I remember writing platform independent code that with the same binary ran on an IBM PC XT, WangVS, AT&T 3B2 and IBM System 370. I remember writing platform independent code for DG Aviion 88Ks and OKI i860s. I remember re-writing the file system code for FlexOS 1.42 because it had too much CP/M cruft in it and it was killing performance on our Nixdorf PT 80186s. I remember prototype OSs coming in that had days of the week numbered 0 to 7. I remember laughing at Xenix.
I remember Acorn's prototype ARM addon for the BBC Micro coming in to the office.
I remember asking the old compiler guys what the switches on the blue Altair box did.
I remember wondering why anyone would want 4 colour CGA when they could have ultra crisp green and an IRMA board to switch to your 3270 session.
I remember writing tech docs for our compiler and handing them to our Xerox Star operators to produce a manual.
I remember the first time CodeView replaced my usual MASM degugger.
I remember Cairo, Pink, Taligent...
I remember running Archie for the first time.
I remember Sun coming to us to licence our 64bit Abstract Machine technology and native code generation for this new thing they had called OAK. And us telling them to shove it. Oops.
Fundamentally, I totally agree with you melgross and from what I can gather you with me. Where we differ is that I think of the 32bit X86 architecture as being fundamentally the same as the 80386 back in the day I was writing protect mode DOS extenders and the 64bit architecture as something else. AMD64 or EMT64 is a bodge for today - Intel have said so even if it's just to diss AMD on a marketing level. And yes it will evolve from there. My mad post above is where I reckon it will evolve based on years of writing platform independent code and compilers that produced it. IMHO the future lies in CPU virtualization not in perpetuating X86. I think you're completely right in predicting there will be a 64bit legacy free core CPU with software emulations of X86. I'm just not convinced the core will be based on EMT64. I think that will be just another software emulation. I don't think we've seen what that will be from Intel yet though.
I cannot see the logic in Apple making a complete jump on the x86-x64 train, and ride that into the future. If that is what Apple is going to do, why not tell developers that is what you are going to do?
Quote:
It does amaze me that many of the Apple 'faithful' were so quick to embrace Intel and the legacy X86 strain at that and immediately dismiss PowerPC.
This is true most everyone on this list was saying it was impossible for Apple for Apple to switch from PPC to Intel. My self included I could not fathom it.
I still however read reports from analysts who still say Power has a much brighter future ahead than x86. The only reason x86 dominates so thoroughly is legacy.
Quote:
'What about the PowerBook' but it's going to take Apple a year to 18 months to get an Intel laptop out and IBM and Apple would have to be seriously incompetent to not be able to get low power PPC out in that timeframe.
That?s true I didn?t think of that.
Quote:
The Developer kit isn't about getting people on to Intel, it's about getting them on to XCode, Cocoa and platform independent code.
I made this argument on another thread. This may be a position to move the code base to XCode in IA-32 then it should be easier to go from there.
What this linear thinking ignores is first Apple and Steve Jobs history. Jobs has always wanted the Macintosh to stand out from the commodity PC crowd, which is more crucial today than it ever was.
Second that Intel seems pretty frustrated with the current situation. Intel has innovations it would like to bring to market but is unable because of Microsoft. Without a doubt Intel has ideas in the labs that they haven?t yet made public.
The future battle for Apple/Intel is Windows Vista based on AMD 64 architecture. The only advantage I can see for an Apple/Intel collaboration is to offer the market something different and better. I don?t see the advantage of offering the same in any way.
The x86 has plenty more life left. They will continue to evolve, just as they are RISC at the core, they will take on whatever they need to compete. In doing that they will slowly change, as does everything.
From what I see the reason the x86 continues to live is because billions of dollars of investment keep it going.
If natural selection were allowed to work the x86 would be gone.
The other obvious question if x86 were such a progressive architecture why has Intel worked so hard to leave it.
I remember writing platform independent code that with the same binary ran on an IBM PC XT, WangVS, AT&T 3B2 and IBM System 370. I remember writing platform independent code for DG Aviion 88Ks and OKI i860s. I remember re-writing the file system code for FlexOS 1.42 because it had too much CP/M cruft in it and it was killing performance on our Nixdorf PT 80186s. I remember prototype OSs coming in that had days of the week numbered 0 to 7. I remember laughing at Xenix.
I remember Acorn's prototype ARM addon for the BBC Micro coming in to the office.
I remember asking the old compiler guys what the switches on the blue Altair box did.
I remember wondering why anyone would want 4 colour CGA when they could have ultra crisp green and an IRMA board to switch to your 3270 session.
I remember writing tech docs for our compiler and handing them to our Xerox Star operators to produce a manual.
I remember the first time CodeView replaced my usual MASM degugger.
I remember Cairo, Pink, Taligent...
I remember running Archie for the first time.
I remember Sun coming to us to licence our 64bit Abstract Machine technology and native code generation for this new thing they had called OAK. And us telling them to shove it. Oops.
Dude, your old. Just kidding.
I'm sure it has been repeated a million times, but the first thought that came to me after Apple bought Next was how many platforms Next ran on.
For the record, my first college computer programming class was Fourtran written using IBM punch cards, no one but the real uber geeks in the main computer room had CRTs in front of them. Punch the code, run it through the compiler then endless pages of computer paper with error messages extolling your complete stupidity(in not so gentle language I might add).
I remember writing platform independent code that with the same binary ran on an IBM PC XT, WangVS, AT&T 3B2 and IBM System 370. I remember writing platform independent code for DG Aviion 88Ks and OKI i860s. I remember re-writing the file system code for FlexOS 1.42 because it had too much CP/M cruft in it and it was killing performance on our Nixdorf PT 80186s. I remember prototype OSs coming in that had days of the week numbered 0 to 7. I remember laughing at Xenix.
I remember Acorn's prototype ARM addon for the BBC Micro coming in to the office.
I remember asking the old compiler guys what the switches on the blue Altair box did.
I remember wondering why anyone would want 4 colour CGA when they could have ultra crisp green and an IRMA board to switch to your 3270 session.
I remember writing tech docs for our compiler and handing them to our Xerox Star operators to produce a manual.
I remember the first time CodeView replaced my usual MASM degugger.
I remember Cairo, Pink, Taligent...
I remember running Archie for the first time.
I remember Sun coming to us to licence our 64bit Abstract Machine technology and native code generation for this new thing they had called OAK. And us telling them to shove it. Oops.
Fundamentally, I totally agree with you melgross and from what I can gather you with me. Where we differ is that I think of the 32bit X86 architecture as being fundamentally the same as the 80386 back in the day I was writing protect mode DOS extenders and the 64bit architecture as something else. AMD64 or EMT64 is a bodge for today - Intel have said so even if it's just to diss AMD on a marketing level. And yes it will evolve from there. My mad post above is where I reckon it will evolve based on years of writing platform independent code and compilers that produced it. IMHO the future lies in CPU virtualization not in perpetuating X86. I think you're completely right in predicting there will be a 64bit legacy free core CPU with software emulations of X86. I'm just not convinced the core will be based on EMT64. I think that will be just another software emulation. I don't think we've seen what that will be from Intel yet though.
Ok, so then we do mostly agree. we're just looking at it from different angles.
I also started a while ago. I first learned Fortran 4 in Stuyvesant H.S. in the mid sixties. We had just dumped the cyclotron in the basement that Kruschov had been shown, for the first computer in an American H.S. I was taught that at the Courant Mathematical Institute in lower Manhattan, near my school. They had what was considered the first supercomputer at the time, though they didn't use that term back then. It was the CDC 6000. The techs used to walk around in white spacesuits without the backpack, or it looked like spacesuits. The rooms had airlocks with positive pressure. Inside (the rare times we were allowed in), it looked just like a sci-fi movie. Before I graduated in '67 they replaced it with the 6600.
I remember someone from IBM coming to our school to give a lecture on future computing. He said that in the future, everyone would be using THE (caps mine) computer. And that therefore everyone would have to learn to program.
Well, now (almost) everyone uses A computer, or owns one, and if you have a fairly new car, two or three dozen. Probably several PPC's.
There is a very compelling reason. The G4 is totally obsolete, and becoming more so every day.
I 90% agree here. It's still a pretty capable CPU for most people except for the 10% of people after desktop performance out of a 1" thick laptop. But yes, the writing is on the wall for it comparatively versus the Wintel competition at the high end. People will buy faster computers even if they don't need the extra speed. The high end Intel is also coming down in price - 1.7Ghz Pentium Ms are getting into iBook territory.
The problem here is, what happens in the next year. Apple need a laptop to compete at the high end with the 2.xGhz Pentium M and they don't have an answer for that on the table that we know of. There's the 7448 which might be ok to stem the tide for a while but Yonah is going to wipe that out next year. The G5 is too slow or too hot and the 8641D is a major change and too late. They better be on schedule with the Intel switch.
Quote:
Originally posted by melgross
The G5 is behind the latest Xenon's, and even further behind the latest Opterons.
Totally disagree.
I presume you mean Xeon and not the Xbox Xenon processor.
Genesi announced dual 970MP workstations recently.
