[quote] Nope, AMD runs X86 code natively, no translation necessary. The core is not a RISC solution, not anymore so than the Pentium 4 anyway. <hr></blockquote>
Kind of... I'm referencing Hannibal's article at<a href="http://www.arstechnica.com/cpu/3q99/k7_theory/k7-two-1.html"; target="_blank">Ars Technica</a>. According to him, Athlon has both a hardware decoder that processes small x86 instrcutions, and a microcode decoder that actually is a form of hardware emulation for big x86 instructions. It all gets turned into "macro ops" that the (in his parlance) post-RISC core processes.
My question is more whether PPC instructions could be decoded into Athlon (or opteron) marco ops. The answer is probably no, and if it's possible it's probably not trivial.
Wow, there's a lot of bad information flying around in this thread!
- All of the Athlon/Operton and Pentium2-4 designs use a decoder which translates x86 instructions into internal "micro-ops" that are then fed to the internal RISC machine (very different machines internally).
- The IBM POWER4 does basically the same thing with the PowerPC instruction set.
- RISC has not "lost". In fact it pretty much won because all the CISC chips are now internally RISC or VLIW RISC. The PowerPC doesn't really fit the original meaning of RISC -- it has many instructions, some of which are quite powerful. Think of PPC's RISC more of as "reduced instruction set complexity", meaning that they redesigned it with newer design techniques in mind.
- The PowerPC camp is "behind" in terms of clock rate because their primary target is the embedded, not desktop, market.
- AMD and nVidia are in no way linked, aside from both being members of the HyperTransport consortium ... along with many others.
- nVidia is not going to build CPUs, and that's that. They are racing ahead building bigger and better GPUs and that may evolve into compute engines, but that is a very different thing that a CPU. A day may be coming where people don't care if the CPU is any faster, but do care that the GPU it controls is.
- I doubt AMD will be building PowerPCs in the foreseeable future, but its not completely impossible. If they were to do so they could probably apply quite a lot of their Operton's technology to building a PowerPC that would look quite a bit like IBM's POWER4.
- AMD doesn't "need" Apple. That would be a lot of design work in a different direction than their main focus, for a relatively small market (i.e. high end Macs -- the low-end & portable Macs would probably best be covered by Moto & IBM's existing designs since the Operton is way too high power).
- IBM is very likely the next builder of PowerPCs for Apple. IBM is fairly agnostic about what they build for their customers so I don't think Apple would have any problems getting them to add proprietary instructions, an AltiVec unit, or even HyperTransport links.
- IBM has some of the best fab technology on the planet, but the competition is stiff (Intel, AMD, NEC). I hesitate to put Moto in that list, as they seem to have fallen somewhat behind.
- Motorola's e500 core itself isn't particluarly modular, but the chip that it lives on is thanks to the OCEAN on-chip fabric. The distinction is important... the e500 core will not give Apple the performance it needs. If the rumours of Moto being on the out are false then a Moto G5 would be using a different core than the e500 that has been publicly discussed.
- The top-of-the-line nVidia boards aren't available on the Mac because of Apple, not nVidia. nVidia makes the chipsets and writes the drivers, it is up to the OEM to decide what boards to build and in what configurations. No, Apple isn't building the fastest possible nVidia-based boards. Yes, I think that is going to change soon.
- I firmly believe that Apple will not move to a x86 based hardware, but since I recently said why in another thread I'm not going to repeat it here.
<strong>Wow, there's a lot of bad information flying around in this thread!
- All of the Athlon/Operton and Pentium2-4 designs use a decoder which translates x86 instructions into internal "micro-ops" that are then fed to the internal RISC machine (very different machines internally).
- The IBM POWER4 does basically the same thing with the PowerPC instruction set.
- RISC has not "lost". In fact it pretty much won because all the CISC chips are now internally RISC or VLIW RISC. The PowerPC doesn't really fit the original meaning of RISC -- it has many instructions, some of which are quite powerful. Think of PPC's RISC more of as "reduced instruction set complexity", meaning that they redesigned it with newer design techniques in mind.
- The PowerPC camp is "behind" in terms of clock rate because their primary target is the embedded, not desktop, market.
- AMD and nVidia are in no way linked, aside from both being members of the HyperTransport consortium ... along with many others.
- nVidia is not going to build CPUs, and that's that. They are racing ahead building bigger and better GPUs and that may evolve into compute engines, but that is a very different thing that a CPU. A day may be coming where people don't care if the CPU is any faster, but do care that the GPU it controls is.
- I doubt AMD will be building PowerPCs in the foreseeable future, but its not completely impossible. If they were to do so they could probably apply quite a lot of their Operton's technology to building a PowerPC that would look quite a bit like IBM's POWER4.
- AMD doesn't "need" Apple. That would be a lot of design work in a different direction than their main focus, for a relatively small market (i.e. high end Macs -- the low-end & portable Macs would probably best be covered by Moto & IBM's existing designs since the Operton is way too high power).
- IBM is very likely the next builder of PowerPCs for Apple. IBM is fairly agnostic about what they build for their customers so I don't think Apple would have any problems getting them to add proprietary instructions, an AltiVec unit, or even HyperTransport links.
- IBM has some of the best fab technology on the planet, but the competition is stiff (Intel, AMD, NEC). I hesitate to put Moto in that list, as they seem to have fallen somewhat behind.
- Motorola's e500 core itself isn't particluarly modular, but the chip that it lives on is thanks to the OCEAN on-chip fabric. The distinction is important... the e500 core will not give Apple the performance it needs. If the rumours of Moto being on the out are false then a Moto G5 would be using a different core than the e500 that has been publicly discussed.
