You realize that puts high-end PC hardware against low-end G5's, right? In other words, my two year old Mac is only 5% slower then the fastest PC on the market (in Photoshop, which is the only test on that page that means something) even though there's Macs that are 30% faster than mine out now?
Fact is, PowerPC creams x86. It's cheaper, cleaner, and has lots of room to grow. If/once Apple achieves an even slightly higher marketshare to balance the tremendous economy of scale benefit x86 has, x86 is going to get more and more costly and PPC is gonna get cheaper and faster.
Good post. I'd just add that the Premier comparisons are an absolute joke and shouldn't be used in any real comparisons. If they wanted to compare they should have used Premier on Windows and Final Cut on Macs. I don't know the results but I would hazard a guess that in identical operations the G5 would perform at least as well as the over priced Windows boxes.
I would be with you on that point, if it were true. The x86 chips in the comparison were released as far back as 2 years ago. This was a snapshot of top of the line performance available at that time. While all CPU providers have grown slowly since then, it's not too hard to extrapolate that given current hardware the results would be roughly the same in terms of the performance differential.
The only problem is that you're looking at older software which isn't going to accurately give you a true estimate of the power of the Intel/AMD or PPC chips. PC World needs to update their benchmarks and stop being jokes.
The PowerPC platform finally got the big break it always wanted: All three of the major consoles will ship with PowerPCs, and at least two of those (the PS3 and XBox2) are not going to be computational slouches. IBM has a modular architecture that makes custom PPCs (relatively) cheap, easy and quick to make. Motorola isn't far behind. The x86 world has no answer to this.
Economies of scale are about to get a lot better for IBM. The G4 is about to lose its biggest disadvantage (MaxBus), and its performance per watt is excellent. Both IBM and Motorola are primed to go dual-core with the rest of the industry. And PPC, unlike x86, has always been 64-bit ready.
I'm not even going to go into all the issues that switching ISAs (and losing AltiVec) would raise.
Why on Earth would Apple drop the PPC now? It doesn't make any sense for them to.
On the other hand, it makes a great deal of sense for certain partisans to be very afraid. Apple and PPC are both scoring one success after another right now.
The PowerPC platform finally got the big break it always wanted: All three of the major consoles will ship with PowerPCs, and at least two of those (the PS3 and XBox2) are not going to be computational slouches. IBM has a modular architecture that makes custom PPCs (relatively) cheap, easy and quick to make. Motorola isn't far behind. The x86 world has no answer to this.
Economies of scale are about to get a lot better for IBM. The G4 is about to lose its biggest disadvantage (MaxBus), and its performance per watt is excellent. Both IBM and Motorola are primed to go dual-core with the rest of the industry. And PPC, unlike x86, has always been 64-bit ready.
I'm not even going to go into all the issues that switching ISAs (and losing AltiVec) would raise.
Why on Earth would Apple drop the PPC now? It doesn't make any sense for them to.
On the other hand, it makes a great deal of sense for certain partisans to be very afraid. Apple and PPC are both scoring one success after another right now.
I do not think the recent question has been whether Apple would drop the PPC for the x86 (though I am sure that the 'Skunkworks' keeps current to have choices if something should happen. I believe Steve said it is good to have choices.).
More recently the question has been whether Apple would release an x86 compatible OS for installation on x86 boxes either as a replacement OS for existing boxes or as an OEM install. The subject of whether Apple could withstand such a choice or whether they would prosper is one that has been debated endlessly.
I will say that even some Apple people (off the record) have said that Apple is a software company that happens to make good hardware).
Many of the comments have come, not surprisingly, from IT or industry people who would like to move to an OS that is not such a pain...but on their existing hardware.
The question I wonder about is how much trouble it would actually be to switch to the cell processor. It sounds attractive, but the question is how much difficulty would be involved. It offers the prospect of multiple co-processors to replace Altivec. What with the expected production volume of cell processors it is not difficult to imagine that the price of the processor would be very attractive to Apple. Perhaps attractive enough to have multiple processors at a retail price point that would be more than just "competitive".