Xeons are so far behind Opterons and G5s that it's frankly embarrassing. Where's the dual core Xeon?
Quote:
Originally posted by melgross
There is almost no development dollars going into either line.
If you go to Freescale's site, you will see that Apple's chips, the 7447a, and the as yet unreleased 7448, are listed not under "desktop" , or "workstation", or "Mobile". All indicating a continued development for Apple, but under "embedded", indicating automobiles, machinery, etc. Apple is using chips for their computers that are being developed for CARS!
It's always been the case. Get over it. I've a washing machine with a faster MIPS processor than the SGI Indy I used to use. Actually, I've a feeling my Playstation has a faster MIPS CPU.
And Apple always get chips that aren't even listed anywhere by Freescale.
Quote:
Originally posted by melgross
IBM is no better, even though the G5 isn't relegated to the embedded category, IBM's interest in appealing to Apple's needs have waned.
Who's fault is that? Not IBMs for sure. Now Apple are trying it on with Intel too if this rumour has any basis in reality.
And if a chip being embedded was any kind of knock back, wtf are IBM doing building supercomputers with their PPC 440 series chip?
Quote:
Originally posted by melgross
IBM has more experience in designing and building multi-core chips than ANYONE else. It was expected that they would have a dual core chip for Apple over a year ago. There is even some vague mention of development work in IBM's documentation. So what happened? IBM has had major problems at its G5 chip plants, it has spread development out over several years instead of rushing to compete with AMD and Intel. Why? Because it isn't worth it to them to spend all that money at once. IBM can afford to do what it wants at its own pace. If that pace doesn't meet Apple's needs, so be it.
Kind of agree there. Intel and AMD are at war with each other so the competition sparks development. If Apple aren't prepared to pay IBM then IBM aren't going to develop faster processors for free.
Quote:
Originally posted by melgross
Remember that Apple's leaving IBM will result in less than a 5% loss in it's chip division, which itself is only about 15% of the whole company. IBM won't even notice the loss. And if IBM was losing money on it, it will benefit from Apple leaving.
I think the problem lies with the way the IBM giant works. Each division effectively sells to the other. The chip division thought it could sell a load of 970s to the blade guys but in the end it made more sense to them to use Opterons and Xeons.
And the chip division got paid a load by Sony, Microsoft and Nintendo to produce chips for them leaving little Apple and the IBM Blade guys out in the cold.
I've come across this loads of times working for IBM. I once worked on IBM POS - Point of Sale not the other POS ;-) and we were updating a DOS based POS system to something based on FlexOS (an evolution of CP/M from DR). This was back when IBM were selling OS/2. It was bizarre that we were having to use a non IBM OS but the POS guys were being charged too much by the IBM OS division. DR were cheaper.
Quote:
Originally posted by melgross
Nah, there are VERY good reasons for Apple leaving.
I'm sure there are. I just don't think we've heard them and I'm not sure we ever will.
I remember writing platform independent code that with the same binary ran on an IBM PC XT, WangVS, AT&T 3B2 and IBM System 370. I remember writing platform independent code for DG Aviion 88Ks and OKI i860s. I remember re-writing the file system code for FlexOS 1.42 because it had too much CP/M cruft in it and it was killing performance on our Nixdorf PT 80186s. I remember prototype OSs coming in that had days of the week numbered 0 to 7. I remember laughing at Xenix.
You used an OS with an 8 day week? I hope they put in an extra weekend day and not a workday.
My first computing experiences were using TRS-80 Model I (remember when Tandy had their computer stores?), the Commodore PET -> VIC-20 -> Commodore 64/128 & the TI99-4/A. Of course I was a kid then so my programming consisted mainly of punching in all those DATA statements from BYTE/Antic/Compute! magazines. Nothing like encasulating ASM into DATA statements so the BASIC interpreter could POKE it all into memory. Ahh, the good old days.
I 90% agree here. It's still a pretty capable CPU for most people except for the 10% of people after desktop performance out of a 1" thick laptop. But yes, the writing is on the wall for it comparatively versus the Wintel competition at the high end. People will buy faster computers even if they don't need the extra speed. The high end Intel is also coming down in price - 1.7Ghz Pentium Ms are getting into iBook territory.
The problem here is, what happens in the next year. Apple need a laptop to compete at the high end with the 2.xGhz Pentium M and they don't have an answer for that on the table that we know of. There's the 7448 which might be ok to stem the tide for a while but Yonah is going to wipe that out next year. The G5 is too slow or too hot and the 8641D is a major change and too late. They better be on schedule with the Intel switch.
Totally disagree.
I presume you mean Xeon and not the Xbox Xenon processor.
Genesi announced dual 970MP workstations recently.
Xeons are so far behind Opterons and G5s that it's frankly embarrassing. Where's the dual core Xeon?
It's always been the case. Get over it. I've a washing machine with a faster MIPS processor than the SGI Indy I used to use. Actually, I've a feeling my Playstation has a faster MIPS CPU.
And Apple always get chips that aren't even listed anywhere by Freescale.
Who's fault is that? Not IBMs for sure. Now Apple are trying it on with Intel too if this rumour has any basis in reality.
And if a chip being embedded was any kind of knock back, wtf are IBM doing building supercomputers with their PPC 440 series chip?
Kind of agree there. Intel and AMD are at war with each other so the competition sparks development. If Apple aren't prepared to pay IBM then IBM aren't going to develop faster processors for free.
I think the problem lies with the way the IBM giant works. Each division effectively sells to the other. The chip division thought it could sell a load of 970s to the blade guys but in the end it made more sense to them to use Opterons and Xeons.
And the chip division got paid a load by Sony, Microsoft and Nintendo to produce chips for them leaving little Apple and the IBM Blade guys out in the cold.
I've come across this loads of times working for IBM. I once worked on IBM POS - Point of Sale not the other POS ;-) and we were updating a DOS based POS system to something based on FlexOS (an evolution of CP/M from DR). This was back when IBM were selling OS/2. It was bizarre that we were having to use a non IBM OS but the POS guys were being charged too much by the IBM OS division. DR were cheaper.
I'm sure there are. I just don't think we've heard them and I'm not sure we ever will.
Some people here are talking up the idea that Apple won't have a laptop for 12 - 18 months. That's nonsense. We know that they will have at least one out by June. No more than 9 months away. Yonah will be fine for starters.
Of course I mean the Xenon workstation ships. Those chips are only from 5%- 15% slower than an equivelent Opteron. It depends on what they are being used for really. Intels chips are optimized for different software than AMD's chips are. They work better for rendering, while AMD's are better for gaming. Half one, half the other. But both beat the G5 at this time.
Get over it? What is that supposed to mean? And it hasn't "always been the case". Only in the past couple of years has it "been the case".
Which chips has Apple gotten that Freescale not announced or listed? There was ony mythical chip that didn't exist. It had a "b" designation on the case. Everyone thought it was some *special* Apple only chip. It was only a test version that never got out. I already went over this several months ago, and proved it to those who were so hot about it.
There no rumors that Intel's interest in Apple have waned as IBM's did. Where did you read that one?
When Apple and IBM drew up their contract for G5 development and production, IBM knew full well how many chips Apple would be needing. Anyone who thinks differently is ignorent of the way business works. When I entered into negotions over part purchasing, we all put our cards on the table. You contract for numbers, you don't hedge, you don't fudge. IBM knew what their costs would be. There is no excuse. They were losing money on this because of their plant failures and the industrywide inability to predict the leakage current rise.
While Intel and AMD depend on fixing that problem as fast as they could, IBM didn't. With the plant failure to supply Apple on time, they had to hit a penalty phase in the contract. I can understand them giving up. But that doesn't make it any better.
Supercomputers are specialized machines. Why are the Apple G5 machines so good for those tasks, yet fall 50% behind on a video render?
Apple shouldn't have to pay IBM for development. They had a contract. That should be enough. If it wasn't, then IBM had no business agreeing to it in the first place. No excuse.
IBM has always had too much internal politics. A little story.
When IBM still had 27% of the PC business, and they were teaming up with Apple on Pink, the software division said to the PC division that instead of installing Windows on their machins with an option of OS 2, they should do it the other way around, the PC people refused.
At the time, Windows had still to capture most of the market. The industry experts at the time were saying that the superiority of OS 2 would win out if IBM made it the default system, especially with an alliance of Apple. It didn't happen, and we are where we are today, possibly because of it.
Remember that IBM is a marketing company. They are not a technology company. They seem to be one, but they are not. They will sell whatever the customer wants, even at the expense of their own home grown products.
From what I see the reason the x86 continues to live is because billions of dollars of investment keep it going.
If natural selection were allowed to work the x86 would be gone.
The other obvious question if x86 were such a progressive architecture why has Intel worked so hard to leave it.
None of that matters. The point is that it WILL be around in a more modernized form.
I think it will matter. The reason x86 enjoys its comfortable place right now is because of Microsoft's dominance. AMD saw an opportnity to advance themselves and thwart Intel's latest efforts because Intel took a chance at leaving x86.
