Intel scientists find wall for Moore's Law

TalkBack 26 of 31:

Next »

« Previous

• Thread View
• Flat View

Speed of Light and Speed of design-to-market

The problem is not so much the limitations that good ol' physics palces on the chip manufacturing. The problem is the design and architecting of chips GIVEN these limits!

Intel processors follow the expected speed improvements in VLSI technologies. That have been predicted by Moore several decades ago.

But, where are the improved chips that are based on ingenuity --inspite of Moore's law? We have seen Intel cranking up the clock speeds with corresponding multiplication of the number of pipeline stages. P4 has some 21 stages so that the time / stage matches the shortened clock period. They had to resort to RISC-style micro-ops for their CISC elephant ISA (as was aptly pointed to by by some responders).

But where is the improvement in chip architecture? The Itanium is Intel's answer to this question. Intel and HP have been trying to crack into the high-end processors space with their IA-64 since 7 years ago. The Itanium is NOT compatible with the classic IA32 ISA. Apparently Intel realized that to convert an x86 into an implementation for IA-64 would require another 5-7 years of R/D and did what they should have done since early 90's: dropped it and started from a better design.

This hasn't been easy for Intel. Only recently they started producing processors that are competitive with high-end 64-bit RISC chips, such as P4 of IBM. But the battle to improve performance is lengthy and strenuous.

Why? If Intel with its HUGE human, tech and financial resources cannot make it, then who can? And what will Itanium 2 look like compared to the upcoming P5 IBM schips?

The long term issue is that, if Intel decides to drop their classic 32 bit chips and push for their Itaniums, then the entire collection of x32 s/w must be recompiled to be able to use it. This will require a humangous h/w and s/w upgrade cycle for the IT departments.

Then maybe AMD with its compatible 32 and 64 bit processors is one answer.

The bottom line is that organizations who have invested on x86 will pretty much have to scrap and start from scratch. Whose fault is it? Intel's silly designs or the decision making of CTOs who chose it? I guess both but everyone is going to pay

BTW going RISC is the best long-term decision We transited from the 32 to the 64 bit world with some effort, which compares NOTHING to the effort needed to extricate the x86 industry from the x86 chips.

-m

-m

Posted by: michael-t   Posted on: 12/01/03 You are currently: a Guest | Members login | Terms of Use

Alert moderator to an offensive message

Subscribe to this discussion via Email or RSS

SponsoredWhite Papers, Webcasts, and Downloads

• The Impact of Virtualization Software on Operating Environments VMware Today's use of virtualization technology allows IT professionals to ... Download Now
• The True Costs of Virtual Server Solutions VMware In an economic environment that is repeatedly heralding the message "do ... Download Now
• Email Security and Archiving - Clearer in the Cloud Google The time is NOW for businesses and organizations of all sizes to implement ... Download Now