Steve Bosell
Registered
forget altivec, is there faster than 1 ghz power4's yet?
why doesn't apple just use the power4 ?
- Thread starter Steve Bosell
- Start date
Not a million but they will/do go into $200,000+ IBM mainframes.
A lot of people here talks about the Power4 as if it is really a possibility which is absurd. You'll see OS X on Intel before you see a Power4 Mac.
They could release the G5 now. It would only be around 600MHz. Apple might as well wait until hell freezes over to release the G5 now. They'll never catch up with AMD or Intel. They are so screwed.
A lot of people here talks about the Power4 as if it is really a possibility which is absurd. You'll see OS X on Intel before you see a Power4 Mac.
They could release the G5 now. It would only be around 600MHz. Apple might as well wait until hell freezes over to release the G5 now. They'll never catch up with AMD or Intel. They are so screwed.
Screwed is _so_ the wrong word here. 
The G4 processor may be falling behind P4 and Athlon lines at some tasks, while it still excels at others (albeit not the important ones). Anyway, the processor development continues and the PowerMacs, as well as the other machines, are a fine choice amongst other PC manufacturers' computers.
And you shouldn't forget that Apple _does_ have a choice. It's a brand name well known in the industry and it has shown many times that it won't just fade away because some analysts think so. Instead, they rise from their ashes like a phoenix. Did so a few times in the past.
So...
1) The G4 is still a viable solution. Maybe Quad-Processor machines can be introduced, too.
2) Apple could start a transition to AMD64 in mid 2003.
3) IBM could, with its powers, create a new chip for Apple. Why not call it a G6, available mid 2003.
4) The X86 path could be chosen.
All those would eventually kick in, but not before mid 2003, of course. But until then, we'll see great G4 machines, a Jaguar of an operating system and some other new devices. They'll all keep Apple alive and healthy.
Please stop to say 'Apple is doomed', because it just ain't true.
The G4 processor may be falling behind P4 and Athlon lines at some tasks, while it still excels at others (albeit not the important ones). Anyway, the processor development continues and the PowerMacs, as well as the other machines, are a fine choice amongst other PC manufacturers' computers.
And you shouldn't forget that Apple _does_ have a choice. It's a brand name well known in the industry and it has shown many times that it won't just fade away because some analysts think so. Instead, they rise from their ashes like a phoenix. Did so a few times in the past.
So...
1) The G4 is still a viable solution. Maybe Quad-Processor machines can be introduced, too.
2) Apple could start a transition to AMD64 in mid 2003.
3) IBM could, with its powers, create a new chip for Apple. Why not call it a G6, available mid 2003.
4) The X86 path could be chosen.
All those would eventually kick in, but not before mid 2003, of course. But until then, we'll see great G4 machines, a Jaguar of an operating system and some other new devices. They'll all keep Apple alive and healthy.
Please stop to say 'Apple is doomed', because it just ain't true.
Paragon
Interstellar traveler
Just out of curiosity...how many programs actually takes advantage of the velocity engine anyway? I know that PS and Jaguar does, but is that it. Why is it that programmers don't use this, is it because it's hard to program to? Why would Apple stay with a chip design where no programmers optimize their products for it?
My concern lies in the fact that G4 processor only gets about a 30% speed increase anually and the G5 is no where in sight. The Athlon and Pentium 4 are increasing exponentially. Where will the G5 be when Intel hits 3GHz? I doubt any where to be found and if by some miracle it does make it to the market, it's only going to debut around 600-1000MHz and cost a fortune on top of that.
The G4 was never intended to be a 1.5GHz processor. Apple is spreading it quite thin as it is at 1GHz. The G5 was supposed to be here by now.
Quad processors are not a viable solution for consumers. Very few home users need SMP and the cost would be outrageous.
Everyone seems disgusted that Apple could go the AMD route. I honestly don't see the problem. As long as it stays a Mac and the performance is top notch, who cares who makes the CPU? Yes, RISC is great at running Photoshop and FCP3, but for everything else, it's just proven to be average. My PB G4 DVI does everything I need it for just the way I want it and I love it but at the same time it's a portable and right now, I would never dream of buying a PowerMac. Performance and pricewise, they just can't compete with my desktop PC.