The situation will not always be this way. Eventually someone will come up with a better idea and find a to implement it.
This has historically been the role that Steve Jobs has played.
None of that matters. The point is that it WILL be around in a more modernized form.
I think it will matter. The reason x86 enjoys its comfortable place right now is because of Microsoft's dominance. AMD saw an opportnity to advance themselves and thwart Intel's latest efforts because Intel took a chance at leaving x86.
The situation will not always be this way. Eventually someone will come up with a better idea and find a to implement it.
This has historically been the role that Steve Jobs has played. [/QUOTE]
Eventually, eventually, eventually. And when might that be? Five years from now? Ten?
Look, this is getting tiring. Apple is going to x86, and that's all there is to it.
People might not like that, but tough. MS isn't going anywhere anytime soon. Don't believe it. If ten years from now MS has 70% of the market, that will still be control.
Apple isn't listening to anyone here. They made their decision based on information that nobody here can claim to know. All we can do is to look at what they have done and try to disern, along with the help of the industry experts in these areas, why they have done it. That's what I try to do. I talk to my friends in Apple engineering as well. If Apple wanted to go to a high end 64 bit chip line that is out and working well, they could have gone to POWER, as some of the rumormongers were saying. But that would cost them $2,000 a chipset and uses almost 200 watts. Not going to happen. Same with IA-64. Same with every other 64 bit design - except for the x86.
Some people here are talking up the idea that Apple won't have a laptop for 12 - 18 months. That's nonsense. We know that they will have at least one out by June. No more than 9 months away. Yonah will be fine for starters.
12 months from when they announced the switch.
Quote:
Originally posted by melgross
Of course I mean the Xenon workstation ships.
XEON = Intel server CHIPS.
XENON = PowerPC chip in the Xbox 360.
Quote:
Originally posted by melgross
Those chips are only from 5%- 15% slower than an equivelent Opteron.
Problem is, there's no dual core Xeon. Dualcore Opterons have been out 5 months. The Opteron 280 and 880 at 2.4Ghz is due next month and the 180 single core at 2.4Ghz still beats the pants off the Xeon.
Intel are getting roasted in the server market now.
And they have massive shortages on some of their support chipsets at the moment.
The Intel switch isn't a bed of roses.
Quote:
Originally posted by melgross
It depends on what they are being used for really. Intels chips are optimized for different software than AMD's chips are. They work better for rendering, while AMD's are better for gaming. Half one, half the other. But both beat the G5 at this time.
I'm not going to argue benchmarks because you've already stated that you don't trust the barefeats benchs, which are the only ones around for cross platform benches.
I don't think you're right there - I've two dual Xeon 3.06Ghz, three dual Opteron 244s, a 242 and an iMac G5 so I couldn't give you ultimate benchmarks anyway but safe to say, I only buy Opteron 244s presently to use as web and database servers.
Quote:
Originally posted by melgross
Get over it? What is that supposed to mean? And it hasn't "always been the case". Only in the past couple of years has it "been the case".
Back as far as at least the 603 I remember Motorola expounding the virtues of it's chips in the embedded market. The IBM 750CX did quite well there too.
That's all I meant - chips get used all over the place so there's no shame in having a computer based on an embedded processor.
Quote:
Originally posted by melgross
Which chips has Apple gotten that Freescale not announced or listed? There was ony mythical chip that didn't exist. It had a "b" designation on the case. Everyone thought it was some *special* Apple only chip. It was only a test version that never got out. I already went over this several months ago, and proved it to those who were so hot about it.
Apple, for some time now, have been running chips faster than the speeds quoted by Moto/freescale. See the current PowerBooks v published 7447A specs for instance.
Quote:
Originally posted by melgross
There no rumors that Intel's interest in Apple have waned as IBM's did. Where did you read that one?
See original story - Apple seeking chips early, Intel saying no way.
Quote:
Originally posted by melgross
When Apple and IBM drew up their contract for G5 development and production, IBM knew full well how many chips Apple would be needing. Anyone who thinks differently is ignorent of the way business works. When I entered into negotions over part purchasing, we all put our cards on the table. You contract for numbers, you don't hedge, you don't fudge. IBM knew what their costs would be. There is no excuse. They were losing money on this because of their plant failures and the industrywide inability to predict the leakage current rise.
While Intel and AMD depend on fixing that problem as fast as they could, IBM didn't. With the plant failure to supply Apple on time, they had to hit a penalty phase in the contract. I can understand them giving up. But that doesn't make it any better.
I also think IBM thought they were on to a winner themselves for their own servers and workstations. It hit them also.
Supercomputers are specialized machines. Why are the Apple G5 machines so good for those tasks, yet fall 50% behind on a video render?
Quote:
Originally posted by melgross
Apple shouldn't have to pay IBM for development. They had a contract. That should be enough. If it wasn't, then IBM had no business agreeing to it in the first place. No excuse.
Sure but that doesn't help Apple. No development and they've no business either.
Quote:
Originally posted by melgross
IBM has always had too much internal politics. A little story.
When IBM still had 27% of the PC business, and they were teaming up with Apple on Pink, the software division said to the PC division that instead of installing Windows on their machins with an option of OS 2, they should do it the other way around, the PC people refused.
At the time, Windows had still to capture most of the market. The industry experts at the time were saying that the superiority of OS 2 would win out if IBM made it the default system, especially with an alliance of Apple. It didn't happen, and we are where we are today, possibly because of it.
If you were an IBM shop back then almost all the PS/2s came with OS/2 installed. IBM's real problem was that the PS/2 was a flop. They held on to the MCA architecture (I think only Olivetti licenced it) at a time when Compaq, HP, Elonex and then Dell were all firmly in the ISA and EISA camp with much cheaper computers. I remember paying nearly 10,000 GBP for a 486-25 upgrade for my Model 80. IBMs were just stupidly expensive.
In the consumer space they had the miserable PS/1. IBM just sucked.
It was immediately obvious from the very first Chicago betas we had as developers that Microsoft were going to win. OS/2 1.2 then needed twice the ram, a faster CPU and was plain ugly. 2.0 was much better but too late. Then NT 3.1 arrived and it was complete. If IBM weren't paying us to write compilers for them then we'd have switched to Windows entirely.
So, I don't think IBM had much say in the matter. They were soundly beaten by Microsoft no matter if their object oriented workplace shell was technically better or not. People wanted pretty and easy - not power.
Quote:
Originally posted by melgross
Remember that IBM is a marketing company. They are not a technology company. They seem to be one, but they are not. They will sell whatever the customer wants, even at the expense of their own home grown products.
That's a little unfair - they do a lot in the technology field before anyone else. But it's also true. IBM are a multi headed beast that more often than not, bites their own tail.
Problem is, there's no dual core Xeon. Dualcore Opterons have been out 5 months. The Opteron 280 and 880 at 2.4Ghz is due next month and the 180 single core at 2.4Ghz still beats the pants off the Xeon.
Intel are getting roasted in the server market now.
And they have massive shortages on some of their support chipsets at the moment.
The Intel switch isn't a bed of roses.
I'm not going to argue benchmarks because you've already stated that you don't trust the barefeats benchs, which are the only ones around for cross platform benches.
I don't think you're right there - I've two dual Xeon 3.06Ghz, three dual Opteron 244s, a 242 and an iMac G5 so I couldn't give you ultimate benchmarks anyway but safe to say, I only buy Opteron 244s presently to use as web and database servers.
Back as far as at least the 603 I remember Motorola expounding the virtues of it's chips in the embedded market. The IBM 750CX did quite well there too.
That's all I meant - chips get used all over the place so there's no shame in having a computer based on an embedded processor.
Apple, for some time now, have been running chips faster than the speeds quoted by Moto/freescale. See the current PowerBooks v published 7447A specs for instance.
See original story - Apple seeking chips early, Intel saying no way.
I also think IBM thought they were on to a winner themselves for their own servers and workstations. It hit them also.
Supercomputers are specialized machines. Why are the Apple G5 machines so good for those tasks, yet fall 50% behind on a video render?
Sure but that doesn't help Apple. No development and they've no business either.
If you were an IBM shop back then almost all the PS/2s came with OS/2 installed. IBM's real problem was that the PS/2 was a flop. They held on to the MCA architecture (I think only Olivetti licenced it) at a time when Compaq, HP, Elonex and then Dell were all firmly in the ISA and EISA camp with much cheaper computers. I remember paying nearly 10,000 GBP for a 486-25 upgrade for my Model 80. IBMs were just stupidly expensive.
In the consumer space they had the miserable PS/1. IBM just sucked.
It was immediately obvious from the very first Chicago betas we had as developers that Microsoft were going to win. OS/2 1.2 then needed twice the ram, a faster CPU and was plain ugly. 2.0 was much better but too late. Then NT 3.1 arrived and it was complete. If IBM weren't paying us to write compilers for them then we'd have switched to Windows entirely.
So, I don't think IBM had much say in the matter. They were soundly beaten by Microsoft no matter if their object oriented workplace shell was technically better or not. People wanted pretty and easy - not power.
