SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?


To: Bill Jackson who wrote (28065)1/22/1998 8:04:00 PM
From: Petz  Read Replies (2) | Respond to of 1574060
 
Bill, re:<I had read that Intel could "fix" bugs with an flash updateable area>

I'd read that too, but I'm dubious of it working more than about half the time. There are over 30 errata on the Pentium II for which Intel says there will never be a fix. If it was so easy to update the flash area, why don't they just do it for new production? It wouldn't add one minute to the process time of a CPU.

I am (last 6 years) a space systems engineer where reliability is paramount and redundancy is the rule rather than the exception. But trying to anticipate design flaws is futile -- there are a billion ways to do something wrong. In a CPU, the speed penalty for using FPGA (programmable hardware) technology all over the place is too great. An old style CPU with bit slice processors, a sequencer and microcode could often be "fixed" easily. Heck, you can add floating point SIMD instructions like KNI or AM3D to such a machine easily, but the modern CISC (complex instruction set) microprocessor is a far cry from that design.

Actually, Intel's Merced (IA-64) architecture, by putting most of the burden of parallel operation on the compiler software rather than the hardware, could be designed to be more "fixable" and programmable than the Pentium and its ilk. I hope they design it that way, because with 3x as many transistors as a Pentium II, it probably won't be perfect the first time.

Petz