Apple wanted to be different with the G4 like Sun and SGI with their SPARC and MIPS CPUs. Unfortunately, Apple doesn't cater to a specialized market like Sun and SGI do, though their market share shows otherwise. Programmers just aren't taking the time to optimize for AltiVec or even SSE. Raw power is the path many programmers take, doing little to nothing to optimize their code. This is why the Athlon is such a speed daemon. It doesn't rely on optimization to get the most out of any specific application.
If Apple wants a piece of the home market, highly optimized, high cost, low yield and low speed processors are not the solution. The G4 and G5 are just this.
MHz won. Apple needs to realize this if they want to compete with the PC. If all they want to sell are servers and graphic workstations, then they shouldn't worry about it. Sun and SGI sure don't. Hopefully though they can get someone other than themselves and Adobe to write programs optimized for AltiVec.
Fair enough, they aren't screwed. Chances are they won't lose anything because of the path they've taken, but to me, the reality is, they sure are not going to gain anything either. Especially not market share. When screwed came to mind, my first though was they are no way positioning themselves to compete with the PC even though everything they've been doing recently would make you think otherwise.
I certainly meant no offense, I see the PC market as screwed as well. Just for different reasons.
The G4 was never intended to be a 1.5GHz processor. Apple is spreading it quite thin as it is at 1GHz. The G5 was supposed to be here by now.
Quad processors are not a viable solution for consumers. Very few home users need SMP and the cost would be outrageous.
Everyone seems disgusted that Apple could go the AMD route. I honestly don't see the problem. As long as it stays a Mac and the performance is top notch, who cares who makes the CPU? Yes, RISC is great at running Photoshop and FCP3, but for everything else, it's just proven to be average. My PB G4 DVI does everything I need it for just the way I want it and I love it but at the same time it's a portable and right now, I would never dream of buying a PowerMac. Performance and pricewise, they just can't compete with my desktop PC.
Apple wanted to be different with the G4 like Sun and SGI with their SPARC and MIPS CPUs. Unfortunately, Apple doesn't cater to a specialized market like Sun and SGI do, though their market share shows otherwise. Programmers just aren't taking the time to optimize for AltiVec or even SSE. Raw power is the path many programmers take, doing little to nothing to optimize their code. This is why the Athlon is such a speed daemon. It doesn't rely on optimization to get the most out of any specific application.
If Apple wants a piece of the home market, highly optimized, high cost, low yield and low speed processors are not the solution. The G4 and G5 are just this.
MHz won. Apple needs to realize this if they want to compete with the PC. If all they want to sell are servers and graphic workstations, then they shouldn't worry about it. Sun and SGI sure don't. Hopefully though they can get someone other than themselves and Adobe to write programs optimized for AltiVec.
Fair enough, they aren't screwed. Chances are they won't lose anything because of the path they've taken, but to me, the reality is, they sure are not going to gain anything either. Especially not market share. When screwed came to mind, my first though was they are no way positioning themselves to compete with the PC even though everything they've been doing recently would make you think otherwise.
I certainly meant no offense, I see the PC market as screwed as well. Just for different reasons.
Honestly, I have no exact answer to that but number is very few in my opinion. Photoshop is just a good example because it's usually the only program used to benchmark a Mac against a PC.
Quake 3 is actually optimized for AltiVec. On my PB G4 DVI, it runs circles around my PC version.
AltiVec is hard to program for. Just like SSE is for the Pentium 4. I liken it to the PlayStation 2. A great machine, very powerful but at the same time, very difficult to write games for. There are very few games for the PS2 that use it's hardware to it's full extent. Just like there are very few programs optimized for AltiVec or SSE.
This is why the Athlon kicks around the P4 in most benchmarks. It doesn't rely on optimization, just raw power and brute force.
Apple has stuck with the G4 I presume because they are kind of married to it at the moment. The G5 is no where to be found and I would imagine they already have quite a bit invested in it to back out now.
My one wish though is that they would start releasing the G3 at higher clock speeds and not worry whether or not it passes up the G4 performance wise. If Apple wasn't so stubborn, we could all have 1.5-1.7GHz G3s right now that would likely spank the G4 and would cost us a lot less than a G4 solution as well.