That's a little unfair - they do a lot in the technology field before anyone else. But it's also true. IBM are a multi headed beast that more often than not, bites their own tail.
Whoa! These posts are getting way too long.
I'll only respond to a few points if that's alright with you.
People here are still talking 12-18 months, that's why I mentioned it.
Sorry, I see the confusion. XEON never comes off my fingertips. The spellcheck isn't much help either.
A dual core Xeon is no better than two single core Xeons, and that's where the comparisons are being made. Dual chip Macs vs. dual chip Xeons and Opterons. Real-world tests show the G%'s to be slower on average - up to 50% in some tests. I wish it weren't so, but it is. It's not all the fault of the chip memory bandwidth is also a problem. While the G5's have great theoretical bandwidth, the memory controller doesn't. This is a well known issue.
X86 still has almost 90% of the x86 server sales. You know, there is no reason to believe that AMD's fortunes will hold. Intel isn't standing still. Now that they're off the GHz kick, their designs are becoming more like IBM and AMD's. Things could easily reverse later in 2006 or early 2007. About the time Apple moves to those designs.
The shortage is in the older low end chip. It's going to be discontinued soon anyway.
About the chips, yes, embedded has always been an avenue that PPC has gone. Most PS Laser printers use PPC chips. No problem there. But Apple's chips were never those embedded designs.
Apple has er, "clocked" it's chips before. Remember the dual 1.42GHz machines? Those were 1.25GHz chips. You could see that on the case if you were brave enough to remove the heatsink.
That story about Apple seeking "early" production runs is from the inquirer. A site that I have regular run-ins with. They are not even close to being a neutral site.
They recently had a story about new graphics boards coming out that could run two 2560x1600 monitors. They said that these boards would be the first boards that could do that.
I pointed out that Apple had a board from Nvidia called the 6800Ultra. They said that yes, they knew but it was a Mac board. I then pointed out that, yes, it was, but that these were Mac monitors, as there are no other monitors that display 2560x1600. Needless to say, there was no point to continuing.
Disregard the article, most have.
The thing about IBM here is that they aren't willing to nurture a product. If IBM hadn't screwed up in the first place as did Motorola, and now Freescale, Apple might be selling 6 million computers this year instead of 4.5 million. Who knows, if the problems hadn't gone back to 1999 with Moto, Apple might now be selling 7 million units.
Look at MS, the junkyard dog that it is. It has lost over $4 billion dollars in the past 4 years itwh the XBox, and is expected to continue to lose money. But they want that market.
Look at the same thing with them and Palm. MS was badly beaten for years, but they persist. Now Palm is on the ropes. MS beat them. How much did it cost?
When Intel's chips were out of steam in the early '90's, they invested billions and got themselves out from under it.
If IBM were to look at the future possibilities and properly fund this, Apple's sales could be much higher. It's sad. I'd much rather see the G5 - 6 - 7, etc, than x86, but it ain't gomma happen. Too bad really.
I think that Apple is rather CPU agnostic (to a limited extent of course). If they choose Intel, it's because for laptops (half of the market) and because it's the number one.
With this switch Apple wanted to have the most common line of chips. In this way they will never be behind nor beyond. Apple want to highlight the best of it's qualities : design, stability, quality.
I think that Apple is rather CPU agnostic (to a limited extent of course). If they choose Intel, it's because for laptops (half of the market) and because it's the number one.
With this switch Apple wanted to have the most common line of chips. In this way they will never be behind nor beyond. Apple want to highlight the best of it's qualities : design, stability, quality.
That's true. They need a vendor who will supply them what they need, when they need it. And if that vendor slips up, it will be for everyone, not just Apple.
Apple is going to x86, and that's all there is to it.
Hmmm. I still look at this as one scenario out of a couple.
Largely because the primary reasons most everyone gives for Apple switching to Intel solves one set of problems, but do not address the new set of problems Apple will have.
Apple will now have a more stable supplier than IBM or Freescale. Intel is in competition with AMD which continues to fuel Intel's need for development. If Intel slips in its projections the entire market slips. All true.
But that is just one part of it, that's not a holistic view of what makes a successful business. Turning to x86 Apple flirts with some very big risks.
Hypethetically Apple does go full x86-x64.
First of all this would reinforce Microsoft as the dominate standards leader of the PC industry. Microsoft is in the best position to chart the course of future x86 design.
If Windows Vista turns out to be on par with OS X or is even perceived to be good enough. This will leave less motivation for people to try OS X or buy a Macintosh. Apple being in hardware lock step will impede their ability to make radical hardware changes to differentiate themselves from all other OEM's. The same as it did in 1999 when Apple abandoned parallel ports and floppy drives. That was an extreme radical change that was rewarded in high volume sales of iMac's.
OS X on x86. There is a small but motivated group that vows to break any protection Apple will use to keep OS X only on Macintosh. This is a group who mostly builds their own computers and would not buy a Macintosh at all. The ultimate goal of this group is to force Apple into selling OS X as a stand alone product. It is very possible this group will successfully hack future versions of OS X, write drivers for it, and spread to any one whom ever wants it. When you look at the fact that much of the world has no problem using pirated copies of Windows, this could be a big problem for Apple.
x86 will mean a Macintosh will have the exact same innards as a Dell or HP Pavilion. Apple would be forced to lower its price and come in line with general PC pricing. Which would lower Apple's revenue margins and profit. And at this particular time when medium and small OEM's are slowly dying or being consolodated into the larger OEM's. There is no way Apple could win a price war with Dell. From a hardware standpoint there is no way Apple could battle Dell in price for performance. This would be exacerbated if Vista was deemed good enough or pirated copies of OS X were rampant and common. Dell would have the ability to lower its hardware prices and essentially wait as Apple dies.
Because of poor hardware sales or piracy Apple decides to release OS X as a product. This sceanrio would pit OS X head to head with Windows. There's no need for the bloody details we all know how that would go.
What does Intel gain from any of this? Apple leaving IBM for Intel is a social feather in Intel's cap. But from a practical or business standpoint what does Intel gain? It might be optimistic if Apple even added 2% to Intel's profits.
Intel would have given up one of its best and possibly last chances to advance technology that will not be supported by Microsoft or the larger PC industry.
x86 will not mean the exact same innards as a Dell or an HP. SGI Oxygen. x86 is not going to turn Apple into a commodity box maker. Apple has always been about the marketing package, not the individual components. x86, PPC, or 6502 this doesn't change. Apple is not a feather in Intels cap without x86 at this point. That is a market reality for Intel. Why does [nearly] everyone all of a sudden think Apple will be competing in the commodity box market? Absurd.
x86 does nothing to reinforce Microsoft in anything. If anything it weakens their market confidence. How does Apple justify developers porting to the specific vaguaries of x86 and then tell Adobe "Oh btw rewrite that optimized code you just rewrote for our new architecture please. We just commited you to 150 man years of labour to hide our corporate plans, you can throw that code away."
Intel's opportunity to update the ISA was before Microsoft coded 64 bit windows. Oncce AMD beat them to the punch, the window closed. Apple is Not the same kind of window, because they do not represent nearly the same install base. Maybe 5 years from now there will be another window.
It's sad. I'd much rather see the G5 - 6 - 7, etc, than x86, but it ain't gomma happen.
Why is IBM the only chip maker who could manufacture exotic high performance chips for Apple.
Intel is actually in a much better position to experiment and take risks that IBM is able.
Hypethetically Apple uses a new Intel architecture.
In Intel, Apple would have a dedicated chip supplier. Intel has just as much at stake in this collaboration as Apple does. Intel is in direct competition with AMD and in many ways with Microsoft. This gives Intel motivation to produce and supply Apple with the chips it need.
Apple and Intel would compete against Microsoft and AMD with a better performing platform.
If Windows Vista were on par or percieved as good enough in comparison to OS X. Jobs can compare the performance of OS X/new Intel architecture to Vista and AMD64. Apple maintains the ability to change hardware specs as subtly or radically as it chooses.
OS X on a new Intel architecture would be impossible to hack and run on x86. Funny however Apple has not really shown much concern over this. That warning a month ago to websites was a slap on the wrist, but no legal action has been taken at all. Think about it. I'm sure all of the developer versions Apple sent out had serial numbers or some type of identification attatched. Would not take much effort for Apple to identify which developer the pirated OS X came from and prosecute them for breaking their NDA. Apple has done none of this and doesn't seem too concerned about it.
Intel would gain the ability to showcase its newest technologies that MS is not willing to adopt. If the performance and productivity of Macintosh, OS X, and the new Intel architecture exceed Microsoft and AMD. This would force both to change or adopt what Intel and Apple are doing.
I'm know Apple executives are mulling over all of these scenarios.
With all of the pro's and con's wieghd in they may choose x86. While this scenario would solve Apple's chip supply problem, it carries many other potential problems in the long run.
Apple could choose to work with a new Intel architecture. Which would have more risk in the beginning but has more potential to benefit the future requirements of both Apple and Intel.
Comments
Put it this way...