- The top-of-the-line nVidia boards aren't available on the Mac because of Apple, not nVidia. nVidia makes the chipsets and writes the drivers, it is up to the OEM to decide what boards to build and in what configurations. No, Apple isn't building the fastest possible nVidia-based boards. Yes, I think that is going to change soon.
- I firmly believe that Apple will not move to a x86 based hardware, but since I recently said why in another thread I'm not going to repeat it here.</strong><hr></blockquote>
goods points the programmer. (as usual )
The e500 core is not fit for the desktop market, he is no more powerful than the 7455 core. The fact that the new core designed by mot is not enough performant for Apple is the best argument to believe the rumor saying that IBM will be the next chip supplier of Apple.
The argument that IBM will not sell his most performant chips to preserve his market of high end servers, is not realist.
Apple have very little chances to become a huge actor in the domain of high end servers like IBM. Apple make only one server the Xserve, that have nothing to compare with the IBM products.
AMD can make a chip using Opteron technology , but i will cost them much more money than IBM to develop a new chip : IBM have already all the technology necessary : their only problems is Altivec (as they are going to license it, or make an Altivec compatible one ?).
AMD is not an expert of PPC chips, and have no clue on ativec either. IBM is more advanced in fab process than AMD.
So the next chip for high end Macs will be certainly made by IBM, or Mot is making a huge work to design a G5 not based upon the e500 core.
Why the praise for nVidia? All I see is that nVidia would become another MOTO in no time. They have stated that they want to "own" the mac market yet they have done nothing about it. They "own" the mac market now purely becuase their average cards are bundled with new macs. They have shown no sign of producing kick ass exlusive graphics hardware for macintosh to lure customers to their brand. If apple got into bed with nVidia we'd find that nVidia would then simply sit back (having the other 95% to concentrate on for "business reasons") and just paying lip service to apple customers. They would have no reason or incentive to produce their best technology purely for mac. Thus we'd find ourselves in another MOTO situation:
"Why can't nVidia make . . . I mean look at the competition, they get the nVidia Killer Force Pro 8 . . ."
I think an nVidia deal whilst maybe grabbing headlines for a year or so would end up in the same situation as we find ourselves in now.
<strong>Wow, there's a lot of bad information flying around in this thread!
- All of the Athlon/Operton and Pentium2-4 designs use a decoder which translates x86 instructions into internal "micro-ops" that are then fed to the internal RISC machine (very different machines internally).
- The IBM POWER4 does basically the same thing with the PowerPC instruction set.
- RISC has not "lost". In fact it pretty much won because all the CISC chips are now internally RISC or VLIW RISC. The PowerPC doesn't really fit the original meaning of RISC -- it has many instructions, some of which are quite powerful. Think of PPC's RISC more of as "reduced instruction set complexity", meaning that they redesigned it with newer design techniques in mind.
- The PowerPC camp is "behind" in terms of clock rate because their primary target is the embedded, not desktop, market.
- AMD and nVidia are in no way linked, aside from both being members of the HyperTransport consortium ... along with many others.
- nVidia is not going to build CPUs, and that's that. They are racing ahead building bigger and better GPUs and that may evolve into compute engines, but that is a very different thing that a CPU. A day may be coming where people don't care if the CPU is any faster, but do care that the GPU it controls is.
- I doubt AMD will be building PowerPCs in the foreseeable future, but its not completely impossible. If they were to do so they could probably apply quite a lot of their Operton's technology to building a PowerPC that would look quite a bit like IBM's POWER4.
- AMD doesn't "need" Apple. That would be a lot of design work in a different direction than their main focus, for a relatively small market (i.e. high end Macs -- the low-end & portable Macs would probably best be covered by Moto & IBM's existing designs since the Operton is way too high power).
- IBM is very likely the next builder of PowerPCs for Apple. IBM is fairly agnostic about what they build for their customers so I don't think Apple would have any problems getting them to add proprietary instructions, an AltiVec unit, or even HyperTransport links.
- IBM has some of the best fab technology on the planet, but the competition is stiff (Intel, AMD, NEC). I hesitate to put Moto in that list, as they seem to have fallen somewhat behind.
- Motorola's e500 core itself isn't particluarly modular, but the chip that it lives on is thanks to the OCEAN on-chip fabric. The distinction is important... the e500 core will not give Apple the performance it needs. If the rumours of Moto being on the out are false then a Moto G5 would be using a different core than the e500 that has been publicly discussed.
- The top-of-the-line nVidia boards aren't available on the Mac because of Apple, not nVidia. nVidia makes the chipsets and writes the drivers, it is up to the OEM to decide what boards to build and in what configurations. No, Apple isn't building the fastest possible nVidia-based boards. Yes, I think that is going to change soon.
- I firmly believe that Apple will not move to a x86 based hardware, but since I recently said why in another thread I'm not going to repeat it here.</strong><hr></blockquote>
I have to take issue with a few points.
The code is read natively, how it is processed is different from X86, but no "translation" is required. The code is simply broken into two streams.
In my points, I said that X86 had incorporated some RISC type features. This is not the same thing as becoming a RISC chip. RISC is more or less a dead issue because bruteforce CISC processing has made it so. This doesn't mean that RISC is inherently worse, it's just that with CPU cycles becoming a non issue on the CISC side, RISC advantages become negligible. This is mostly due to the failure to ramp up clock speed on the RISC side. If they had, this battle would be different.
Mark my words: nVidia WILL build a CPU in the future. Given the complexity of the current and upcoming 3D chips, a CPU would be a piece of cake. nVidia has said more or less the same thing.
AMD will not build PowerPC chips. They already have capacity issues. Taking a plant or part of a plant off-line for a year or two to produce a PPC chip just wouldn't make sense, especially when they have a unit coming out that has many more advantages (the Hammer line).