Paragon
Interstellar traveler
One of the problems that I see is that Motorola is making chips for all sort of products and not just for the Apple computers. Intel and AMD are focusing intensely at the PC market which Motorola is not. Now if Apple where to get Motorola off their butt's maybe we would see some real performance increase in their chip's. Unfortunately this will not happen unless Apple buy's a big chunk of Motorola or the design of the chip and have Intel or AMD make them.
As for the VE I don't get why Apple don't make their SW programmers use this more because it looks like a good design.
As for the VE I don't get why Apple don't make their SW programmers use this more because it looks like a good design.
The Velocity Engine, or AltiVec, *is* being used by many applications. Even small ones. Your DivX encoder makes use of it, for example. iTunes makes use of it. Saying that Jaguar is 'only' the operating system here is a bit odd, since you'll be glad to have AltiVec when using it. QuickTime uses AltiVec. But okay, let's just say it doesn't _really_ provide us with real world advantages, since we could have 1.5 GHz G3s by now (which is _not_ true, the fastest G3 runs at 1 GHz at the moment, just like the G4). THAT would be hard to tell the market. We've just gone one generation back, erhm, to bring you those MHz you've been telling us about. Wouldn't look too good. What Apple _needs_ is a 1.8 GHz G4 *right* now. That would help a bit, although the P4 would still dance around our noses MHz-wise.
Btw.: Those numbers, 600-1000 MHz for the G5, they stem from a communications processor used in routers or mobile phones, we're not talking desktop computers there. The G5, would it arrive today as a processor for desktop Macs, would be something between 1.2 and 1.6 GHz. It wouldn't run Mac OS X very stable, but it would. Only: Apple doesn't want to sell you something that isn't stable.
For benchmarks and stuff, go to http://www.queru.com/articles/Benchmarks2.html and you'll find some interesting stuff.
Btw.: Those numbers, 600-1000 MHz for the G5, they stem from a communications processor used in routers or mobile phones, we're not talking desktop computers there. The G5, would it arrive today as a processor for desktop Macs, would be something between 1.2 and 1.6 GHz. It wouldn't run Mac OS X very stable, but it would. Only: Apple doesn't want to sell you something that isn't stable.
For benchmarks and stuff, go to http://www.queru.com/articles/Benchmarks2.html and you'll find some interesting stuff.
Moving to the G5 would be extremely bold, but in the end would just screw things up even more.
What you fail to realize that the G5 is 64 bit. Apple has enough to worry about than making its OS 64 bit clean, making developer tools to offer 64 bit Apps and have it worked out so you can have cooperation between running classic, 32bit carbon/cocoa apps on top of a 64 bit os and lets ALSO not forget that we already shucked out a chunk of money to Adobe and MS for native MacOS 10 apps, now we are going to have to buy all new apps that will be 64 bit clean to take advantage of the G5.
You might as well point a gun to Apple's head and say "guess where the bullet is"
Business wise this would kill apple. We don't even know how much the G5 costs.
I bet you Apple was told by motorola, "Sell more G4's or we can't give you the chips at low costs and won't be able to further development of the G4" Apple then got off its hands and started putting the G4 in every model it could.
I predeict 1.1, 1.3, and 1.5 ghz offerings of the G4 by September and 1.7, 1.9, and 2.1 ghz by Macworld SF. Apple will unveil plans for the first MacOS 10.2 update at MWSF. By May's developer conf, Apple will announce an all new 64 bit MacOS 10 that is in development based around.... ???chip??? this one is a mystery. Apple has 4 choices at that point. G5, Intel, AMD, or SPARQ. My hope is they move to sparq.
What you fail to realize that the G5 is 64 bit. Apple has enough to worry about than making its OS 64 bit clean, making developer tools to offer 64 bit Apps and have it worked out so you can have cooperation between running classic, 32bit carbon/cocoa apps on top of a 64 bit os and lets ALSO not forget that we already shucked out a chunk of money to Adobe and MS for native MacOS 10 apps, now we are going to have to buy all new apps that will be 64 bit clean to take advantage of the G5.
You might as well point a gun to Apple's head and say "guess where the bullet is"
Business wise this would kill apple. We don't even know how much the G5 costs.
I bet you Apple was told by motorola, "Sell more G4's or we can't give you the chips at low costs and won't be able to further development of the G4" Apple then got off its hands and started putting the G4 in every model it could.