On the other hand, MS has done OK just selling software and a little hardware. I am not sure but what there is room for both strategies to work.
Yeah. Ilike how the Athlons were all set up with RAID and 256 MB of RAM, while the G5s were all standard single drive and 128 MB of RAM.
Pretty poor excuse for a true comparison.
If they are trying to see CPU power, they would use similar configs for the PCs. It is only to prove a bias.
At the bottom of the page, they note that they retested the Alienware to be more fair without raid and with 128 MB. The G5 comes up about even with the PC-centric software.
For what it's worth, I think Intel dumped MS on the Xbox 360, not the other way around. Intel was either losing money or just breaking even on those embedded P3s, so losing that deal was no big loss to them (not to mention that there's no Intel advertising on the Xbox). MS was basically selling a PC for $300 at launch, and now only get $150. I think MS has always lost money on the hardware--their profits come from game licenses. This is also why the new Xbox has foregone the hard drive as standard equipment.
Back to the subject, I think it's very possible for Apple to adopt future x86 processors without doing much. AMD and Intel both plan to put virtualization into their next line of CPUs, which means they could run OSX code through the hardware and NOT through a software emulator. I don't know all the details about VT, but I certainly see this as an outlet for Apple without having to change a single "legacy" app.
I also don't think AMD making 970s would help. AMD and IBM collaborate on process technology, and I think both have been a bit stumped on getting to 3.0ghz with their transistor type. However, they did just implement DSL (Dual strained layer) SOI, which allows for a boost in clock speed. I doubt current G5s are seeing this tech, as AMD is just now getting these out (I think, details are sketchy at this time). It COULD mean that a 3.0ghz G5 might be less of a wait than the 2.7. It also might mean that later 2.7s will run cooler than current models.
I do not think the recent question has been whether Apple would drop the PPC for the x86 (though I am sure that the 'Skunkworks' keeps current to have choices if something should happen. I believe Steve said it is good to have choices.).
He did. And I'm sure Apple has investigated them. And I know that there is a faction within Apple that wants this to happen.
But I don't see it. (Of course, I didn't see the Mac mini either, so caveat lector...)
Quote:
I will say that even some Apple people (off the record) have said that Apple is a software company that happens to make good hardware).
That's true. It's also true that it's much easier to write good software for a small, consistent set of hardware. The idea that you can throw any random collection of hardware together sounds great until you run into driver issues, and until you discover that the tradeoff is that the software assumes the lowest common denominator and thus uses a fraction of the potential speed and capability of your hardware.
Quote:
Many of the comments have come, not surprisingly, from IT or industry people who would like to move to an OS that is not such a pain...but on their existing hardware.
I feel their pain. But I also look at the Herculean efforts MS has made to deal with their hardware ecosystem, and I see a problem big enough to kneecap Apple's software development.
Quote:
The question I wonder about is how much trouble it would actually be to switch to the cell processor. It sounds attractive, but the question is how much difficulty would be involved. It offers the prospect of multiple co-processors to replace Altivec.
Cell is a customizable architecture. If you mean the exact processor that will be going into Sony PlayStations, then you're talking about a fair amount of work. Porting would be easy, because Cell contains a PowerPC with AltiVec; but getting good performance, and using the PPUs efficiently, would be a whole other ball of wax.
Something like the variant that's going into the XBox would be a simpler problem: That's just three PPC cores, each of which has an AltiVec unit. Given that OS X runs just fine on G4s, which have minimal out-of-order execution capabilities, I would expect OS X to run well with a small, if not trivial, amount of tuning.
Apple can order up their own variant, of course. The point, generally, is that since Cell is built around at least one PPC processor, and since that processor happens to have AltiVec on board, it's not much work to get OS X bootstrapped on Cell. The questions would be: How well does it run? and how much advantage can it take of how many PPEs (the PPC cores)? How many SPEs (the additional SIMD units)?