I disagree with you. AMD does need Apple. They have a heckuva chip coming out with nobody willing to publicly say they will support it. All the major OEMs are afraid of intel. Apple has no such fear. Furthermore, AMD would have very little added development since the chip has already been designed. Apple would have to port OS X to the new platform and then work with AMD to design a motherboard or logic chip that would make the motherboard proprietary. That would not take as long as converting a chip plant.
Top of the Line Nvidia boards ARE available for Apple. Check their store, they have a GeForce 4 ti 4600 for sale now.
I don't beleive Apple will move to X86 for philosophical reasons (because of intel). I don't think they will have the same problem with X86-64 since this was developed by AMD. Unless there is some real development really soon, Apple may have no choice.
AMD doesn't need Apple because the potential market inside of Apple's product line for an Opteron based processor is the PowerMac line exclusively. That 5% they'd be after is in reality maybe 0.5%. Too much trouble for both parties. No way is AMD making the G5, it just ain't happening.
Mot ain't happening either because they're on life support anyhow so Apple would be stupid to put all their eggs in that basket. Anyway, Mot's not interested in making a high end CPU exclusively for Apple, hell they won't even put RIO on the G4. At least that would make it a decent idea for Apple to start offering 4 and 8 way G4 PowerMacs that would be pretty damn powerful, but as long as the G4 is on the MPX bus, that ain't happening becasue there's no bandwidth. Theres a chance that Apple could convince Mot to design a G4 with RIO capability, but the chances of that are pretty slim. Not impossible though, and a Powermac with quad 1.2Ghz G4s, 333mhz FSB, and 333mhz DDRAM would go a long way to closing the performance gap.
So it's IBM from the looks of it. They're huge, not going out of business, and have the most powerful chip in the world in their repetiore. I honestly don't know how likely it is that Apple has convinced IBM to make a cooler version of the Power4, it all depends when Apple came to IBM asking for a next gen chip design. If Apple came to IBM when the G4 fiasco started 3 years ago then the Power4 would have still been in development and it would have been easier to make a cooler, less power hungry version possible. If that's the case, then that chip should be ready within the next year. But who the heck knows, not us that's for sure. Whatever the outcome is, it's most likely IBM will be behind it, other than that we have no clue and can only speculate.
PS. What kind of bus does IBM use for the Power4 and other high end processors? How does it compare to RIO or Hypertransport?
<strong>Oh, there may be a third obstacle. Will Steve ever get real and realize that Motorola is a dead end?
</strong><hr></blockquote>
Are you serious? Or was that supposed to be ironic?
After the 500->450 debacle, and the subsequent 500Mhz stall, I'd bet dollars to donuts SJ would have gotten out of that dead end THEN, let alone now. And remember, if this switch is going to happen at all, the decision was made 12-18 months ago, at least - when the Great Stall was still a fresh and infuriating memory.
If there's one thing SJ cannot stand, it's incompetence.
This excerpt from the Wired piece on Nvidia amused me greatly:
(about Huang)
But he has little tolerance for screwups. In one legendary meeting, he's said to have ripped into a project team for its tendency to repeat mistakes. "Do you suck?" he asked the stunned employees. "Because if you suck, just get up and say you suck."
Are you serious? Or was that supposed to be ironic?
After the 500->450 debacle, and the subsequent 500Mhz stall, I'd bet dollars to donuts SJ would have gotten out of that dead end THEN, let alone now. And remember, if this switch is going to happen at all, the decision was made 12-18 months ago, at least - when the Great Stall was still a fresh and infuriating memory.
If there's one thing SJ cannot stand, it's incompetence.
This excerpt from the Wired piece on Nvidia amused me greatly:
(about Huang)
But he has little tolerance for screwups. In one legendary meeting, he's said to have ripped into a project team for its tendency to repeat mistakes. "Do you suck?" he asked the stunned employees. "Because if you suck, just get up and say you suck."
Ok, just to be the odd man out, and for the sae of picking the underdog, I'm going to bet on Motorola being the source for the G5. They'll make the first batch of G5s, and possibly later they'll contract out some of the production to IBM.
Motorola has done the most work on the G5. It's fine and dandy to talk about how other companies could really build a killer G5, but the reality is that Moto's the one that's actually been working on it.
Also, you say that AMD needs Apple. This may be true, but Motorola needs Apple a lot more. IBM is big, and reliable, yes, but they're also in the same category as AMD and NVidia. They have so much else on their plate right now that they just don't care that much about the Mac market.
Motorola cares about the Mac. Motorola will be the source for the G5.
And if I'm wrong, well, then I'm wrong. But if I'm right, I'm going to get to watch a lot of people eat their hats. :-)
If you look at the current state of Apple hardware, it's apparent that Motorola doesn't give a dam about Apple's needs.
If Motorola won't build Apple a competitive desktop CPU, then Apple needs to take its business elsewhere.
One thing is certain however, the G4's successor will be a PowerPC chip. PowerPC has a long life ahead of it while x86 is on its last legs.
I think IBM would build Apple a custom chip, as long as Apple helped finance the project. IBM did Gecko for Nintendo, so I'm sure they'd be happy to work with Apple as well.
<strong>If you look at the current state of Apple hardware, it's apparent that Motorola doesn't give a dam about Apple's needs.
If Motorola won't build Apple a competitive desktop CPU, then Apple needs to take its business elsewhere.
One thing is certain however, the G4's successor will be a PowerPC chip. PowerPC has a long life ahead of it while x86 is on its last legs.
I think IBM would build Apple a custom chip, as long as Apple helped finance the project. IBM did Gecko for Nintendo, so I'm sure they'd be happy to work with Apple as well.</strong><hr></blockquote>
Um, no.