I predeict 1.1, 1.3, and 1.5 ghz offerings of the G4 by September and 1.7, 1.9, and 2.1 ghz by Macworld SF. Apple will unveil plans for the first MacOS 10.2 update at MWSF. By May's developer conf, Apple will announce an all new 64 bit MacOS 10 that is in development based around.... ???chip??? this one is a mystery. Apple has 4 choices at that point. G5, Intel, AMD, or SPARQ. My hope is they move to sparq.
600-1000MHz for stability of course and every AltiVec program you mentioned was made by Apple. Not exactly a broad list.
I've never encoded DivX on a Mac but I know DivX movies play like crap on one.
Here's a G3 vs G4 guide. A few years old but a good read. Also, this is when the G3 used to be released on comparable hardware to the G4.
http://www.xlr8yourmac.com/G3CARDS/XLR8G4/G4vsG3.html
The G3 is a great chip. Apple should have never abandoned it. And yes, I am fully confident that it's ready to ship at a lot higher clock speeds than the G4. Apple is just holding it back. It's a conspiracy!
Okay now. Tell me what you've had for breakfast. It's a conspiracy, sure. And to whose profits? If IBM had 1.6 GHz G3s, it would surely want to sell them, too. And if Apple could have chips at 1.6 GHz, it would SURELY want to sell them, too. There's no SENSE in a conspiracy, here. Unless you're implying Intel & Microsoft are behind that one, which is complete nonsense, either.
IBM _does_ have the 1 GHz Sahara processor (G3), that's about it. It might be good for a TiBook, but as I mentioned before, it's bad marketing-wise to make a generation-step back! And Apple can't just change this, not even for you, azosx...
I haven't had anything for breakfast and your last paragraph answered your own question.
Apple has nothing to gain from letting IBM release a 1.6GHz G3 on one of their platforms. It would steal the G4's thunder. The G3 is a low cost solution and Apple wouldn't make the money off it like they do the G4.
IBM announced their 1GHz G3 almost a year ago. Surely you don't believe they haven't been doing anything but sitting on their hands since then.
IBM develops this technology for themselves. Just because they don't sell a 1.6GHz G3 CPU exclusively, doesn't mean they don't profit from the technology behind it. I believe G3 technology goes into their POWER3/4 chips or visa versa.
IBM is getting fat off of all the G3s going into the Nintendo GameCube. I believe they're 450MHz and think that the little beasts smokes the Xbox pound for pound.
You can download the PowerPC 750FX (Sahara) product sheet from http://www-3.ibm.com/chips/techlib/techlib.nsf/techdocs/2FF4861D6755A6CA87256BB1006B1DE6 ... You'll find a PDF there, laying out the chip. It's from 600 to 1000 MHz. Now go to www.chips.ibm.com yourself and prove me wrong. Somehow. Go find me a faster G3 from IBM and I'll say you're right.
I'm surprised that so many are so quick to dismiss the power4.
First off, a huge part of the cost of the chips that you see in IBM's big/medium iron is the staggering 32MB(!) L3 cache, this is not something you would see on consumer level machines, so the cost would be signifigantly reduced (and reduced further by volume pricing, assuming that IBM becomes the primary supplier for mac CPUs). Furthermore, it has been sugested that Apple could partner with IBM to provide either some sort of new chip ("G6") or a customized, high-clock G3. If you're doing that kind R&D anyway, why not go for a pared down Power4, which exists now (unlike the "G6") and excels at SMP (unlike the G3).
Compared to Intel/AMD, Power4 is a no-brainer; it runs powerpc instructions (more or less) so there's no need to (once again) force the mac user base into a major paradigm shift... most code would require a simple recompile, and with a little creative customization (see above) not even that.
The real downside to the Power4 is twofold:
1) Heat. The Power4 draws even more power than a P4 toaster oven and throws off heat to match. This is not a portable chip.
2) Altivec. Apple has a woody for Velocity Engine(TM) and IBM has a bug up its butt about not putting special purpose vector units on its chips. This may well be the biggest impediment, and as usual it's not technical, but political.
I think the Power4 is a great opportunity for Apple. They would still need some other chip for use in portables, but heck, even if I conceed the price issue (which I do not) and say that the power4 is unsuitable for consumer machines, it would still be a great way for apple to offer _real_ workstation class, or even data-center class machines without switching architectures or waiting 2 or 3 millenia for motorolla to get with the program.
Just my $0.02...