Quote:
On the other hand, MS has done OK just selling software and a little hardware. I am not sure but what there is room for both strategies to work.
They have, but not necessarily on terms that Apple would find acceptable. MS hasn't released anything that really lit up the landscape in the last decade or so. It's much harder to release (not to mention, bundle) software like iLife when you have no control over whether the machines your system runs on has the necessary ports, drives, or processing power...
It's interesting that people looking to technological orthodoxy as a salve for Apple's ills. If we should argue for any orthodoxy, then let it be a market (or retail, or consumer) orthodoxy -- I'm searching for the right word. But it's in the wrapper, not the guts, if you catch my meaning.
For example, Apple has doggedly stuck to the AIO form factor for consumer desktops. They believe it's the right way. The problem, a big one, is that consumers do not. Now, after many years, we have the mini -- consumer desktop, cheapish, headless -- and something of a sales explosion. This despite some glaring omissions. Still, it's conventional enough (headless). So, Apple markets-away the problems; in fact, turns them into pluses. It's not expandable/upgradable -- NO, it's small, and simple. No keyboard and mouse -- NO, it's plug and play, it's efficient, it uses what you know and like and is worth saving. etc etc... In truth, they couldn't just market the problems out of mind, if there wasn't some thought to design as well. They made it small, which helps make those argumnents, but the number one thing they did to it was cut off the head. It's a form factor people know and like, and BUY!
What goes inside doesn't matter, so long as it works well. PPCs are working well right now, makes software development and control easier too. No reason to change that.
"Intel Inside", or AMD, or IBM. People don't care. ATI or nVidia? People don't care. PCIe or AGP? People don't care unless salesmen tell they should. But even then, that's a limited crowd of people. Look at how many no-name parts go into PC towers that sell very well at Costco, or Best Buy, or "you name it" electronics Superstore! People don't care. They care only that:
1.) They can buy what they want, when they want...
2.) It doesn't cost more than they want to pay...
3.) It works
That's it, that's the order. You cannot surprise people without first appealing to them. You cannot appeal to them without first acknowledging them.
What does that mean? You have to give them what they want. People want what they know. Before you can teach them something different, you must submit to that discipline.
Matsu is right on how customers think and shop. Apples problem is they have all these crippled down prebuilt machines. They dont and wont make a Mac your way . They make it apples way. Its why 3% market. Gamers are frustrated at Apple because they wont sell 1 fast CPU and if you go the all in one you settle for old,obsolete cheap video. Customer is king not Apple. Once they figure this out and build systems for real people marketshare will go up. At the moment its Mac vs Mac and everything is waterdowned to prop up the top dual machine.
They now need to take the Mac mini concept (headless) and create another Mac between the mini and the Power Mac G5 that has expansion capabilities. $999.
They now need to take the Mac mini concept (headless) and create another Mac between the mini and the Power Mac G5 that has expansion capabilities. $999.
*Actualy, that's not a bad idea. Another headless cheeper computer.
If they just eliminate a lot of the things like dual cpu (just one) less memory slots, expansion slots, only one hard drive bay. They could probably make it smaller and cheaper.
Matsu is right on how customers think and shop. Apples problem is they have all these crippled down prebuilt machines. They dont and wont make a Mac your way . They make it apples way. Its why 3% market. Gamers are frustrated at Apple because they wont sell 1 fast CPU and if you go the all in one you settle for old,obsolete cheap video. Customer is king not Apple. Once they figure this out and build systems for real people marketshare will go up. At the moment its Mac vs Mac and everything is waterdowned to prop up the top dual machine.
They now need to take the Mac mini concept (headless) and create another Mac between the mini and the Power Mac G5 that has expansion capabilities. $999.
