alfismoney said:
i'm not saying that every chip on the market is supposed to be bug free or that the core is bad for having errors, my point is that intel isn't about to sit on its laurels and let what is essentially a public beta chip sit at the head of its platform for longer than it has to. unlike ibm, intel and amd are constantly replacing their processors with newer (and hopefully better) technology. hence the G5 never appearing as a low power mobile unit and revisions of the G4 still shipping in quantity after 7 years.
If Intel fixes the problems, it becomes a different chip - so in essence your statement that they won't fix the existing chip is true - even if they fixed it!
My point is that "errata" are common, and is a standard way of dealing with minor, trivial flaws in microprocessors. IBM does it, Moto/Freescale does it, AMD does it, everybody does it.
One would never go back and make a new pass of silicon if a couple of lines of BIOS/EFI or HAL code can ensure that the erroneous situation cannot occur.
alfismoney said:
my guess is that intel is ramping up production because they're unhappy with what they're shipping today, whether it's due to architecture problems or the fact that it was forced to fall back on a 32 bit solution...
What architecture problems? There are a few errata.
Apple is the one that "fell back" to a 32-bit solution, not Intel.
Like many big companies, Intel has multiple teams developing products in parallel. The Yonah project was a relatively modest enhancement to the Pentium M line (add dual core, SSE3, minor bumps).
Merom/Conroe/Woodcrest are part of the much more ambitious Next Generation Architecture project. It's been under development for several years (note this March 2004 article
http://www.theregister.co.uk/2004/03/17/intel_plots_4mb_l2_64bit/).
Look at Yonah as an "insurance policy" - had NGMA slipped, Intel would still be able to ship a dual-core chip in the notebook space in the interim. (A fairly cheap insurance policy, especially since Merom chips can be used in existing Yonah motherboards!)
alfismoney said:
...because they are cutting down on their test time and might have a large bug count on the final products. personally, i think pushing up the release is a good thing.
Even though that "large bug count" is 3/4 of the number of bugs in the PPC chip in the iBook G4 ???
The point of testing is to find and document the errata, not to continue to go back to the silicon drawing board and try to make a perfect chip. You only do that when an erratum is so serious that it cannot be worked around in low-level software (the Pentium FP bug being a good example of that).
Moving up the release is a good thing because it will get 64-bit dual-core laptops and low-power desktops here sooner. It's a good thing (for Intel) because it will give them an advantage over AMD. (I wonder what Sun's next marketing campaign will be, once Intel chips are faster and lower power than AMD. Ooops!)
Moving up the release because of a few errata is not an issue - because who's to say that Merom won't have *more* errata?