X86 is making a transition to 64 bit, that's hardly being "on its last legs". In fact, both intel and AMD are transitioning to a .13 micron fab process. There are plans to go to a .09 micron process within a year. Motorola isn't there yet. CPU clock cycles on X86 will be over 3 times what is available on the G platform within about 2 months. Say what you will about processor efficiency, but AMD's Athlon line is only slightly less efficient than the G series. Both AMD and Motorola processors are much more efficient than intels processors.
If anything, the PPC is on its last legs in the desktop environment, but only because of the neglect of Motorola. Had they spent the time and money on it, it would still be a contender.
IBM may be to big to care about Apple that much but they're goals are the same as Apple's as far as processor performance. Power4 is awesome, Power5 is gonna be awesome. The question is whether Apple/IBM can find a way to somehow get those processors or a derivative of them running in a Powermac. I haven't really heard a definitive argument on the chances of this happening.
I would tend to think that the next Apple chip will be a derivative on the Power5. This depends on the existence of a RIO enabled G4 though for Apple to put in quad and octo processor systems in the meantime.
I'll ask again, does anyone know what kind of bus IBM uses for the Power4 and how it compares to RIO and Hypertransport?
Will it be custom (like Gekko) or a Power4 derivative? I don't know. I'm working on finding out.
Moto will have no part. Why? Steve despises Moto for loving the embedded market so much and not seeing and buying into his vision. Moto will source G4s for the next gen iMac, eMac, and iBook (and Powerbook for a while).
Apple will indeed never use an x86 chip, nor a PPC chip from AMD, because AMD will not build one.
I agree that Nvidia will never build a CPU, but that the focus in computing will become the GPU, and that the CPU will be a side-mention.
BTW, Moto's tech is called BookE, not eBook.
RISC isn't dead, and neither is CISC. But both are changing. RISC-based chips like PPC have taken on some CISC-like properties, and CISC architectures like x86 have taken a lot of RISC-like properties.
Apple and nVidia have a good relationship now. Apple does have the highest nVidia boards, and nVidia is in every iMac and eMac sold. Added to that that indeed nVidia has said they are working on some great things with Apple...
But for the love of Steve, AMD doesn't need Apple, and vice versa.
Not to point fingers, but it seems that Thresher is very much defending the x86 architecture and AMD in general...and while I certainly welcome all cross-platform people, Thresher....I think that defending something like that in here is kinda a hopeless battle...kinda like selling Bill Gates on a new iMac
EDIT: Oh, and Thresher, the last thing you want to do in here is trash the PPC saying its on it last legs. More of that and you'll unleash the monster inside of me.
The PPC is nice and healthy, thanks to IBM. I spent this morning reading up on the 750FX chip from IBM, which despite being a G3, is a very nice chip, and should serve the iBook well, since it scales to 1 Ghz...
Just wait and see what the next IBM PPC chip looks like
[ 07-07-2002: Message edited by: The All Knowing 1 ]</p>
[quote]I disagree with you. AMD does need Apple. They have a heckuva chip coming out with nobody willing to publicly say they will support it. All the major OEMs are afraid of intel.<hr></blockquote>
You living on Planet B*llsh*t or what. MS has already committed to 64-bit versions of desktop and server OS in support of the Opteron. Linux port by SuSe is in progress.
On the hardware side, Dell has not committed to the Itanium 2, and major component mfgs have already announced support for the Opteron--nVidia, ATI, Matrox, 3D Labs, VIA, ASUS etc.
Afraid of Intel--nope--where MS goes everyone else falls in line.
ps. IDC announced a couple of weeks back that "white box" mfgs accounted for over 58% of PC sales. If the Opteron lives up to spec, AMD could live without a major mfg and still sell lots of chips. Though given the Itanium debacle and Intel's pricing of Xeon chips, at least one major will commit to the Opteron.
Settle down folks, everyone is entitled to their own opion, no matter how wrong they may be.
I am new to the platform, I just bought my first Powerbook last month. I've been on PCs for about 12 years. My Dell Inspiron went on eBay today.
I think there is a serious lack of urgency about the development of the G series CPUs. This is going to be a real problem if it is not rectified soon. The newest X86 processors ARE faster than what we have available on the Mac platform. Ignoring it or going into denial is not going to solve the problem.
For those of you who think I'm on some sort of covert mission to evangalize the x86, you couldn't be more wrong. I could care less who wins, as long as OS X and Apple succeed. I know Windows inside and out, which is the problem, you shouldn't HAVE to know all the wierd crap in an OS to just use it. If it will help my credentials, I am moderator of the Mac forum at <a href="http://www.tech-report.com."; target="_blank">www.tech-report.com.</a> Not an easy job considering the site caters to the PC crowd.
Personally, I do think (my opinion folks), that the Hammer and OS X would be a match made in heaven. It helps both parties. AMD gets a mainstream OEM that isn't afraid of intel and expands their potential market, Apple gets a really fast CPU with a lot of headroom for further improvements and a hardcore group of enthusiasts.
As for living in la-la land about acceptance of the Hammer, when you see Dell stick one in their top of the line Dimension, then you'll know that AMD has gotten acceptance. HP/Compaq, Micron, and Sony all put Athlons in their bottom of the line equipment despite the fact that Athlons are generally faster CPUs and offer a superior price/peformance ratio. They don't really advertise them (where's the Genuine Athlon sticker?) and Gateway can't make up it's mind whether or not to sell them at all. Dell avoids them like the plague.
No doubt MS is doing a good thing for AMD by supporting the Hammer and that will go a long way to supporting the platform, but until major OEMs (not just the whitebox and enthusiast markets) start using Athlons, AMD will remain a bit player.
