[hmurchison2001]

Out of respect for Bonkers I moved this response to its own thread to respond to lateralus.

No... The number of people who oppose the Intel switch is indeed on the rise around the net. Immediately after the announcement, everybody seemed to be in rather good spirits about it. But people are digging into the deficiencies of x86 and realizing Intel's usual approach to design, and the second guessing has begun.

Why was the move to Intel smart? I'll agree that it was a good move for portables, as low power consumption has become a focus of Intel. But you'd be hard pressed to come up with any reasons to support the switch on the desktop and server front.

More over, the fact that there is no clearly attractive PowerPC route to go down for portable chips is a temporary sympton, it has nothing to do with any flaw or draw back to PowerPC. It's simply a matter of IBM not wanting to custom tailor low-power chips for Apple's portable line without some R&D help from Apple. Which, in reality, Apple should be willing to provide given the cost of developing the things.

The only people I've run into who don't have an opposition to the Intel switch are those who don't know the differences between the two architectures.

[hmurchison2001]

My respones

I find the opposite to be true. I find the ones that oppose the switch seemed to be the least informed of the architecture. For instance….my knowledge is just that of a fan however here is what I know.

Intel has two current architectures developed by two different teams. The Cali team created the Netburst architecture which is and has been the basis for the Pentium 4. The Israeli team created the Pentium M Banias and Dothan cores. The Israeli team has hit paydirt. The Pentium M is a rousing success and Netburst is cancelled. That’s the preface for why I believe Apple is moving to Intel.

Apple have carried the PowerPC torch for over a decade but lately it’s been abundantly clear that IBM is a bit reticent to devote the time and funds necessary to make a splash in the consumer/SMB markets for CPUs. Remember it was IBM that left AIM to pursue clocking G3s higher and eschewing Altivec SIMD. Anyone remember those blazing G3s. Well they oversold themselves again when they told Steve the 970 would have legs and frankly it doesn’t. In fact not only doesn’t the G5 have legs but it cannot match the Intel CPUs of today in Integer performance and in memory bandwidth (see Anandtech’s reports..we got demolished) .

Thus when I look at IBM’s history of overpromising and underdelivering I get dejavu. Fool me once shame on you…fool me twice shame on me. The Intel Israeli team has been on schedule and on target. I don’t doubt that they’ll have the Dual-Core Yonah chips in full production this last qtr and laptops will be shipping with them in Q1 2006.

Next up will be their revamped Merom core which supports EM64T and is approx %30 faster per clock. Now ask yourself. If the PPC can’t keep up today…what makes us think it’s going to match Intel’s next champion core?

What we know about Merom/Conroe

64-bit
Total TDP of only 65-70 watts for both cores and 2 or 4MB L2 cache
Shared L2 cache which is faster than the 970mp’s communication over the FSB
65nm process
Intel tech like. Vanderpool virtualization, support for FB-DIMM, iAMT2, EBD, TCPA

I think the real question is. After 10 yrs of disappointmens, why should many of us believe in IBM anymore?

Remember this “We’ll be at 3Ghz in a year!” how long ago was that?

[Eug Wanker]

the fact that there is no clearly attractive PowerPC route to go down for portable chips is a temporary sympton, it has nothing to do with any flaw or draw back to PowerPC

That is a pretty big "flaw" IMO, considering laptop sales are starting to overtake desktop sales nowadays.

However, probably the more accurate reason is that Steve just got sick of the underdog status in CPUs. Apple had to pay IBM to come up with these chips, and even after money was spent, IBM still lagged behind on the portable side.

So why swim upstream and pay for it?

To be frank, I was surprised that Steve went this route however, NOT because I think PowerPC is inherently better, cuz I don't, but because of the migration costs, and because I didn't realize that OS X was as polished as it was on Intel. I don't think x86 is inherently better either, but it does have the advantages of better portable options with Intel, and more importantly, TWO MAJOR MANUFACTURERS.

And no, Freescale doesn't count. Their parts are just too far behind. If you were to count Freescale as a PowerPC option for Macs, then you have to count Cyrix as an option on the x86 side.

What this means is now Apple has the ability to focus on other things instead of having to worry constantly about the future of its CPUs.

I find the opposite to be true. I find the ones that oppose the switch seemed to be the least informed of the architecture.

I'm not expert either, but from what I've read around the net, the switch is not as big a deal as some try to make it out to be. The small guys on Cocoa port within hours, and the big guys on Code Warrior are already committed to making the switch. Now, there are a very focal few who have built their lives and apps around Altivec, and they will be hit hard, but they are few and far between.

Remember this “We’ll be at 3Ghz in a year!” how long ago was that?