I remember writing platform independent code that with the same binary ran on an IBM PC XT, WangVS, AT&T 3B2 and IBM System 370. I remember writing platform independent code for DG Aviion 88Ks and OKI i860s. I remember re-writing the file system code for FlexOS 1.42 because it had too much CP/M cruft in it and it was killing performance on our Nixdorf PT 80186s. I remember prototype OSs coming in that had days of the week numbered 0 to 7. I remember laughing at Xenix.
I remember Acorn's prototype ARM addon for the BBC Micro coming in to the office.
I remember asking the old compiler guys what the switches on the blue Altair box did.
I remember wondering why anyone would want 4 colour CGA when they could have ultra crisp green and an IRMA board to switch to your 3270 session.
I remember writing tech docs for our compiler and handing them to our Xerox Star operators to produce a manual.
I remember the first time CodeView replaced my usual MASM degugger.
I remember Cairo, Pink, Taligent...
I remember running Archie for the first time.
I remember Sun coming to us to licence our 64bit Abstract Machine technology and native code generation for this new thing they had called OAK. And us telling them to shove it. Oops.
Fundamentally, I totally agree with you melgross and from what I can gather you with me. Where we differ is that I think of the 32bit X86 architecture as being fundamentally the same as the 80386 back in the day I was writing protect mode DOS extenders and the 64bit architecture as something else. AMD64 or EMT64 is a bodge for today - Intel have said so even if it's just to diss AMD on a marketing level. And yes it will evolve from there. My mad post above is where I reckon it will evolve based on years of writing platform independent code and compilers that produced it. IMHO the future lies in CPU virtualization not in perpetuating X86. I think you're completely right in predicting there will be a 64bit legacy free core CPU with software emulations of X86. I'm just not convinced the core will be based on EMT64. I think that will be just another software emulation. I don't think we've seen what that will be from Intel yet though.
It does amaze me that many of the Apple 'faithful' were so quick to embrace Intel and the legacy X86 strain at that and immediately dismiss PowerPC.
This is true most everyone on this list was saying it was impossible for Apple for Apple to switch from PPC to Intel. My self included I could not fathom it.
I still however read reports from analysts who still say Power has a much brighter future ahead than x86. The only reason x86 dominates so thoroughly is legacy.
'What about the PowerBook' but it's going to take Apple a year to 18 months to get an Intel laptop out and IBM and Apple would have to be seriously incompetent to not be able to get low power PPC out in that timeframe.
That?s true I didn?t think of that.
The Developer kit isn't about getting people on to Intel, it's about getting them on to XCode, Cocoa and platform independent code.
I made this argument on another thread. This may be a position to move the code base to XCode in IA-32 then it should be easier to go from there.
What this linear thinking ignores is first Apple and Steve Jobs history. Jobs has always wanted the Macintosh to stand out from the commodity PC crowd, which is more crucial today than it ever was.
Second that Intel seems pretty frustrated with the current situation. Intel has innovations it would like to bring to market but is unable because of Microsoft. Without a doubt Intel has ideas in the labs that they haven?t yet made public.
The future battle for Apple/Intel is Windows Vista based on AMD 64 architecture. The only advantage I can see for an Apple/Intel collaboration is to offer the market something different and better. I don?t see the advantage of offering the same in any way.
The x86 has plenty more life left. They will continue to evolve, just as they are RISC at the core, they will take on whatever they need to compete. In doing that they will slowly change, as does everything.
From what I see the reason the x86 continues to live is because billions of dollars of investment keep it going.
If natural selection were allowed to work the x86 would be gone.
The other obvious question if x86 were such a progressive architecture why has Intel worked so hard to leave it.
Originally posted by aegisdesign
How old am I?
Put it this way...
I remember writing platform independent code that with the same binary ran on an IBM PC XT, WangVS, AT&T 3B2 and IBM System 370. I remember writing platform independent code for DG Aviion 88Ks and OKI i860s. I remember re-writing the file system code for FlexOS 1.42 because it had too much CP/M cruft in it and it was killing performance on our Nixdorf PT 80186s. I remember prototype OSs coming in that had days of the week numbered 0 to 7. I remember laughing at Xenix.
I remember Acorn's prototype ARM addon for the BBC Micro coming in to the office.
I remember asking the old compiler guys what the switches on the blue Altair box did.
I remember wondering why anyone would want 4 colour CGA when they could have ultra crisp green and an IRMA board to switch to your 3270 session.
I remember writing tech docs for our compiler and handing them to our Xerox Star operators to produce a manual.
I remember the first time CodeView replaced my usual MASM degugger.
I remember Cairo, Pink, Taligent...
I remember running Archie for the first time.
I remember Sun coming to us to licence our 64bit Abstract Machine technology and native code generation for this new thing they had called OAK. And us telling them to shove it. Oops.
Dude, your old. Just kidding.
I'm sure it has been repeated a million times, but the first thought that came to me after Apple bought Next was how many platforms Next ran on.
For the record, my first college computer programming class was Fourtran written using IBM punch cards, no one but the real uber geeks in the main computer room had CRTs in front of them. Punch the code, run it through the compiler then endless pages of computer paper with error messages extolling your complete stupidity(in not so gentle language I might add).
Originally posted by aegisdesign
How old am I?
Put it this way...
I remember writing platform independent code that with the same binary ran on an IBM PC XT, WangVS, AT&T 3B2 and IBM System 370. I remember writing platform independent code for DG Aviion 88Ks and OKI i860s. I remember re-writing the file system code for FlexOS 1.42 because it had too much CP/M cruft in it and it was killing performance on our Nixdorf PT 80186s. I remember prototype OSs coming in that had days of the week numbered 0 to 7. I remember laughing at Xenix.
I remember Acorn's prototype ARM addon for the BBC Micro coming in to the office.
I remember asking the old compiler guys what the switches on the blue Altair box did.
I remember wondering why anyone would want 4 colour CGA when they could have ultra crisp green and an IRMA board to switch to your 3270 session.
I remember writing tech docs for our compiler and handing them to our Xerox Star operators to produce a manual.
I remember the first time CodeView replaced my usual MASM degugger.
I remember Cairo, Pink, Taligent...
I remember running Archie for the first time.
I remember Sun coming to us to licence our 64bit Abstract Machine technology and native code generation for this new thing they had called OAK. And us telling them to shove it. Oops.
Fundamentally, I totally agree with you melgross and from what I can gather you with me. Where we differ is that I think of the 32bit X86 architecture as being fundamentally the same as the 80386 back in the day I was writing protect mode DOS extenders and the 64bit architecture as something else. AMD64 or EMT64 is a bodge for today - Intel have said so even if it's just to diss AMD on a marketing level. And yes it will evolve from there. My mad post above is where I reckon it will evolve based on years of writing platform independent code and compilers that produced it. IMHO the future lies in CPU virtualization not in perpetuating X86. I think you're completely right in predicting there will be a 64bit legacy free core CPU with software emulations of X86. I'm just not convinced the core will be based on EMT64. I think that will be just another software emulation. I don't think we've seen what that will be from Intel yet though.
Ok, so then we do mostly agree. we're just looking at it from different angles.
I also started a while ago. I first learned Fortran 4 in Stuyvesant H.S. in the mid sixties. We had just dumped the cyclotron in the basement that Kruschov had been shown, for the first computer in an American H.S. I was taught that at the Courant Mathematical Institute in lower Manhattan, near my school. They had what was considered the first supercomputer at the time, though they didn't use that term back then. It was the CDC 6000. The techs used to walk around in white spacesuits without the backpack, or it looked like spacesuits. The rooms had airlocks with positive pressure. Inside (the rare times we were allowed in), it looked just like a sci-fi movie. Before I graduated in '67 they replaced it with the 6600.
I remember someone from IBM coming to our school to give a lecture on future computing. He said that in the future, everyone would be using THE (caps mine) computer. And that therefore everyone would have to learn to program.
Well, now (almost) everyone uses A computer, or owns one, and if you have a fairly new car, two or three dozen. Probably several PPC's.
How things have changed!
Originally posted by TenoBell
From what I see the reason the x86 continues to live is because billions of dollars of investment keep it going.
If natural selection were allowed to work the x86 would be gone.
The other obvious question if x86 were such a progressive architecture why has Intel worked so hard to leave it.
None of that matters. The point is that it WILL be around in a more modernized form.
Originally posted by melgross
There is a very compelling reason. The G4 is totally obsolete, and becoming more so every day.
I 90% agree here. It's still a pretty capable CPU for most people except for the 10% of people after desktop performance out of a 1" thick laptop. But yes, the writing is on the wall for it comparatively versus the Wintel competition at the high end. People will buy faster computers even if they don't need the extra speed. The high end Intel is also coming down in price - 1.7Ghz Pentium Ms are getting into iBook territory.
The problem here is, what happens in the next year. Apple need a laptop to compete at the high end with the 2.xGhz Pentium M and they don't have an answer for that on the table that we know of. There's the 7448 which might be ok to stem the tide for a while but Yonah is going to wipe that out next year. The G5 is too slow or too hot and the 8641D is a major change and too late. They better be on schedule with the Intel switch.
Originally posted by melgross
The G5 is behind the latest Xenon's, and even further behind the latest Opterons.