It is amazing to me to watch both sides of the X86 versus Mac debate. I see a lot of irrational discussions on both sides. Neither side seems willing to believe that there may be some merits to the other side's arguments. We are right, the PC is fundamentally broken because they are hindered by Windows. They are right that we are hindered by the performance of our hardware. Which is worse?
Personally, I just use my PCs for gaming now. I use the Powerbook for everything else. This kind of proves my point. PCs perform better (when they aren't crashing) and Macs just plain work, even if they aren't speed demons.
You guys should check out the info I've been gathering over the past year. I posted it in a new forum discussion titled: "G5 Speculation Revisited" on the main forum page. There is some interesting information there.
Comments
Kind of... I'm referencing Hannibal's article at<a href="http://www.arstechnica.com/cpu/3q99/k7_theory/k7-two-1.html"; target="_blank">Ars Technica</a>. According to him, Athlon has both a hardware decoder that processes small x86 instrcutions, and a microcode decoder that actually is a form of hardware emulation for big x86 instructions. It all gets turned into "macro ops" that the (in his parlance) post-RISC core processes.
My question is more whether PPC instructions could be decoded into Athlon (or opteron) marco ops. The answer is probably no, and if it's possible it's probably not trivial.
- All of the Athlon/Operton and Pentium2-4 designs use a decoder which translates x86 instructions into internal "micro-ops" that are then fed to the internal RISC machine (very different machines internally).
- The IBM POWER4 does basically the same thing with the PowerPC instruction set.
- RISC has not "lost". In fact it pretty much won because all the CISC chips are now internally RISC or VLIW RISC. The PowerPC doesn't really fit the original meaning of RISC -- it has many instructions, some of which are quite powerful. Think of PPC's RISC more of as "reduced instruction set complexity", meaning that they redesigned it with newer design techniques in mind.
- The PowerPC camp is "behind" in terms of clock rate because their primary target is the embedded, not desktop, market.
- AMD and nVidia are in no way linked, aside from both being members of the HyperTransport consortium ... along with many others.
- nVidia is not going to build CPUs, and that's that. They are racing ahead building bigger and better GPUs and that may evolve into compute engines, but that is a very different thing that a CPU. A day may be coming where people don't care if the CPU is any faster, but do care that the GPU it controls is.
- I doubt AMD will be building PowerPCs in the foreseeable future, but its not completely impossible. If they were to do so they could probably apply quite a lot of their Operton's technology to building a PowerPC that would look quite a bit like IBM's POWER4.
- AMD doesn't "need" Apple. That would be a lot of design work in a different direction than their main focus, for a relatively small market (i.e. high end Macs -- the low-end & portable Macs would probably best be covered by Moto & IBM's existing designs since the Operton is way too high power).
- IBM is very likely the next builder of PowerPCs for Apple. IBM is fairly agnostic about what they build for their customers so I don't think Apple would have any problems getting them to add proprietary instructions, an AltiVec unit, or even HyperTransport links.
- IBM has some of the best fab technology on the planet, but the competition is stiff (Intel, AMD, NEC). I hesitate to put Moto in that list, as they seem to have fallen somewhat behind.
- Motorola's e500 core itself isn't particluarly modular, but the chip that it lives on is thanks to the OCEAN on-chip fabric. The distinction is important... the e500 core will not give Apple the performance it needs. If the rumours of Moto being on the out are false then a Moto G5 would be using a different core than the e500 that has been publicly discussed.
- The top-of-the-line nVidia boards aren't available on the Mac because of Apple, not nVidia. nVidia makes the chipsets and writes the drivers, it is up to the OEM to decide what boards to build and in what configurations. No, Apple isn't building the fastest possible nVidia-based boards. Yes, I think that is going to change soon.
- I firmly believe that Apple will not move to a x86 based hardware, but since I recently said why in another thread I'm not going to repeat it here.
<strong>Wow, there's a lot of bad information flying around in this thread!
- All of the Athlon/Operton and Pentium2-4 designs use a decoder which translates x86 instructions into internal "micro-ops" that are then fed to the internal RISC machine (very different machines internally).
- The IBM POWER4 does basically the same thing with the PowerPC instruction set.
- RISC has not "lost". In fact it pretty much won because all the CISC chips are now internally RISC or VLIW RISC. The PowerPC doesn't really fit the original meaning of RISC -- it has many instructions, some of which are quite powerful. Think of PPC's RISC more of as "reduced instruction set complexity", meaning that they redesigned it with newer design techniques in mind.
- The PowerPC camp is "behind" in terms of clock rate because their primary target is the embedded, not desktop, market.
- AMD and nVidia are in no way linked, aside from both being members of the HyperTransport consortium ... along with many others.
- nVidia is not going to build CPUs, and that's that. They are racing ahead building bigger and better GPUs and that may evolve into compute engines, but that is a very different thing that a CPU. A day may be coming where people don't care if the CPU is any faster, but do care that the GPU it controls is.
- I doubt AMD will be building PowerPCs in the foreseeable future, but its not completely impossible. If they were to do so they could probably apply quite a lot of their Operton's technology to building a PowerPC that would look quite a bit like IBM's POWER4.
- AMD doesn't "need" Apple. That would be a lot of design work in a different direction than their main focus, for a relatively small market (i.e. high end Macs -- the low-end & portable Macs would probably best be covered by Moto & IBM's existing designs since the Operton is way too high power).
- IBM is very likely the next builder of PowerPCs for Apple. IBM is fairly agnostic about what they build for their customers so I don't think Apple would have any problems getting them to add proprietary instructions, an AltiVec unit, or even HyperTransport links.
- IBM has some of the best fab technology on the planet, but the competition is stiff (Intel, AMD, NEC). I hesitate to put Moto in that list, as they seem to have fallen somewhat behind.