To be fair, IBM never made that statement (publicly). As far as I concerned, Steve Jobs should never have made that statement either.

[Randman]

You know, a side effect of all this will be the demise of the gigahertz myth. With Leopard and Vista basically going toe-to-toe, people will see Windows vs Macs using one measurement. I think that will sway a good number of users to Apple along with the iPod halo effect.

[Lateralus]

Originally Posted by Eug Wanker

That is a pretty big "flaw" IMO, considering laptop sales are starting to overtake desktop sales nowadays.

However, probably the more accurate reason is that Steve just got sick of the underdog status in CPUs. Apple had to pay IBM to come up with these chips, and even after money was spent, IBM still lagged behind on the portable side.

So why swim upstream and pay for it?

To be frank, I was surprised that Steve went this route however, NOT because I think PowerPC is inherently better, cuz I don't, but because of the migration costs, and because I didn't realize that OS X was as polished as it was on Intel. I don't think x86 is inherently better either, but it does have the advantages of better portable options with Intel, and more importantly, TWO MAJOR MANUFACTURERS.

And no, Freescale doesn't count. Their parts are just too far behind. If you were to count Freescale as a PowerPC option for Macs, then you have to count Cyrix as an option on the x86 side.

What this means is now Apple has the ability to focus on other things instead of having to worry constantly about the future of its CPUs.

Well, echo my post why don't you.

Originally Posted by Lateralus

The general concensus is that the Intel switch is happening because of the following two reasons;

1) With Freescale/Motorola out of the picture as a serious provider of PowerPCs for Macs, Apple felt boxed in as far as notebooks and small form factor machines go since neither of these types of machines have really been a concern of IBM's when designing processors. IBM's bread and butter is mainframes and servers, which only benefits Apple on the Power Mac front. But with notebooks now outselling desktops, notebooks have become a larger focus for Apple than ever.

2) Steve threw a fit because IBM isn't super-motivated to crank out the custom-tailored 'best processors in the world' that Apple wants. Reason being that as large of a company as Apple is, they still only account for around 5% of IBM's total microprocessor sales. However, development of processors for Apple accounts for a significantly larger percentage of IBM's total R&D for microprocessors (I've heard more than 20%).

I'm opposed to the switch. IBM, on the whole, has had better offerings than Intel across the board on the desktop/server front for years and the PowerPC ISA is far and away a better architecture than x86 is. And with the foundation of OS X being as 'portable' as it is, there's no reason why Apple couldn't continue using IBM's PowerPCs for Xserves and Power Macs and reserve the Intel switch for the only lines that actually need it; the portables.

In fact, I'm hoping they will do so.

[Lateralus]

Originally Posted by hmurchison2001

I find the opposite to be true. I find the ones that oppose the switch seemed to be the least informed of the architecture. For instance….my knowledge is just that of a fan however here is what I know.

Intel has two current architectures developed by two different teams. The Cali team created the Netburst architecture which is and has been the basis for the Pentium 4. The Israeli team created the Pentium M Banias and Dothan cores. The Israeli team has hit paydirt. The Pentium M is a rousing success and Netburst is cancelled. That’s the preface for why I believe Apple is moving to Intel.

The problems with x86 go far deeper than NetBurst. Which, considering the depths of x86s defects, can only be considered 'cosmetic'.

As jcadam so succinctly pointed out in the other thread, x86 was a defective child at birth. Even Intel admits that they've wanted to kill it off several times in the past but haven't had the resolve to because of the software breaking that would be involved with engineering a new architecture. So, instead they've laid patch upon patch with each subsequent x86 CPU to force the architecture to do things it was never meant to do.

The fact of the matter is that of all the major ISA's, x86 is the most ass backwards and the least progressive.

I cannot support an architecture that revolves around dragging legacy protocols on and on through the decades.

[tooki]

Well, it came as no surprise to me that OS X had been "leading a double life" -- there was never any other reasonable explanation for Apple consistently releasing an Intel build of Darwin on the coat tails of every major OS X release. I have stated in the past that "I fully believe that Apple keeps Mac OS X running on Intel" behind closed doors. (What DOES amaze me is that they were able to successfully prevent even a single person from leaking even the tiniest bit of hard evidence of this for 8 years!)

Considering that Apple even had System 7 running on Intel (reportedly with high performance) before the PPC switch, I suspect that Apple has always kept Mac OS running, to some degree, on Intel.

tooki

[Eug Wanker]

Originally Posted by Lateralus

Well, echo my post why don't you.

Hehe. Didn't see that one.