Totally disagree.
I presume you mean Xeon and not the Xbox Xenon processor.
Genesi announced dual 970MP workstations recently.
Xeons are so far behind Opterons and G5s that it's frankly embarrassing. Where's the dual core Xeon?
Originally posted by melgross
There is almost no development dollars going into either line.
If you go to Freescale's site, you will see that Apple's chips, the 7447a, and the as yet unreleased 7448, are listed not under "desktop" , or "workstation", or "Mobile". All indicating a continued development for Apple, but under "embedded", indicating automobiles, machinery, etc. Apple is using chips for their computers that are being developed for CARS!
It's always been the case. Get over it. I've a washing machine with a faster MIPS processor than the SGI Indy I used to use. Actually, I've a feeling my Playstation has a faster MIPS CPU.
And Apple always get chips that aren't even listed anywhere by Freescale.
Originally posted by melgross
IBM is no better, even though the G5 isn't relegated to the embedded category, IBM's interest in appealing to Apple's needs have waned.
Who's fault is that? Not IBMs for sure. Now Apple are trying it on with Intel too if this rumour has any basis in reality.
And if a chip being embedded was any kind of knock back, wtf are IBM doing building supercomputers with their PPC 440 series chip?
Originally posted by melgross
IBM has more experience in designing and building multi-core chips than ANYONE else. It was expected that they would have a dual core chip for Apple over a year ago. There is even some vague mention of development work in IBM's documentation. So what happened? IBM has had major problems at its G5 chip plants, it has spread development out over several years instead of rushing to compete with AMD and Intel. Why? Because it isn't worth it to them to spend all that money at once. IBM can afford to do what it wants at its own pace. If that pace doesn't meet Apple's needs, so be it.
Kind of agree there. Intel and AMD are at war with each other so the competition sparks development. If Apple aren't prepared to pay IBM then IBM aren't going to develop faster processors for free.
Originally posted by melgross
Remember that Apple's leaving IBM will result in less than a 5% loss in it's chip division, which itself is only about 15% of the whole company. IBM won't even notice the loss. And if IBM was losing money on it, it will benefit from Apple leaving.
I think the problem lies with the way the IBM giant works. Each division effectively sells to the other. The chip division thought it could sell a load of 970s to the blade guys but in the end it made more sense to them to use Opterons and Xeons.
And the chip division got paid a load by Sony, Microsoft and Nintendo to produce chips for them leaving little Apple and the IBM Blade guys out in the cold.
I've come across this loads of times working for IBM. I once worked on IBM POS - Point of Sale not the other POS ;-) and we were updating a DOS based POS system to something based on FlexOS (an evolution of CP/M from DR). This was back when IBM were selling OS/2. It was bizarre that we were having to use a non IBM OS but the POS guys were being charged too much by the IBM OS division. DR were cheaper.
Originally posted by melgross
Nah, there are VERY good reasons for Apple leaving.
I'm sure there are. I just don't think we've heard them and I'm not sure we ever will.
Originally posted by rickag
Dude, your old. Just kidding.
Just started young. ;-)
Originally posted by aegisdesign
How old am I?
Put it this way...
I remember writing platform independent code that with the same binary ran on an IBM PC XT, WangVS, AT&T 3B2 and IBM System 370. I remember writing platform independent code for DG Aviion 88Ks and OKI i860s. I remember re-writing the file system code for FlexOS 1.42 because it had too much CP/M cruft in it and it was killing performance on our Nixdorf PT 80186s. I remember prototype OSs coming in that had days of the week numbered 0 to 7. I remember laughing at Xenix.
You used an OS with an 8 day week? I hope they put in an extra weekend day and not a workday.
My first computing experiences were using TRS-80 Model I (remember when Tandy had their computer stores?), the Commodore PET -> VIC-20 -> Commodore 64/128 & the TI99-4/A. Of course I was a kid then so my programming consisted mainly of punching in all those DATA statements from BYTE/Antic/Compute! magazines. Nothing like encasulating ASM into DATA statements so the BASIC interpreter could POKE it all into memory.
Originally posted by aegisdesign
I 90% agree here. It's still a pretty capable CPU for most people except for the 10% of people after desktop performance out of a 1" thick laptop. But yes, the writing is on the wall for it comparatively versus the Wintel competition at the high end. People will buy faster computers even if they don't need the extra speed. The high end Intel is also coming down in price - 1.7Ghz Pentium Ms are getting into iBook territory.
The problem here is, what happens in the next year. Apple need a laptop to compete at the high end with the 2.xGhz Pentium M and they don't have an answer for that on the table that we know of. There's the 7448 which might be ok to stem the tide for a while but Yonah is going to wipe that out next year. The G5 is too slow or too hot and the 8641D is a major change and too late. They better be on schedule with the Intel switch.
Totally disagree.
I presume you mean Xeon and not the Xbox Xenon processor.
Genesi announced dual 970MP workstations recently.
Xeons are so far behind Opterons and G5s that it's frankly embarrassing. Where's the dual core Xeon?
It's always been the case. Get over it. I've a washing machine with a faster MIPS processor than the SGI Indy I used to use. Actually, I've a feeling my Playstation has a faster MIPS CPU.
And Apple always get chips that aren't even listed anywhere by Freescale.
Who's fault is that? Not IBMs for sure. Now Apple are trying it on with Intel too if this rumour has any basis in reality.
And if a chip being embedded was any kind of knock back, wtf are IBM doing building supercomputers with their PPC 440 series chip?
Kind of agree there. Intel and AMD are at war with each other so the competition sparks development. If Apple aren't prepared to pay IBM then IBM aren't going to develop faster processors for free.
I think the problem lies with the way the IBM giant works. Each division effectively sells to the other. The chip division thought it could sell a load of 970s to the blade guys but in the end it made more sense to them to use Opterons and Xeons.
And the chip division got paid a load by Sony, Microsoft and Nintendo to produce chips for them leaving little Apple and the IBM Blade guys out in the cold.
I've come across this loads of times working for IBM. I once worked on IBM POS - Point of Sale not the other POS ;-) and we were updating a DOS based POS system to something based on FlexOS (an evolution of CP/M from DR). This was back when IBM were selling OS/2. It was bizarre that we were having to use a non IBM OS but the POS guys were being charged too much by the IBM OS division. DR were cheaper.
I'm sure there are. I just don't think we've heard them and I'm not sure we ever will.
Some people here are talking up the idea that Apple won't have a laptop for 12 - 18 months. That's nonsense. We know that they will have at least one out by June. No more than 9 months away. Yonah will be fine for starters.
Of course I mean the Xenon workstation ships. Those chips are only from 5%- 15% slower than an equivelent Opteron. It depends on what they are being used for really. Intels chips are optimized for different software than AMD's chips are. They work better for rendering, while AMD's are better for gaming. Half one, half the other. But both beat the G5 at this time.
Get over it? What is that supposed to mean? And it hasn't "always been the case". Only in the past couple of years has it "been the case".
Which chips has Apple gotten that Freescale not announced or listed? There was ony mythical chip that didn't exist. It had a "b" designation on the case. Everyone thought it was some *special* Apple only chip. It was only a test version that never got out. I already went over this several months ago, and proved it to those who were so hot about it.
There no rumors that Intel's interest in Apple have waned as IBM's did. Where did you read that one?
When Apple and IBM drew up their contract for G5 development and production, IBM knew full well how many chips Apple would be needing. Anyone who thinks differently is ignorent of the way business works. When I entered into negotions over part purchasing, we all put our cards on the table. You contract for numbers, you don't hedge, you don't fudge. IBM knew what their costs would be. There is no excuse. They were losing money on this because of their plant failures and the industrywide inability to predict the leakage current rise.
While Intel and AMD depend on fixing that problem as fast as they could, IBM didn't. With the plant failure to supply Apple on time, they had to hit a penalty phase in the contract. I can understand them giving up. But that doesn't make it any better.
Supercomputers are specialized machines. Why are the Apple G5 machines so good for those tasks, yet fall 50% behind on a video render?
Apple shouldn't have to pay IBM for development. They had a contract. That should be enough. If it wasn't, then IBM had no business agreeing to it in the first place. No excuse.
IBM has always had too much internal politics. A little story.
When IBM still had 27% of the PC business, and they were teaming up with Apple on Pink, the software division said to the PC division that instead of installing Windows on their machins with an option of OS 2, they should do it the other way around, the PC people refused.
At the time, Windows had still to capture most of the market. The industry experts at the time were saying that the superiority of OS 2 would win out if IBM made it the default system, especially with an alliance of Apple. It didn't happen, and we are where we are today, possibly because of it.
Remember that IBM is a marketing company. They are not a technology company. They seem to be one, but they are not. They will sell whatever the customer wants, even at the expense of their own home grown products.
From what I see the reason the x86 continues to live is because billions of dollars of investment keep it going.
If natural selection were allowed to work the x86 would be gone.
The other obvious question if x86 were such a progressive architecture why has Intel worked so hard to leave it.
None of that matters. The point is that it WILL be around in a more modernized form.