- Motorola's e500 core itself isn't particluarly modular, but the chip that it lives on is thanks to the OCEAN on-chip fabric. The distinction is important... the e500 core will not give Apple the performance it needs. If the rumours of Moto being on the out are false then a Moto G5 would be using a different core than the e500 that has been publicly discussed.
- The top-of-the-line nVidia boards aren't available on the Mac because of Apple, not nVidia. nVidia makes the chipsets and writes the drivers, it is up to the OEM to decide what boards to build and in what configurations. No, Apple isn't building the fastest possible nVidia-based boards. Yes, I think that is going to change soon.
- I firmly believe that Apple will not move to a x86 based hardware, but since I recently said why in another thread I'm not going to repeat it here.</strong><hr></blockquote>
goods points the programmer. (as usual
The e500 core is not fit for the desktop market, he is no more powerful than the 7455 core. The fact that the new core designed by mot is not enough performant for Apple is the best argument to believe the rumor saying that IBM will be the next chip supplier of Apple.
The argument that IBM will not sell his most performant chips to preserve his market of high end servers, is not realist.
Apple have very little chances to become a huge actor in the domain of high end servers like IBM. Apple make only one server the Xserve, that have nothing to compare with the IBM products.
AMD can make a chip using Opteron technology , but i will cost them much more money than IBM to develop a new chip : IBM have already all the technology necessary : their only problems is Altivec (as they are going to license it, or make an Altivec compatible one ?).
AMD is not an expert of PPC chips, and have no clue on ativec either. IBM is more advanced in fab process than AMD.
So the next chip for high end Macs will be certainly made by IBM, or Mot is making a huge work to design a G5 not based upon the e500 core.
"Why can't nVidia make . . . I mean look at the competition, they get the nVidia Killer Force Pro 8 . . ."
I think an nVidia deal whilst maybe grabbing headlines for a year or so would end up in the same situation as we find ourselves in now.
<strong>Wow, there's a lot of bad information flying around in this thread!
- All of the Athlon/Operton and Pentium2-4 designs use a decoder which translates x86 instructions into internal "micro-ops" that are then fed to the internal RISC machine (very different machines internally).
- The IBM POWER4 does basically the same thing with the PowerPC instruction set.
- RISC has not "lost". In fact it pretty much won because all the CISC chips are now internally RISC or VLIW RISC. The PowerPC doesn't really fit the original meaning of RISC -- it has many instructions, some of which are quite powerful. Think of PPC's RISC more of as "reduced instruction set complexity", meaning that they redesigned it with newer design techniques in mind.
- The PowerPC camp is "behind" in terms of clock rate because their primary target is the embedded, not desktop, market.
- AMD and nVidia are in no way linked, aside from both being members of the HyperTransport consortium ... along with many others.
- nVidia is not going to build CPUs, and that's that. They are racing ahead building bigger and better GPUs and that may evolve into compute engines, but that is a very different thing that a CPU. A day may be coming where people don't care if the CPU is any faster, but do care that the GPU it controls is.
- I doubt AMD will be building PowerPCs in the foreseeable future, but its not completely impossible. If they were to do so they could probably apply quite a lot of their Operton's technology to building a PowerPC that would look quite a bit like IBM's POWER4.
- AMD doesn't "need" Apple. That would be a lot of design work in a different direction than their main focus, for a relatively small market (i.e. high end Macs -- the low-end & portable Macs would probably best be covered by Moto & IBM's existing designs since the Operton is way too high power).
- IBM is very likely the next builder of PowerPCs for Apple. IBM is fairly agnostic about what they build for their customers so I don't think Apple would have any problems getting them to add proprietary instructions, an AltiVec unit, or even HyperTransport links.
- IBM has some of the best fab technology on the planet, but the competition is stiff (Intel, AMD, NEC). I hesitate to put Moto in that list, as they seem to have fallen somewhat behind.
- Motorola's e500 core itself isn't particluarly modular, but the chip that it lives on is thanks to the OCEAN on-chip fabric. The distinction is important... the e500 core will not give Apple the performance it needs. If the rumours of Moto being on the out are false then a Moto G5 would be using a different core than the e500 that has been publicly discussed.
- The top-of-the-line nVidia boards aren't available on the Mac because of Apple, not nVidia. nVidia makes the chipsets and writes the drivers, it is up to the OEM to decide what boards to build and in what configurations. No, Apple isn't building the fastest possible nVidia-based boards. Yes, I think that is going to change soon.
- I firmly believe that Apple will not move to a x86 based hardware, but since I recently said why in another thread I'm not going to repeat it here.</strong><hr></blockquote>
I have to take issue with a few points.
The code is read natively, how it is processed is different from X86, but no "translation" is required. The code is simply broken into two streams.
In my points, I said that X86 had incorporated some RISC type features. This is not the same thing as becoming a RISC chip. RISC is more or less a dead issue because bruteforce CISC processing has made it so. This doesn't mean that RISC is inherently worse, it's just that with CPU cycles becoming a non issue on the CISC side, RISC advantages become negligible. This is mostly due to the failure to ramp up clock speed on the RISC side. If they had, this battle would be different.
Mark my words: nVidia WILL build a CPU in the future. Given the complexity of the current and upcoming 3D chips, a CPU would be a piece of cake. nVidia has said more or less the same thing.
AMD will not build PowerPC chips. They already have capacity issues. Taking a plant or part of a plant off-line for a year or two to produce a PPC chip just wouldn't make sense, especially when they have a unit coming out that has many more advantages (the Hammer line).