I see we are switching topics. I tend to agree with Matsu too and hope Apple is getting the message that many, ever most, customers like a separate monitor, not built-in. If Apple does get this message, I hope it splits the products up by markets. The Mac mini sells because it is cute and cheap. I think Apple needs two more products. One a little better than the Mac mini for business and other who like it. The second a small tower for gamers and lighter duty professionals.
The business model would have built in power supply, full size hard and optical drives and a couple expansion slots. The small tower would have a single processor chip, and possibly downsize the current power Mac in other ways too.
Comments
Originally posted by gregmightdothat
You realize that puts high-end PC hardware against low-end G5's, right? In other words, my two year old Mac is only 5% slower then the fastest PC on the market (in Photoshop, which is the only test on that page that means something) even though there's Macs that are 30% faster than mine out now?
Fact is, PowerPC creams x86. It's cheaper, cleaner, and has lots of room to grow. If/once Apple achieves an even slightly higher marketshare to balance the tremendous economy of scale benefit x86 has, x86 is going to get more and more costly and PPC is gonna get cheaper and faster.
Good post. I'd just add that the Premier comparisons are an absolute joke and shouldn't be used in any real comparisons. If they wanted to compare they should have used Premier on Windows and Final Cut on Macs. I don't know the results but I would hazard a guess that in identical operations the G5 would perform at least as well as the over priced Windows boxes.
Originally posted by nowayout11
I would be with you on that point, if it were true. The x86 chips in the comparison were released as far back as 2 years ago. This was a snapshot of top of the line performance available at that time. While all CPU providers have grown slowly since then, it's not too hard to extrapolate that given current hardware the results would be roughly the same in terms of the performance differential.
The only problem is that you're looking at older software which isn't going to accurately give you a true estimate of the power of the Intel/AMD or PPC chips. PC World needs to update their benchmarks and stop being jokes.
Economies of scale are about to get a lot better for IBM. The G4 is about to lose its biggest disadvantage (MaxBus), and its performance per watt is excellent. Both IBM and Motorola are primed to go dual-core with the rest of the industry. And PPC, unlike x86, has always been 64-bit ready.
I'm not even going to go into all the issues that switching ISAs (and losing AltiVec) would raise.
Why on Earth would Apple drop the PPC now? It doesn't make any sense for them to.
On the other hand, it makes a great deal of sense for certain partisans to be very afraid. Apple and PPC are both scoring one success after another right now.
Originally posted by Amorph
The PowerPC platform finally got the big break it always wanted: All three of the major consoles will ship with PowerPCs, and at least two of those (the PS3 and XBox2) are not going to be computational slouches. IBM has a modular architecture that makes custom PPCs (relatively) cheap, easy and quick to make. Motorola isn't far behind. The x86 world has no answer to this.
Economies of scale are about to get a lot better for IBM. The G4 is about to lose its biggest disadvantage (MaxBus), and its performance per watt is excellent. Both IBM and Motorola are primed to go dual-core with the rest of the industry. And PPC, unlike x86, has always been 64-bit ready.
I'm not even going to go into all the issues that switching ISAs (and losing AltiVec) would raise.
Why on Earth would Apple drop the PPC now? It doesn't make any sense for them to.
On the other hand, it makes a great deal of sense for certain partisans to be very afraid. Apple and PPC are both scoring one success after another right now.
I do not think the recent question has been whether Apple would drop the PPC for the x86 (though I am sure that the 'Skunkworks' keeps current to have choices if something should happen. I believe Steve said it is good to have choices.).
More recently the question has been whether Apple would release an x86 compatible OS for installation on x86 boxes either as a replacement OS for existing boxes or as an OEM install. The subject of whether Apple could withstand such a choice or whether they would prosper is one that has been debated endlessly.
I will say that even some Apple people (off the record) have said that Apple is a software company that happens to make good hardware).
Many of the comments have come, not surprisingly, from IT or industry people who would like to move to an OS that is not such a pain...but on their existing hardware.