I think it will matter. The reason x86 enjoys its comfortable place right now is because of Microsoft's dominance. AMD saw an opportnity to advance themselves and thwart Intel's latest efforts because Intel took a chance at leaving x86.
The situation will not always be this way. Eventually someone will come up with a better idea and find a to implement it.
This has historically been the role that Steve Jobs has played.
Originally posted by TenoBell
None of that matters. The point is that it WILL be around in a more modernized form.
I think it will matter. The reason x86 enjoys its comfortable place right now is because of Microsoft's dominance. AMD saw an opportnity to advance themselves and thwart Intel's latest efforts because Intel took a chance at leaving x86.
The situation will not always be this way. Eventually someone will come up with a better idea and find a to implement it.
This has historically been the role that Steve Jobs has played. [/QUOTE]
Eventually, eventually, eventually. And when might that be? Five years from now? Ten?
Look, this is getting tiring. Apple is going to x86, and that's all there is to it.
People might not like that, but tough. MS isn't going anywhere anytime soon. Don't believe it. If ten years from now MS has 70% of the market, that will still be control.
Apple isn't listening to anyone here. They made their decision based on information that nobody here can claim to know. All we can do is to look at what they have done and try to disern, along with the help of the industry experts in these areas, why they have done it. That's what I try to do. I talk to my friends in Apple engineering as well. If Apple wanted to go to a high end 64 bit chip line that is out and working well, they could have gone to POWER, as some of the rumormongers were saying. But that would cost them $2,000 a chipset and uses almost 200 watts. Not going to happen. Same with IA-64. Same with every other 64 bit design - except for the x86.
Speculate all you want, but this is set in stone.
Originally posted by aegisdesign
Otherwise I fail to see any compelling reason to switch from PowerPC to Intel.
Intel's manufacturing prowess is the compelling reason, if not the only reason. It is the only thing that really matters.
Originally posted by melgross
Some people here are talking up the idea that Apple won't have a laptop for 12 - 18 months. That's nonsense. We know that they will have at least one out by June. No more than 9 months away. Yonah will be fine for starters.
12 months from when they announced the switch.
Quote:
Originally posted by melgross
Of course I mean the Xenon workstation ships.
XEON = Intel server CHIPS.
XENON = PowerPC chip in the Xbox 360.
Quote:
Originally posted by melgross
Those chips are only from 5%- 15% slower than an equivelent Opteron.
Problem is, there's no dual core Xeon. Dualcore Opterons have been out 5 months. The Opteron 280 and 880 at 2.4Ghz is due next month and the 180 single core at 2.4Ghz still beats the pants off the Xeon.
Intel are getting roasted in the server market now.
And they have massive shortages on some of their support chipsets at the moment.
The Intel switch isn't a bed of roses.
Quote:
Originally posted by melgross
It depends on what they are being used for really. Intels chips are optimized for different software than AMD's chips are. They work better for rendering, while AMD's are better for gaming. Half one, half the other. But both beat the G5 at this time.
I'm not going to argue benchmarks because you've already stated that you don't trust the barefeats benchs, which are the only ones around for cross platform benches.
I don't think you're right there - I've two dual Xeon 3.06Ghz, three dual Opteron 244s, a 242 and an iMac G5 so I couldn't give you ultimate benchmarks anyway but safe to say, I only buy Opteron 244s presently to use as web and database servers.
Quote:
Originally posted by melgross
Get over it? What is that supposed to mean? And it hasn't "always been the case". Only in the past couple of years has it "been the case".
Back as far as at least the 603 I remember Motorola expounding the virtues of it's chips in the embedded market. The IBM 750CX did quite well there too.
That's all I meant - chips get used all over the place so there's no shame in having a computer based on an embedded processor.
Quote:
Originally posted by melgross
Which chips has Apple gotten that Freescale not announced or listed? There was ony mythical chip that didn't exist. It had a "b" designation on the case. Everyone thought it was some *special* Apple only chip. It was only a test version that never got out. I already went over this several months ago, and proved it to those who were so hot about it.
Apple, for some time now, have been running chips faster than the speeds quoted by Moto/freescale. See the current PowerBooks v published 7447A specs for instance.
Quote:
Originally posted by melgross
There no rumors that Intel's interest in Apple have waned as IBM's did. Where did you read that one?
See original story - Apple seeking chips early, Intel saying no way.
Quote:
Originally posted by melgross
When Apple and IBM drew up their contract for G5 development and production, IBM knew full well how many chips Apple would be needing. Anyone who thinks differently is ignorent of the way business works. When I entered into negotions over part purchasing, we all put our cards on the table. You contract for numbers, you don't hedge, you don't fudge. IBM knew what their costs would be. There is no excuse. They were losing money on this because of their plant failures and the industrywide inability to predict the leakage current rise.
While Intel and AMD depend on fixing that problem as fast as they could, IBM didn't. With the plant failure to supply Apple on time, they had to hit a penalty phase in the contract. I can understand them giving up. But that doesn't make it any better.
I also think IBM thought they were on to a winner themselves for their own servers and workstations. It hit them also.
Supercomputers are specialized machines. Why are the Apple G5 machines so good for those tasks, yet fall 50% behind on a video render?
Quote:
Originally posted by melgross
Apple shouldn't have to pay IBM for development. They had a contract. That should be enough. If it wasn't, then IBM had no business agreeing to it in the first place. No excuse.
Sure but that doesn't help Apple. No development and they've no business either.
Quote:
Originally posted by melgross
IBM has always had too much internal politics. A little story.
When IBM still had 27% of the PC business, and they were teaming up with Apple on Pink, the software division said to the PC division that instead of installing Windows on their machins with an option of OS 2, they should do it the other way around, the PC people refused.
At the time, Windows had still to capture most of the market. The industry experts at the time were saying that the superiority of OS 2 would win out if IBM made it the default system, especially with an alliance of Apple. It didn't happen, and we are where we are today, possibly because of it.
If you were an IBM shop back then almost all the PS/2s came with OS/2 installed. IBM's real problem was that the PS/2 was a flop. They held on to the MCA architecture (I think only Olivetti licenced it) at a time when Compaq, HP, Elonex and then Dell were all firmly in the ISA and EISA camp with much cheaper computers. I remember paying nearly 10,000 GBP for a 486-25 upgrade for my Model 80. IBMs were just stupidly expensive.
In the consumer space they had the miserable PS/1. IBM just sucked.
It was immediately obvious from the very first Chicago betas we had as developers that Microsoft were going to win. OS/2 1.2 then needed twice the ram, a faster CPU and was plain ugly. 2.0 was much better but too late. Then NT 3.1 arrived and it was complete. If IBM weren't paying us to write compilers for them then we'd have switched to Windows entirely.
So, I don't think IBM had much say in the matter. They were soundly beaten by Microsoft no matter if their object oriented workplace shell was technically better or not. People wanted pretty and easy - not power.
Quote:
Originally posted by melgross
Remember that IBM is a marketing company. They are not a technology company. They seem to be one, but they are not. They will sell whatever the customer wants, even at the expense of their own home grown products.
That's a little unfair - they do a lot in the technology field before anyone else. But it's also true. IBM are a multi headed beast that more often than not, bites their own tail.
Originally posted by aegisdesign
12 months from when they announced the switch.
XEON = Intel server CHIPS.
XENON = PowerPC chip in the Xbox 360.
Problem is, there's no dual core Xeon. Dualcore Opterons have been out 5 months. The Opteron 280 and 880 at 2.4Ghz is due next month and the 180 single core at 2.4Ghz still beats the pants off the Xeon.
Intel are getting roasted in the server market now.
And they have massive shortages on some of their support chipsets at the moment.
The Intel switch isn't a bed of roses.
I'm not going to argue benchmarks because you've already stated that you don't trust the barefeats benchs, which are the only ones around for cross platform benches.
I don't think you're right there - I've two dual Xeon 3.06Ghz, three dual Opteron 244s, a 242 and an iMac G5 so I couldn't give you ultimate benchmarks anyway but safe to say, I only buy Opteron 244s presently to use as web and database servers.
Back as far as at least the 603 I remember Motorola expounding the virtues of it's chips in the embedded market. The IBM 750CX did quite well there too.
That's all I meant - chips get used all over the place so there's no shame in having a computer based on an embedded processor.
Apple, for some time now, have been running chips faster than the speeds quoted by Moto/freescale. See the current PowerBooks v published 7447A specs for instance.
See original story - Apple seeking chips early, Intel saying no way.
I also think IBM thought they were on to a winner themselves for their own servers and workstations. It hit them also.
Supercomputers are specialized machines. Why are the Apple G5 machines so good for those tasks, yet fall 50% behind on a video render?
Sure but that doesn't help Apple. No development and they've no business either.
If you were an IBM shop back then almost all the PS/2s came with OS/2 installed. IBM's real problem was that the PS/2 was a flop. They held on to the MCA architecture (I think only Olivetti licenced it) at a time when Compaq, HP, Elonex and then Dell were all firmly in the ISA and EISA camp with much cheaper computers. I remember paying nearly 10,000 GBP for a 486-25 upgrade for my Model 80. IBMs were just stupidly expensive.