I disagree with you. AMD does need Apple. They have a heckuva chip coming out with nobody willing to publicly say they will support it. All the major OEMs are afraid of intel. Apple has no such fear. Furthermore, AMD would have very little added development since the chip has already been designed. Apple would have to port OS X to the new platform and then work with AMD to design a motherboard or logic chip that would make the motherboard proprietary. That would not take as long as converting a chip plant.
Top of the Line Nvidia boards ARE available for Apple. Check their store, they have a GeForce 4 ti 4600 for sale now.
I don't beleive Apple will move to X86 for philosophical reasons (because of intel). I don't think they will have the same problem with X86-64 since this was developed by AMD. Unless there is some real development really soon, Apple may have no choice.
Mot ain't happening either because they're on life support anyhow so Apple would be stupid to put all their eggs in that basket. Anyway, Mot's not interested in making a high end CPU exclusively for Apple, hell they won't even put RIO on the G4. At least that would make it a decent idea for Apple to start offering 4 and 8 way G4 PowerMacs that would be pretty damn powerful, but as long as the G4 is on the MPX bus, that ain't happening becasue there's no bandwidth. Theres a chance that Apple could convince Mot to design a G4 with RIO capability, but the chances of that are pretty slim. Not impossible though, and a Powermac with quad 1.2Ghz G4s, 333mhz FSB, and 333mhz DDRAM would go a long way to closing the performance gap.
So it's IBM from the looks of it. They're huge, not going out of business, and have the most powerful chip in the world in their repetiore. I honestly don't know how likely it is that Apple has convinced IBM to make a cooler version of the Power4, it all depends when Apple came to IBM asking for a next gen chip design. If Apple came to IBM when the G4 fiasco started 3 years ago then the Power4 would have still been in development and it would have been easier to make a cooler, less power hungry version possible. If that's the case, then that chip should be ready within the next year. But who the heck knows, not us that's for sure. Whatever the outcome is, it's most likely IBM will be behind it, other than that we have no clue and can only speculate.
PS. What kind of bus does IBM use for the Power4 and other high end processors? How does it compare to RIO or Hypertransport?
<strong>AMD and Apple need each other.
AMD has ...
(...)
Apple needs horsepower desperately.
Oh, there may be a third obstacle. Will Steve ever get real and realize that Motorola is a dead end?
[ 07-05-2002: Message edited by: Thresher ]</strong><hr></blockquote>
Good point at all.
<strong>Oh, there may be a third obstacle. Will Steve ever get real and realize that Motorola is a dead end?
</strong><hr></blockquote>
Are you serious? Or was that supposed to be ironic?
After the 500->450 debacle, and the subsequent 500Mhz stall, I'd bet dollars to donuts SJ would have gotten out of that dead end THEN, let alone now. And remember, if this switch is going to happen at all, the decision was made 12-18 months ago, at least - when the Great Stall was still a fresh and infuriating memory.
If there's one thing SJ cannot stand, it's incompetence.
This excerpt from the Wired piece on Nvidia amused me greatly:
(about Huang)
But he has little tolerance for screwups. In one legendary meeting, he's said to have ripped into a project team for its tendency to repeat mistakes. "Do you suck?" he asked the stunned employees. "Because if you suck, just get up and say you suck."
(end quote)
Sound like anyone you know?
[ 07-06-2002: Message edited by: arbitrary ]</p>
<strong>
Are you serious? Or was that supposed to be ironic?
After the 500->450 debacle, and the subsequent 500Mhz stall, I'd bet dollars to donuts SJ would have gotten out of that dead end THEN, let alone now. And remember, if this switch is going to happen at all, the decision was made 12-18 months ago, at least - when the Great Stall was still a fresh and infuriating memory.
If there's one thing SJ cannot stand, it's incompetence.
This excerpt from the Wired piece on Nvidia amused me greatly:
(about Huang)
But he has little tolerance for screwups. In one legendary meeting, he's said to have ripped into a project team for its tendency to repeat mistakes. "Do you suck?" he asked the stunned employees. "Because if you suck, just get up and say you suck."
(end quote)
Sound like anyone you know?
[ 07-06-2002: Message edited by: arbitrary ]</strong><hr></blockquote>
I don't know, Jobs can be pretty obstinate. I think he has better options now though.
Motorola has done the most work on the G5. It's fine and dandy to talk about how other companies could really build a killer G5, but the reality is that Moto's the one that's actually been working on it.
Also, you say that AMD needs Apple. This may be true, but Motorola needs Apple a lot more. IBM is big, and reliable, yes, but they're also in the same category as AMD and NVidia. They have so much else on their plate right now that they just don't care that much about the Mac market.
Motorola cares about the Mac. Motorola will be the source for the G5.
And if I'm wrong, well, then I'm wrong. But if I'm right, I'm going to get to watch a lot of people eat their hats. :-)
If Motorola won't build Apple a competitive desktop CPU, then Apple needs to take its business elsewhere.
One thing is certain however, the G4's successor will be a PowerPC chip. PowerPC has a long life ahead of it while x86 is on its last legs.
I think IBM would build Apple a custom chip, as long as Apple helped finance the project. IBM did Gecko for Nintendo, so I'm sure they'd be happy to work with Apple as well.
<strong>If you look at the current state of Apple hardware, it's apparent that Motorola doesn't give a dam about Apple's needs.
If Motorola won't build Apple a competitive desktop CPU, then Apple needs to take its business elsewhere.
One thing is certain however, the G4's successor will be a PowerPC chip. PowerPC has a long life ahead of it while x86 is on its last legs.
I think IBM would build Apple a custom chip, as long as Apple helped finance the project. IBM did Gecko for Nintendo, so I'm sure they'd be happy to work with Apple as well.</strong><hr></blockquote>
Um, no.
