowerMac - 1420 BTU/h maximum
Single 1.8ghz G5 PowerMac - 1466 BTU/h maximum
Dual 1.8ghz G5 PowerMAc - 2060BTU/h maximum

It's bigger than you think.

Finally, I think we should stop trying to compare the FSB speed potential on the G5 Power Mac with the Athlon 64. Saying that the Power Mac should be able to run its FSB at speeds similar to the Athlon 64 is a big assumption (IMO). A better comparison, I think, would be to the Pentium 4 architecture which is currently running at 800MHz (with, I guess, 1066MHz coming in the next chip-set generation).

I'm not at all sure why you think the Pentium 4 is a better comparison, since it uses a more tradiitional bus architecure, when Apples is far more like the one used by AMD. The Athlon 64 uses a double-pumped bidirectional split bus, just as Apple does, but their doesn't have to carry the memory subsystem as much because of the on-die controller. The Pentium 4 still uses a Northbridge, so it's remotely like the G5, but not very much at all.

The Opteron is about to scale to 2.4ghz system bus, and the Athlon 64 to 1.6 ghz, both of them on the same kind of system Apple uses, with that one niggling difference. It's a better comparison.

If you want, you can now begin the debate on why the Pentium 4 can't run its FSB as fast as the Athlon 64. I mean, can't those Intel engineers do anything right?

Oh, I don't know. I'd say that the Pentium M was a pretty good bit of engineering, and that it's what I've been thinking Intel would have to do for a while now. It's even more RISC-like than anything of theirs but the Itanium was.

You are right that the power saving feature of the 7447A will reduce the typical watt usage below what Motorola mentions. However, this feature works by changing the frequency of the processor dynamically. So, if the processor is pushed to its full potential then the maximum power use that Motorola lists should be correct at 30 watts for a 1.42GHz 7447A. Afterall, the processor would not have a change in frequency if it is constantly pushed to maximum performance.

It still keeps the wattage down, since you're probably far better off with a desktop if you're doing anything that pushes the processor that hard. It's a good way to have no battery life, no matter what's under the hood (and that includes the Penitum M). Basically, I think that it's more than a little ridiculous to be using a portable machine for anything that processor intensive, especially since it's vastly cheaper to just invest in a powerful desktop and a machine to do your mobile tasks on. I'm not a professional film editor or graphics artist, though, so I guess I have no appreciation of how impoprtant it is to render that project while you're on the plane with eight spare batteries so that your processor doesn't croak.

The Power5 derived 970 processor would have enough performance advantage over the 970FX that Apple could easily move the 970FX to the consumer line and keep the bigger die sized/more expensive, higher power use chip for the pro line. IBM claims the Power5 is getting an average of 40% performance improvement by adding dual-threading. I would assume that a 970 implementation of that would also have a significant performance boost. In which case there would be enough G5 processor performance differentiation between the pro and consumer line, even at the same frequencies. Plus, it is likely that the Power5 derived G5 would get a larger L2 cache than the 970FX, in which case there could be a 10% boost in speed just from that.

Well, that's basically what I see happening within a year, actually. The IBM/Apple partnership will produce a P5D chip that will become Apple's professional desktop chip and a low to midrange server chip for Big Blue. Features I expect to see cross the gap are greater cache and cache-coherency, SMT, on-die memory control, higher I/O speeds, and possibly even the multi-core design.

I'd really like to see Apple and IBM beat the Intel world to the punch on consumer and professional mutli-core SMT systems.

PC makers have also been putting Pentium M chips into 2 1/2 lb systems with 6.5 hrs of battery life (http://www.toshibadirect.com/td/b2c/cmod.to?seg=ENT&coid=-26371&sel=0&rcid=-26367&ccid=1291021).

It's got even less powerful graphics than the older powerbooks. Hell, it's even less powerful than my iBook, has a 1.0ghz processor, uses PC2100 RAM, a 12.1" display that only reaches 1024x768, a 4200RPM HD, 802.11 b instead of 802.11g, no optical drive, no FireWire, and no DVI or S-Video. As I've pointed out numerous times in the past... It's not exactly impressive to cut corners and charge prices that roughly equal or exceed Apple's ($1,999 and $2,099 base prices)

It's all about choice. Those 12 pound monsters are selling well, because some people like them.

Some people bought Pintos, too.

I'd wager that the big laptops are more likely competing with the iMac than with PowerBooks. While the iMac makes a statement on your desk, many people prefer to be able to make the computer disappear when not in use. Therefore, they'll buy a big laptop that's about as powerful as a tower - yet can be closed and put in a drawer or on a shelf when not in use.

Actually, I could see this a lot more easily than I could calling them some kind of actual portable. They're certainly not intended for long flights where you need to get work done, with that kind of weight and ridiculous battery life. I'd take a PowerBook (for preference) or a Centrino (if I had to use a PC) over a "desktop replacement" any day.

--

Last, but not least, the mac is not, will not be, and probably shouldn't even try to be, a gaming machine. It's like buying a Porsche to haul lumber. Sure, it's powerful, but it's a different kind of power, and it's not exactly built for the task.

Go buy a pickup if that's what you want.

 

[AidenShaw]

It's not exactly impressive to cut corners and charge prices that roughly equal or exceed Apple's

Small is often more expensive than big (witness the iPod vs. some other audio players).

You're comparing the Toshiba to an iBook that's twice as thick and twice as heavy. Is it not surprising that some features were sacrificed for the size? The 12 pound notebook has even more power and features than your iBook - why don't you buy one?

For some people, the size is more important than the other things that you mention. (I know one woman who absolutely loves her little Toshiba - it fits in a file pocket in her standard folio, so she doesn't even need a laptop bag!)

It's about choice, you don't have to buy one - but it's there for the people who do want a very light and compact notebook.