-alex.
First off, a huge part of the cost of the chips that you see in IBM's big/medium iron is the staggering 32MB(!) L3 cache, this is not something you would see on consumer level machines, so the cost would be signifigantly reduced (and reduced further by volume pricing, assuming that IBM becomes the primary supplier for mac CPUs). Furthermore, it has been sugested that Apple could partner with IBM to provide either some sort of new chip ("G6") or a customized, high-clock G3. If you're doing that kind R&D anyway, why not go for a pared down Power4, which exists now (unlike the "G6") and excels at SMP (unlike the G3).
Compared to Intel/AMD, Power4 is a no-brainer; it runs powerpc instructions (more or less) so there's no need to (once again) force the mac user base into a major paradigm shift... most code would require a simple recompile, and with a little creative customization (see above) not even that.
The real downside to the Power4 is twofold:
1) Heat. The Power4 draws even more power than a P4 toaster oven and throws off heat to match. This is not a portable chip.
2) Altivec. Apple has a woody for Velocity Engine(TM) and IBM has a bug up its butt about not putting special purpose vector units on its chips. This may well be the biggest impediment, and as usual it's not technical, but political.
I think the Power4 is a great opportunity for Apple. They would still need some other chip for use in portables, but heck, even if I conceed the price issue (which I do not) and say that the power4 is unsuitable for consumer machines, it would still be a great way for apple to offer _real_ workstation class, or even data-center class machines without switching architectures or waiting 2 or 3 millenia for motorolla to get with the program.
Just my $0.02...
-alex.
Well, maybe the 32 MB Cache is what makes this processor that expensive, but then again it's also what makes it that powerful. Could well be that investing in R&D could spill out a sibling of the Power4 that is well suited for further PowerMacs down the line, but then again why not team up with IBM (ditching Mot totally) and develop the next generation PowerPC processor for desktop machines?
Like Steve Jobs said: The G4 has still some life in it. It's not a bad processor design, and Motorola will take it into the next year.
If Steve Jobs thinks they still have time to think about the next options, one option could be to let IBM take control over the PowerPC development. The G3s IBM produces aren't that bad, either. But they're not an option for the future, I believe. They might have been one for present and past - and I still believe in the G3 for notebooks. The G4 is too hot for them.
Like Steve Jobs said: The G4 has still some life in it. It's not a bad processor design, and Motorola will take it into the next year.
If Steve Jobs thinks they still have time to think about the next options, one option could be to let IBM take control over the PowerPC development. The G3s IBM produces aren't that bad, either. But they're not an option for the future, I believe. They might have been one for present and past - and I still believe in the G3 for notebooks. The G4 is too hot for them.
The point about whopping cache => performance is well taken, but I think it's a open question. I don't know how well a Power4 with, say, 3MB of L3 cache would perform (certainly not as well as with 32MB) but that's a question for the Apple and IBM engineers, not a bunch of schmoes on some message board*
As to the next generation chip, that's not too far off from my feelings. Mostly I'm just suggesting that the Power4 may be better starting point for that project than the G3 Sahara or an x86 variant (and perhaps better than any motorola offering as well).
-alex
* - In case it's not abundantly clear: I'm just as big a schmoe as anyone here
My point was not that IBM has 1.5GHz G3s sitting in some warehouse. My point was that common sense dictates several things concerning this issue.
#1, a 1GHz G3 was announced almost a year ago yet to this day, the most we are seeing from Apple is 700MHz. It's obviously being held behind the G4. IBM doesn't have a yield problem, Motorola does. If we can't get more than 100MHz out of a chip per year, then Apple was pretty stupid for going with both the G3 and G4.
#2, IBM has little demand from Apple for the G3 yet they have no problem putting higher clock speed G3 variants in their highend mainframes. If Apple would put a little effort behind pushing the higher clock speed, lower power consumption G3 compared to the G4, I know we'd be seeing a 1.5GHz G3 today. We'd also be saving a lot of money and be seeing better overall performance.
No, I don't think the G3 is the answer to Apple's problems but I would have liked to see it used more. Similiar to what the Celeron is to the P4. I very much would like to see Apple succeed but I would like someone to explain to me how they intend to do this when right now they are trying to directly compete with the PC. It's like they are preparing to go to the Daytona 500 with a bicycle. How the hell do they expect to win?