Originally Posted by Lateralus

The problems with x86 go far deeper than NetBurst. Which, considering the depths of x86s defects, can only be considered 'cosmetic'.

As jcadam so succinctly pointed out in the other thread, x86 was a defective child at birth. Even Intel admits that they've wanted to kill it off several times in the past but haven't had the resolve to because of the software breaking that would be involved with engineering a new architecture. So, instead they've laid patch upon patch with each subsequent x86 CPU to force the architecture to do things it was never meant to do.

The fact of the matter is that of all the major ISA's, x86 is the most ass backwards and the least progressive.

I cannot support an architecture that revolves around dragging legacy protocols on and on through the decades.

Not that I have a clue, but there's a lot more to chip design than philosphical online hand-waving as you very well know. The proof is in the pudding. Intel has been wildly successful with its x86 products, both in terms of speed and in terms of profits. And it's not as if the x86 chips of decades ago bears any true resemblance to the x86 of today.

Originally Posted by tooki

Well, it came as no surprise to me that OS X had been "leading a double life" -- there was never any other reasonable explanation for Apple consistently releasing an Intel build of Darwin on the coat tails of every major OS X release. I have stated in the past that "I fully believe that Apple keeps Mac OS X running on Intel" behind closed doors. (What DOES amaze me is that they were able to successfully prevent even a single person from leaking even the tiniest bit of hard evidence of this for 8 years!)

Considering that Apple even had System 7 running on Intel (reportedly with high performance) before the PPC switch, I suspect that Apple has always kept Mac OS running, to some degree, on Intel.

tooki

I agree. I knew Apple had something deep in the labs. However, I was still surprised at how well it worked. I had guessed it was Darwin with a buggy OS X, but apparently the OS X on Intel is quite polished.

And despite Rosetta's drawbacks, I was shocked to see how well it worked. So far the reports are that it is nearly invisible (assuming you don't depend on Altivec).

[Big Mac]

Originally Posted by Eug Wanker

Hehe. Didn't see that one.

Not that I have a clue, but there's a lot more to chip design than philosphical online hand-waving as you very well know. The proof is in the pudding. Intel has been wildly successful with its x86 products, both in terms of speed and in terms of profits. And it's not as if the x86 chips of decades ago bears any true resemblance to the x86 of today.

The proof is indeed in the pudding - just witness the amount of negativity PC geeks feel toward the current level of Intel performance. Precisely when pessimism about Intel's near to medium term prospects is at its height, Apple decides to marry Intel.

[Eug Wanker]

Originally Posted by Big Mac

The proof is indeed in the pudding - just witness the amount of negativity PC geeks feel toward the current level of Intel performance. Precisely when pessimism about Intel's near to medium term prospects is at its height, Apple decides to marry Intel.

???

AMD fanboys complain about Intel. Otherwise most PC guys seem to consider Intel desktop chips reasonable, and Intel laptop chips superior, compared to AMD's offerings.

[hmurchison2001]

Intel and Microsoft are both victims of their success. While they love having a relatively captive audience that same success makes it very difficult to make significant changes without breaking everything.

Funny thing is the X86 is indeed old and crusty yet Intel still manages to make it faster than most of their competitors including Apple.

Intel and Apple is a good matchup because Intel so desperately wants to move away from Bios and ATX and the same boring form factors that are entrenched. Apple is willing and almost mandated to develop new and exciting products.

We really can't blame Intel...they have pumped billions into Itanium only to have software vendors ignore it. The status quo is always safer than the bleeding edge.

[Lateralus]

Intel could force a migration to IA-64, they're just afraid that it would involve ceding marketshare to AMD in the progress. Which it probably would, since you can bet AMD would ride x86-64 for all it's worth during any transition Intel makes, just to grab market share by supplying to companies that drag their heels in going along with a transition.

But Intel is a staple in the industry, so any marketshare concessions would be temporary. And elevating the standard of an entire industry would probably be worth it, not only for consumers and developers, but for Intel itself. Intel would probably much rather have their competitive standing based on something newer and more versatile than x86 through the coming decade. But, again, they're too paranoid of AMD for it to happen.

I have no problem with Intel as a company, I think they have developed some great tech. And I very much admire IA-64/Itanium, and it's sad that it was drowned by bureaucracy.

[hmurchison2001]

http://www.digitimes.com/bits_chips/a20051018PR201.html

First mentioned by sources in 2004, Merom was announced by Intel in August this year at IDF Fall in San Francisco. The processor will be manufactured utilizing 65nm technology. Compared to Yonah, it will have a larger level-two cache (presumably up to 4MB) and will contain some micro-architecture innovations, according to Eden. One of these innovations will be a higher performance 4-issue out-of-order engine with deeper buffers and a pipeline extended to 14 stages. Another promised features include more power efficiency, direct L1-to-L1 cache transfer and improved memory access, Eden pointed out. He also said that Merom will feature some substantial improvements in floating-point unit (FPU) performance, but turned down questions about performance measurements

Man I just have a good feeling about the Merom core. It has to be good if Intel scrapped Netburst.