In the consumer space they had the miserable PS/1. IBM just sucked.
It was immediately obvious from the very first Chicago betas we had as developers that Microsoft were going to win. OS/2 1.2 then needed twice the ram, a faster CPU and was plain ugly. 2.0 was much better but too late. Then NT 3.1 arrived and it was complete. If IBM weren't paying us to write compilers for them then we'd have switched to Windows entirely.
So, I don't think IBM had much say in the matter. They were soundly beaten by Microsoft no matter if their object oriented workplace shell was technically better or not. People wanted pretty and easy - not power.
That's a little unfair - they do a lot in the technology field before anyone else. But it's also true. IBM are a multi headed beast that more often than not, bites their own tail.
Whoa! These posts are getting way too long.
I'll only respond to a few points if that's alright with you.
People here are still talking 12-18 months, that's why I mentioned it.
Sorry, I see the confusion. XEON never comes off my fingertips. The spellcheck isn't much help either.
A dual core Xeon is no better than two single core Xeons, and that's where the comparisons are being made. Dual chip Macs vs. dual chip Xeons and Opterons. Real-world tests show the G%'s to be slower on average - up to 50% in some tests. I wish it weren't so, but it is. It's not all the fault of the chip memory bandwidth is also a problem. While the G5's have great theoretical bandwidth, the memory controller doesn't. This is a well known issue.
X86 still has almost 90% of the x86 server sales. You know, there is no reason to believe that AMD's fortunes will hold. Intel isn't standing still. Now that they're off the GHz kick, their designs are becoming more like IBM and AMD's. Things could easily reverse later in 2006 or early 2007. About the time Apple moves to those designs.
The shortage is in the older low end chip. It's going to be discontinued soon anyway.
About the chips, yes, embedded has always been an avenue that PPC has gone. Most PS Laser printers use PPC chips. No problem there. But Apple's chips were never those embedded designs.
Apple has er, "clocked" it's chips before. Remember the dual 1.42GHz machines? Those were 1.25GHz chips. You could see that on the case if you were brave enough to remove the heatsink.
That story about Apple seeking "early" production runs is from the inquirer. A site that I have regular run-ins with. They are not even close to being a neutral site.
They recently had a story about new graphics boards coming out that could run two 2560x1600 monitors. They said that these boards would be the first boards that could do that.
I pointed out that Apple had a board from Nvidia called the 6800Ultra. They said that yes, they knew but it was a Mac board. I then pointed out that, yes, it was, but that these were Mac monitors, as there are no other monitors that display 2560x1600. Needless to say, there was no point to continuing.
Disregard the article, most have.
The thing about IBM here is that they aren't willing to nurture a product. If IBM hadn't screwed up in the first place as did Motorola, and now Freescale, Apple might be selling 6 million computers this year instead of 4.5 million. Who knows, if the problems hadn't gone back to 1999 with Moto, Apple might now be selling 7 million units.
Look at MS, the junkyard dog that it is. It has lost over $4 billion dollars in the past 4 years itwh the XBox, and is expected to continue to lose money. But they want that market.
Look at the same thing with them and Palm. MS was badly beaten for years, but they persist. Now Palm is on the ropes. MS beat them. How much did it cost?
When Intel's chips were out of steam in the early '90's, they invested billions and got themselves out from under it.
If IBM were to look at the future possibilities and properly fund this, Apple's sales could be much higher. It's sad. I'd much rather see the G5 - 6 - 7, etc, than x86, but it ain't gomma happen. Too bad really.
With this switch Apple wanted to have the most common line of chips. In this way they will never be behind nor beyond. Apple want to highlight the best of it's qualities : design, stability, quality.
Originally posted by Powerdoc
I think that Apple is rather CPU agnostic (to a limited extent of course). If they choose Intel, it's because for laptops (half of the market) and because it's the number one.
With this switch Apple wanted to have the most common line of chips. In this way they will never be behind nor beyond. Apple want to highlight the best of it's qualities : design, stability, quality.
That's true. They need a vendor who will supply them what they need, when they need it. And if that vendor slips up, it will be for everyone, not just Apple.
Apple is going to x86, and that's all there is to it.
Hmmm. I still look at this as one scenario out of a couple.
Largely because the primary reasons most everyone gives for Apple switching to Intel solves one set of problems, but do not address the new set of problems Apple will have.
Apple will now have a more stable supplier than IBM or Freescale. Intel is in competition with AMD which continues to fuel Intel's need for development. If Intel slips in its projections the entire market slips. All true.
But that is just one part of it, that's not a holistic view of what makes a successful business. Turning to x86 Apple flirts with some very big risks.
Hypethetically Apple does go full x86-x64.
First of all this would reinforce Microsoft as the dominate standards leader of the PC industry. Microsoft is in the best position to chart the course of future x86 design.
If Windows Vista turns out to be on par with OS X or is even perceived to be good enough. This will leave less motivation for people to try OS X or buy a Macintosh. Apple being in hardware lock step will impede their ability to make radical hardware changes to differentiate themselves from all other OEM's. The same as it did in 1999 when Apple abandoned parallel ports and floppy drives. That was an extreme radical change that was rewarded in high volume sales of iMac's.
OS X on x86. There is a small but motivated group that vows to break any protection Apple will use to keep OS X only on Macintosh. This is a group who mostly builds their own computers and would not buy a Macintosh at all. The ultimate goal of this group is to force Apple into selling OS X as a stand alone product. It is very possible this group will successfully hack future versions of OS X, write drivers for it, and spread to any one whom ever wants it. When you look at the fact that much of the world has no problem using pirated copies of Windows, this could be a big problem for Apple.
x86 will mean a Macintosh will have the exact same innards as a Dell or HP Pavilion. Apple would be forced to lower its price and come in line with general PC pricing. Which would lower Apple's revenue margins and profit. And at this particular time when medium and small OEM's are slowly dying or being consolodated into the larger OEM's. There is no way Apple could win a price war with Dell. From a hardware standpoint there is no way Apple could battle Dell in price for performance. This would be exacerbated if Vista was deemed good enough or pirated copies of OS X were rampant and common. Dell would have the ability to lower its hardware prices and essentially wait as Apple dies.
Because of poor hardware sales or piracy Apple decides to release OS X as a product. This sceanrio would pit OS X head to head with Windows. There's no need for the bloody details we all know how that would go.
What does Intel gain from any of this? Apple leaving IBM for Intel is a social feather in Intel's cap. But from a practical or business standpoint what does Intel gain? It might be optimistic if Apple even added 2% to Intel's profits.
Intel would have given up one of its best and possibly last chances to advance technology that will not be supported by Microsoft or the larger PC industry.
x86 does nothing to reinforce Microsoft in anything. If anything it weakens their market confidence. How does Apple justify developers porting to the specific vaguaries of x86 and then tell Adobe "Oh btw rewrite that optimized code you just rewrote for our new architecture please. We just commited you to 150 man years of labour to hide our corporate plans, you can throw that code away."
Intel's opportunity to update the ISA was before Microsoft coded 64 bit windows. Oncce AMD beat them to the punch, the window closed. Apple is Not the same kind of window, because they do not represent nearly the same install base. Maybe 5 years from now there will be another window.
It's sad. I'd much rather see the G5 - 6 - 7, etc, than x86, but it ain't gomma happen.
Why is IBM the only chip maker who could manufacture exotic high performance chips for Apple.
Intel is actually in a much better position to experiment and take risks that IBM is able.
Hypethetically Apple uses a new Intel architecture.
In Intel, Apple would have a dedicated chip supplier. Intel has just as much at stake in this collaboration as Apple does. Intel is in direct competition with AMD and in many ways with Microsoft. This gives Intel motivation to produce and supply Apple with the chips it need.
Apple and Intel would compete against Microsoft and AMD with a better performing platform.
If Windows Vista were on par or percieved as good enough in comparison to OS X. Jobs can compare the performance of OS X/new Intel architecture to Vista and AMD64. Apple maintains the ability to change hardware specs as subtly or radically as it chooses.
OS X on a new Intel architecture would be impossible to hack and run on x86. Funny however Apple has not really shown much concern over this. That warning a month ago to websites was a slap on the wrist, but no legal action has been taken at all. Think about it. I'm sure all of the developer versions Apple sent out had serial numbers or some type of identification attatched. Would not take much effort for Apple to identify which developer the pirated OS X came from and prosecute them for breaking their NDA. Apple has done none of this and doesn't seem too concerned about it.
Intel would gain the ability to showcase its newest technologies that MS is not willing to adopt. If the performance and productivity of Macintosh, OS X, and the new Intel architecture exceed Microsoft and AMD. This would force both to change or adopt what Intel and Apple are doing.
I'm know Apple executives are mulling over all of these scenarios.
With all of the pro's and con's wieghd in they may choose x86. While this scenario would solve Apple's chip supply problem, it carries many other potential problems in the long run.
Apple could choose to work with a new Intel architecture. Which would have more risk in the beginning but has more potential to benefit the future requirements of both Apple and Intel.