[Giv]

Originally posted by rmendis:
In otherwords isn't on chip code morphing software likely to benefit from more powerful micro-architectures available in the future?
I expect there ought to be a convergence (in performance) between native and code-morphed software....IMHO the technology is in it's infancy and it shouldn't be dismissed so quickly. Some technologies take time to develop.

I am not dismissing "code morphing" (I prefer the term dynamic code generation). I did some work in that domain and I am aware of the practical challenges. It's just that the hype generated by Transmeta and their zealots is a bit much for my taste.

Originally posted by rmendis:
So much so that perhaps soon, even with the performance penalty of code morphing, a VLIW/EPIC powered x86 processor could outperform one based on its native architecture?

By definition this can't ever happen. Native is native is native. You can't beat "native".

Where dynamic code generation may (or will) win is that "being native" is becoming too expensive, so micro-architectures can be hobbled by the need to be backward-compatible. If you can count on a VM / JITC to execute 99% of the code, then you can probably evolve the processor faster or in more aggressive ways. That's the theory. In real life evolving a JIT to follow hardware changes can also be expensive and difficult.

I can't provide any info on Transmeta or HP Aries future. If you are in position to place orders ($$$), ask the companies for more information, they may even tell you under an NDA. If not, read the published papers or wait for the next press releases to appear.

[rmendis]

Originally posted by Giv:
I am sure the 970 will be an excellent processor when running Java.

I was taking a look at a couple old NEXTWORLD Extra magazines i had sitting on my shelf.

They were the issues just before and just after NeXT went software only.
A very harsh and bold move that split the company...and as it so happens led the company down a path that led to...it's aquisition by Apple

The interesting thing to note is that what caused this to happen to NeXT was in fact the Motorola processors.

Just like the PowerPC after it, the 68k series was very powerful and leaps and bounds ahead of it's Intel x86 equivalents. i.e the 020, 030 and 040 were faster than the 286, 386 and 486 respectively.

However, Motorola got stuck in a rut after the 040,
pretty much like how it is at the moment with the G4. Intel leapfrogged Motorola with fater 486s and then the Pentium. Motorola had by then decided to give up on developing the 68k line in favour of the new PowerPC RISC processors it was developing with Apple and IBM.

(In fact, all this could have been the result of Steve Jobs' NeXT-IBM pact/investment. Scully later very obviously outdid Steve by forming the Apple-IBM alliance which blossomed the PowerPC. As a result Motorola gave up on the 88k and even the 68k lines, which later left NeXT in a quandary).

One wonders how different the IT world would look now if NeXT had chosen SPARC as it's processor for the NeXT computer instead? (In fact much of the original NEXTSTEP was developed on old Sun machines if i remember by NeXT history correctly. Sun had started to open up the SPARC architecture to third parties, so this would certainly have been a possibility for NeXT. Or did they not want to make glorified SPARCstations? At least better than making what were essentially glorified Macs

Back to the thread: Apple (in place of NeXT) and Motorola appear to be at a similar junction in history again. Who says history doesn't repeat itself?

Motorola has left Apple in a quandary, this time with no proper future in desktop and workstation microprocessors planned out. Is it because of Mac OS X and it's portablity? Is this because of the threat of Intel and the Itanium? Is it because Apple indeed planned on making a transition to another architecture. Many of the plans for Rhapsody were carved out during Gil Amelio's term. Perhaps back then they had indicated to Motorola that the PowerPC's days were numbered? And that they would be moving to a different architecture once the transition to the new/modern OS was complete?

Is the PowerPC 970 (and 980) simply stop gap processors until the transition to Mac OS X is complete and Intel's IA64 architecture hits the desktop. Reports suggest Intel not providing a 64 bit (i.e IA64) solution for the desktop for another 3 or 4 years at least. The PowerPC 970 (and possibly the 980) would suffice until then.

What is for sure is that Apple is at a similar junction to NeXT before it had to abandon hardware. This time be sure for Steve Jobs to take a different route...and let's hope that this time he gets it right

[rmendis]

Originally posted by freakboy2:
If OSX could run on intel hardware, I'd go build myself a 400$ machine that was 2x as fast as the apple machine that costs 3,000$ and then I'd spend 120 bucks on jaguar and be done with apple.

Well, Apple knows if they had taken up Bill Gates offer to port Mac OS to PCs in the mid to late 80s (?) then there would be no Windows today.

When Apple finally got around to licensing it was too little too late.

So *timing* is key!

Intel have suggested that it will be at least another 2-3 years before IA64 hits the desktop. That is before it will offer a 64 bit desktop solution...and bet your bottom dollar it will be based on I!64 even though it may not be called an Itanium. Maybe Xeon or even a Pentium.

So porting Mac OS X to Intel/IA64 if it is ever going to happen commercially, ought to happen before then. Before IA64 boxes become cheaper than Apple PowerPC hardware.

If Mac OS X were ported to Intel/Itanium now, it realistically speaking would not be a threat at all to Apple hardware. The cheapest Itanium processor is $4000+ casting it outside the reach of most consumer Mac users. Intel has reported that it will maintain that price for the Itanium for the next year or two.

So in actual fact, this may happen sooner than people imagine.
I'd bet WWDC next year or if the PowerPC 970 is ahead of schedule and Apple delivers 64-bit Mac OS X soon, then maybe even at WWDC this June!?

What is for sure, Apple will either ink a deal with a big name Itanium server shop (like HP or even IBM) or purhaps purchase a company like SGI and sell Mac OS X Itanium SGI-branded servers (effectively) for itself. Another option i had not considered, though i believe is less (or un)likely, is for Apple to release their own Itanium Xserve....however this would mean it would conflict/overlap with its G4 and possibly 970 Xserve and it's already a tight and new market. It would make more sense to have Mac OS X on a high end server classed outside the reach of Apple's brand and product protfolio.

SGI is a very good match for Apple and is bargain basement. It would make a perfect "server brand" for Apple, eventually migrating SGI servers and superclusters to Itanium Mac OS X servers.

--

Intel's Itanium offers Apple an opportunity to sync itself with the rest of the IT industry (in terms of price & performance of its PCs). As PC vendors migrate to IA64 in a few years, so can Apple.

Apple has that time to broaden its portfolio as it has with more software (Mac OS X, iLife, iWorks, AppleScript Studio, WebObjects, FinalCut, Shake, etc..) and broader range of hardware iPod, Xserve, Xserve RAID. And new services ranging from .Mac to iTunes music service.

Once it is has done that, it will be less reliant on hardware sales and therefore be able to compete better with PC vendors such as Dell on the same footing - that is with commodity PC microprocessors and parts.