The question I wonder about is how much trouble it would actually be to switch to the cell processor. It sounds attractive, but the question is how much difficulty would be involved. It offers the prospect of multiple co-processors to replace Altivec. What with the expected production volume of cell processors it is not difficult to imagine that the price of the processor would be very attractive to Apple. Perhaps attractive enough to have multiple processors at a retail price point that would be more than just "competitive".
On the other hand, MS has done OK just selling software and a little hardware. I am not sure but what there is room for both strategies to work.
Originally posted by nowayout11
Depends who you ask, I guess.
Yeah. Ilike how the Athlons were all set up with RAID and 256 MB of RAM, while the G5s were all standard single drive and 128 MB of RAM.
Pretty poor excuse for a true comparison.
If they are trying to see CPU power, they would use similar configs for the PCs. It is only to prove a bias.
At the bottom of the page, they note that they retested the Alienware to be more fair without raid and with 128 MB. The G5 comes up about even with the PC-centric software.
Back to the subject, I think it's very possible for Apple to adopt future x86 processors without doing much. AMD and Intel both plan to put virtualization into their next line of CPUs, which means they could run OSX code through the hardware and NOT through a software emulator. I don't know all the details about VT, but I certainly see this as an outlet for Apple without having to change a single "legacy" app.
I also don't think AMD making 970s would help. AMD and IBM collaborate on process technology, and I think both have been a bit stumped on getting to 3.0ghz with their transistor type. However, they did just implement DSL (Dual strained layer) SOI, which allows for a boost in clock speed. I doubt current G5s are seeing this tech, as AMD is just now getting these out (I think, details are sketchy at this time). It COULD mean that a 3.0ghz G5 might be less of a wait than the 2.7. It also might mean that later 2.7s will run cooler than current models.
Originally posted by RBR
I do not think the recent question has been whether Apple would drop the PPC for the x86 (though I am sure that the 'Skunkworks' keeps current to have choices if something should happen. I believe Steve said it is good to have choices.).
He did. And I'm sure Apple has investigated them. And I know that there is a faction within Apple that wants this to happen.
But I don't see it. (Of course, I didn't see the Mac mini either, so caveat lector...)
I will say that even some Apple people (off the record) have said that Apple is a software company that happens to make good hardware).
That's true. It's also true that it's much easier to write good software for a small, consistent set of hardware. The idea that you can throw any random collection of hardware together sounds great until you run into driver issues, and until you discover that the tradeoff is that the software assumes the lowest common denominator and thus uses a fraction of the potential speed and capability of your hardware.
Many of the comments have come, not surprisingly, from IT or industry people who would like to move to an OS that is not such a pain...but on their existing hardware.
I feel their pain. But I also look at the Herculean efforts MS has made to deal with their hardware ecosystem, and I see a problem big enough to kneecap Apple's software development.
The question I wonder about is how much trouble it would actually be to switch to the cell processor. It sounds attractive, but the question is how much difficulty would be involved. It offers the prospect of multiple co-processors to replace Altivec.
Cell is a customizable architecture. If you mean the exact processor that will be going into Sony PlayStations, then you're talking about a fair amount of work. Porting would be easy, because Cell contains a PowerPC with AltiVec; but getting good performance, and using the PPUs efficiently, would be a whole other ball of wax.
Something like the variant that's going into the XBox would be a simpler problem: That's just three PPC cores, each of which has an AltiVec unit. Given that OS X runs just fine on G4s, which have minimal out-of-order execution capabilities, I would expect OS X to run well with a small, if not trivial, amount of tuning.
Apple can order up their own variant, of course. The point, generally, is that since Cell is built around at least one PPC processor, and since that processor happens to have AltiVec on board, it's not much work to get OS X bootstrapped on Cell. The questions would be: How well does it run? and how much advantage can it take of how many PPEs (the PPC cores)? How many SPEs (the additional SIMD units)?