[Lateralus]

Everybody harps on about Merom and Conroe as if IBM intends to become stagnant in their own developments.

I'm much more excited about the 980 than I am about anything from Intel. The 980 will be a POWER5 derivative, which among other things, provides 50% more performance at the same clock speed than the POWER4.

[Catfish_Man]

Originally Posted by Lateralus

Everybody harps on about Merom and Conroe as if IBM intends to become stagnant in their own developments.

I'm much more excited about the 980 than I am about anything from Intel. The 980 will be a POWER5 derivative, which among other things, provides 50% more performance at the same clock speed than the POWER4.

1) With Apple switching, what makes you think the 980 will ever exist? Why should IBM spend money with no hope of making more money from it?
2) You're aware, I hope, that most of the POWER5's improved speed comes from its VERY extensive cache and memory system, which the 980 would not have.

In summary: Yes, the 970 will stagnate. All processors stagnate without money, and the 970 has never sold in the kind of quantities necessary to support a large high end design, even before Apple announced they were going to stop using it.

I do have to agree though, that the expectations for Conroe/Merom are set rather high. Intel has been known to miss its predictions as well We'll have to wait and see.

[Eug Wanker]

Originally Posted by hmurchison2001

Man I just have a good feeling about the Merom core. It has to be good if Intel scrapped Netburst.

I'm looking forward to Merom as well, specifically in an iMac.

I probably won't buy in 2006 though even if it does get a dual-core Merom, unless Apple releases a 23" Intel iMac, with a GPU that supports H.264 acceleration.

Originally Posted by Lateralus

Everybody harps on about Merom and Conroe as if IBM intends to become stagnant in their own developments.

I'm much more excited about the 980 than I am about anything from Intel. The 980 will be a POWER5 derivative, which among other things, provides 50% more performance at the same clock speed than the POWER4.

I'd be surprised if the 980 (or whatever it would have been called) will ever appear. There is essentially no market for it. The game console guys don't want it. Apple doesn't seem to want it. And IBM itself doesn't want it either. I wouldn't be surprised to see 970MP to get one more speed bump in 2006, but then it's all she wrote for PowerPC on the desktop.

That said, it would be nice to see it appear, just for one last hurrah.

[mduell]

the fact that there is no clearly attractive PowerPC route to go down for portable chips is a temporary sympton, it has nothing to do with any flaw or draw back to PowerPC

I'd say the "flaw" is that no one with money is interested in PPC. Namely building low-power high-performance PPC chips.

Originally Posted by Lateralus

The problems with x86 go far deeper than NetBurst. Which, considering the depths of x86s defects, can only be considered 'cosmetic'.

As jcadam so succinctly pointed out in the other thread, x86 was a defective child at birth. Even Intel admits that they've wanted to kill it off several times in the past but haven't had the resolve to because of the software breaking that would be involved with engineering a new architecture. So, instead they've laid patch upon patch with each subsequent x86 CPU to force the architecture to do things it was never meant to do.

The fact of the matter is that of all the major ISA's, x86 is the most ass backwards and the least progressive.

I cannot support an architecture that revolves around dragging legacy protocols on and on through the decades.

Meanwhile dragging around all that ass backward legacy cruft, x86 still matches or beats every other platform in speed and power consumption. With the new Opteron chipsets it can even do massive SMP.

Originally Posted by Big Mac

The proof is indeed in the pudding - just witness the amount of negativity PC geeks feel toward the current level of Intel performance. Precisely when pessimism about Intel's near to medium term prospects is at its height, Apple decides to marry Intel.

PC geeks feel great about the current level of Intel performance and power consumption in the mobile chips. And in case you haven't noticed, people are buying more laptops than desktops these days.
NetBurst was a good idea at the time, but it's time has passed. It was short lived, but for most of its life it lived well.

Originally Posted by Lateralus

Everybody harps on about Merom and Conroe as if IBM intends to become stagnant in their own developments.

I'm much more excited about the 980 than I am about anything from Intel. The 980 will be a POWER5 derivative, which among other things, provides 50% more performance at the same clock speed than the POWER4.

IBM said 980 was to be developed alongside POWER5, but no one has seen hide nor hair of it. And POWER5 is making its way into some lower end boxes that may crowd out the 980 market anyway.