X86 is making a transition to 64 bit, that's hardly being "on its last legs". In fact, both intel and AMD are transitioning to a .13 micron fab process. There are plans to go to a .09 micron process within a year. Motorola isn't there yet. CPU clock cycles on X86 will be over 3 times what is available on the G platform within about 2 months. Say what you will about processor efficiency, but AMD's Athlon line is only slightly less efficient than the G series. Both AMD and Motorola processors are much more efficient than intels processors.
If anything, the PPC is on its last legs in the desktop environment, but only because of the neglect of Motorola. Had they spent the time and money on it, it would still be a contender.
I would tend to think that the next Apple chip will be a derivative on the Power5. This depends on the existence of a RIO enabled G4 though for Apple to put in quad and octo processor systems in the meantime.
I'll ask again, does anyone know what kind of bus IBM uses for the Power4 and how it compares to RIO and Hypertransport?
IBM will source the G5.
Will it be custom (like Gekko) or a Power4 derivative? I don't know. I'm working on finding out.
Moto will have no part. Why? Steve despises Moto for loving the embedded market so much and not seeing and buying into his vision. Moto will source G4s for the next gen iMac, eMac, and iBook (and Powerbook for a while).
Apple will indeed never use an x86 chip, nor a PPC chip from AMD, because AMD will not build one.
I agree that Nvidia will never build a CPU, but that the focus in computing will become the GPU, and that the CPU will be a side-mention.
BTW, Moto's tech is called BookE, not eBook.
RISC isn't dead, and neither is CISC. But both are changing. RISC-based chips like PPC have taken on some CISC-like properties, and CISC architectures like x86 have taken a lot of RISC-like properties.
Apple and nVidia have a good relationship now. Apple does have the highest nVidia boards, and nVidia is in every iMac and eMac sold. Added to that that indeed nVidia has said they are working on some great things with Apple...
But for the love of Steve, AMD doesn't need Apple, and vice versa.
Not to point fingers, but it seems that Thresher is very much defending the x86 architecture and AMD in general...and while I certainly welcome all cross-platform people, Thresher....I think that defending something like that in here is kinda a hopeless battle...kinda like selling Bill Gates on a new iMac
EDIT: Oh, and Thresher, the last thing you want to do in here is trash the PPC saying its on it last legs. More of that and you'll unleash the monster inside of me.
The PPC is nice and healthy, thanks to IBM. I spent this morning reading up on the 750FX chip from IBM, which despite being a G3, is a very nice chip, and should serve the iBook well, since it scales to 1 Ghz...
Just wait and see what the next IBM PPC chip looks like
[ 07-07-2002: Message edited by: The All Knowing 1 ]</p>
You living on Planet B*llsh*t or what. MS has already committed to 64-bit versions of desktop and server OS in support of the Opteron. Linux port by SuSe is in progress.
On the hardware side, Dell has not committed to the Itanium 2, and major component mfgs have already announced support for the Opteron--nVidia, ATI, Matrox, 3D Labs, VIA, ASUS etc.
Afraid of Intel--nope--where MS goes everyone else falls in line.
ps. IDC announced a couple of weeks back that "white box" mfgs accounted for over 58% of PC sales. If the Opteron lives up to spec, AMD could live without a major mfg and still sell lots of chips. Though given the Itanium debacle and Intel's pricing of Xeon chips, at least one major will commit to the Opteron.
[ 07-07-2002: Message edited by: cowerd ]</p>
I am new to the platform, I just bought my first Powerbook last month. I've been on PCs for about 12 years. My Dell Inspiron went on eBay today.
I think there is a serious lack of urgency about the development of the G series CPUs. This is going to be a real problem if it is not rectified soon. The newest X86 processors ARE faster than what we have available on the Mac platform. Ignoring it or going into denial is not going to solve the problem.
For those of you who think I'm on some sort of covert mission to evangalize the x86, you couldn't be more wrong. I could care less who wins, as long as OS X and Apple succeed. I know Windows inside and out, which is the problem, you shouldn't HAVE to know all the wierd crap in an OS to just use it. If it will help my credentials, I am moderator of the Mac forum at <a href="http://www.tech-report.com."; target="_blank">www.tech-report.com.</a> Not an easy job considering the site caters to the PC crowd.
Personally, I do think (my opinion folks), that the Hammer and OS X would be a match made in heaven. It helps both parties. AMD gets a mainstream OEM that isn't afraid of intel and expands their potential market, Apple gets a really fast CPU with a lot of headroom for further improvements and a hardcore group of enthusiasts.
As for living in la-la land about acceptance of the Hammer, when you see Dell stick one in their top of the line Dimension, then you'll know that AMD has gotten acceptance. HP/Compaq, Micron, and Sony all put Athlons in their bottom of the line equipment despite the fact that Athlons are generally faster CPUs and offer a superior price/peformance ratio. They don't really advertise them (where's the Genuine Athlon sticker?) and Gateway can't make up it's mind whether or not to sell them at all. Dell avoids them like the plague.
No doubt MS is doing a good thing for AMD by supporting the Hammer and that will go a long way to supporting the platform, but until major OEMs (not just the whitebox and enthusiast markets) start using Athlons, AMD will remain a bit player.
It is amazing to me to watch both sides of the X86 versus Mac debate. I see a lot of irrational discussions on both sides. Neither side seems willing to believe that there may be some merits to the other side's arguments. We are right, the PC is fundamentally broken because they are hindered by Windows. They are right that we are hindered by the performance of our hardware. Which is worse?
Personally, I just use my PCs for gaming now. I use the Powerbook for everything else. This kind of proves my point. PCs perform better (when they aren't crashing) and Macs just plain work, even if they aren't speed demons.
[ 07-07-2002: Message edited by: Thresher ]</p>
Best
--
Ed M.