On the other hand, MS has done OK just selling software and a little hardware. I am not sure but what there is room for both strategies to work.
They have, but not necessarily on terms that Apple would find acceptable. MS hasn't released anything that really lit up the landscape in the last decade or so. It's much harder to release (not to mention, bundle) software like iLife when you have no control over whether the machines your system runs on has the necessary ports, drives, or processing power...
Originally posted by hmurchison
Typically the OS X on Intel crowd loves this topic because it's their way of fantasizing about low cost hardware with a great OS.
... Doesn't everyone?
Originally posted by slughead
... Doesn't everyone?
That's affirmative. But then good PC hardware isn't that much cheaper unless you're building it yourself.
Originally posted by RBR
Apple sells crippled hardware all the time. Wake up!
The 970 has a fixed 2:1 bus multiplier. . .
Read your datasheets!
For example, Apple has doggedly stuck to the AIO form factor for consumer desktops. They believe it's the right way. The problem, a big one, is that consumers do not. Now, after many years, we have the mini -- consumer desktop, cheapish, headless -- and something of a sales explosion. This despite some glaring omissions. Still, it's conventional enough (headless). So, Apple markets-away the problems; in fact, turns them into pluses. It's not expandable/upgradable -- NO, it's small, and simple. No keyboard and mouse -- NO, it's plug and play, it's efficient, it uses what you know and like and is worth saving. etc etc... In truth, they couldn't just market the problems out of mind, if there wasn't some thought to design as well. They made it small, which helps make those argumnents, but the number one thing they did to it was cut off the head. It's a form factor people know and like, and BUY!
What goes inside doesn't matter, so long as it works well. PPCs are working well right now, makes software development and control easier too. No reason to change that.
"Intel Inside", or AMD, or IBM. People don't care. ATI or nVidia? People don't care. PCIe or AGP? People don't care unless salesmen tell they should. But even then, that's a limited crowd of people. Look at how many no-name parts go into PC towers that sell very well at Costco, or Best Buy, or "you name it" electronics Superstore! People don't care. They care only that:
1.) They can buy what they want, when they want...
2.) It doesn't cost more than they want to pay...
3.) It works
That's it, that's the order. You cannot surprise people without first appealing to them. You cannot appeal to them without first acknowledging them.
What does that mean? You have to give them what they want. People want what they know. Before you can teach them something different, you must submit to that discipline.
Originally posted by DHagan4755
They now need to take the Mac mini concept (headless) and create another Mac between the mini and the Power Mac G5 that has expansion capabilities. $999.
*Actualy, that's not a bad idea. Another headless cheeper computer.
If they just eliminate a lot of the things like dual cpu (just one) less memory slots, expansion slots, only one hard drive bay. They could probably make it smaller and cheaper.
Originally posted by Aurora
Matsu is right on how customers think and shop. Apples problem is they have all these crippled down prebuilt machines. They dont and wont make a Mac your way . They make it apples way. Its why 3% market. Gamers are frustrated at Apple because they wont sell 1 fast CPU and if you go the all in one you settle for old,obsolete cheap video. Customer is king not Apple. Once they figure this out and build systems for real people marketshare will go up. At the moment its Mac vs Mac and everything is waterdowned to prop up the top dual machine.
BRAVO!!!
Originally posted by DHagan4755
They now need to take the Mac mini concept (headless) and create another Mac between the mini and the Power Mac G5 that has expansion capabilities. $999.
I see we are switching topics. I tend to agree with Matsu too and hope Apple is getting the message that many, ever most, customers like a separate monitor, not built-in. If Apple does get this message, I hope it splits the products up by markets. The Mac mini sells because it is cute and cheap. I think Apple needs two more products. One a little better than the Mac mini for business and other who like it. The second a small tower for gamers and lighter duty professionals.
The business model would have built in power supply, full size hard and optical drives and a couple expansion slots. The small tower would have a single processor chip, and possibly downsize the current power Mac in other ways too.