On GameFAQs: What causes the Red Ring of Death?
BNET Business Network:
BNET
TechRepublic
ZDNet
TalkBack 9 of 10:
Next »
« Previous
Let me elucidate
``You ask when they are going to end the X86 quagmire but then you scoff at Itanium which does just that. ''

I scoffed at BOTH the Dynosauric x86 (IA32) (and its short-term reprieve 64-bit extensions) AND the Itanium (IA64).

``The resistance to the Itanium has been primarely cost not performance. It was also hurt by the delay in launch of Windows Server 2003 which was the first Microsoft server OS to support Itanium. There was also resistance because of the need initially to replace all the apps that ran on it.''

I think you 've just answered to your own reply: Resistance was DUE TO LACK of a decently smooth way to MIGRATE application code from the x86 platform to the IA64 platform.

Elaborations/Reasons: the EPIC model is semantically very far from that of the IA32 that was pervasive (but crappy). In the beginning 1) good compilers to IA64 were missing AND 2) a reasonably smooth UPGRADE path to this new platform (run-time libraries, 32->64 environment, etc.)

Due to insistance of Intel to stay in the 32 bit world and extend it / feed it with IV, developers had NO easy eay to migrate the 32-bit apps to the 64-bit world of the IA64. UNIX went ovr this transition during the early 90's and the transition was MOSTLY smooth. The lack ofcourse of a working IA64 MS windows was another factor. If MS has such a hard time fixing its 32-bit 'kernel' then transitioning the same to the 64-bit model must be proven an extremely nerve-wraking task (for the MS 'engineers'). Lack of the tools for this 32->64 bit transition is PARTLY to blame.

A little reminder on performance: When Itanium-I came out it was a very lackluster performer at best. Only the manufacturing of Itanium-2 gave some glimmer of hope to Intel to its supporters (HP mainly).

I believe that the Itanium-2 is a very decent 64-bit platform but its main problem is that it is an EPIC: this requires the optimizer to jump through hoops to schedule statically the instructions, otherwise any performance AFVANTAGES of the EPIC and the Itanium datapath WILL NOT be leveraged.


``Still all things considered it was a valiant attempt to lose the legacy X86 code and come up with a new processor.''

Yes, it really is a valiant attempt but apparently it was one against Intel's phylosophy of sticking with the old, lest we break the legacy code.

One needs to break away from the obsolete schemes
and methods when clearly they do not work anymore. But if you want people to start using a new faster high-way then you need to build a passable path to get there from the one they have access, instead of hopping that they will just walk on a tricky trail to get to it.

Apparently, Intel / HP were more confident or less wise on the complexities of designing a 64-bit replacement for the IA32. HP has no excuse since the travelled that path with their PA processor and they had 'access' to DEC Alpha's designers.

Bottom line: the heavy burden of the CISC legacy that Intel worked so hard to create and impose on all x86 users is very HARD to get rid of (poetic justice). Even HP started turning their back to the Itanium, as they realize that huge and quick profits will materialize only AFTER huge investment is made on this platform.

Summary: Intel stuck to the x86 and now its left with 99% legacy systems that can only survive with temporary extensions, but which clearly have a very finite useful life. Ity opted to go ALL THE way the other direction and now it has created a polarization competing basically with its own legacy systems. Had they be mo wise they would have thought of a reasonable migration path from the x86 to their new platform. IA64 needs TONS of investment on the part of system and appl developers and h/w manufacturers.

Although I would use IA64 at any expence vs x86, its future is uncertain.

-m
Posted by: michael-t   Posted on: 03/01/05 You are currently: a Guest | Members login | Terms of Use

Alert moderator to an offensive message

Subscribe to this discussion via Email or RSS

Intel to spotlight new  michael-t | 02/28/05
Your post is confusing to say the least  ShadeTree | 02/28/05
Good point but...  Prognosticator | 02/28/05
Good vision  Roger Ramjet | 03/01/05
There is no P4 core at the ...  ShadeTree | 03/01/05
I disagree  Roger Ramjet | 03/01/05
Read my lips ....  ShadeTree | 03/01/05
I looked it up  Roger Ramjet | 03/02/05
Let me elucidate  michael-t | 03/01/05
Clarification accepted(nt)  ShadeTree | 03/01/05

What do you think?

SponsoredWhite Papers, Webcasts, and Downloads

advertisement

SmartPlanet

Click Here