Silicon Investor (SI) -- The First Internet Community

STOCKTALK

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: Tenchusatsu who wrote (74851)	10/11/1999 1:48:00 AM
From: Saturn V	Read Replies (1) \| Respond to of 1573364

Ref < IBM...., they did their best to cast doubt on pushing ILP (instruction-level parallelism) any further. And it's definitely no coincidence that ILP is the basic premise behind IA-64 and EPIC technology. > The IBM argument has a ring of truth. On today's X86 processors the performance is not constrained by the number of registers,ALUs or Floating Point Units, but by the processor stalls while waiting for the data to be loaded from the main memory. Thus the Athlon with much greater ILP (Instruction Level parallelism) will be running neck and neck with Coppermine at the same clock speed. Thus IBM is posing a seemingly legitimate question, "Who needs ILP ?". However Merced appears to alleviate the memory bottlenecks by employing L3, and by explicitly prefetching of data by the compiler. Since it appears reasonable that the compiler will do a good job of prefetching data for most applications,IBM's motives appear suspect. Strategically IBM is in a quandry at a critical juncture. The Merced poses as significant a threat to IBM's core business, as 386 posed in 1985/87. The 32 bit memory of 386 allowed most computing applications to be executed on 386, and thereby ended IBM's hegemony on computing.[Luckily Microsoft's inept and slow migration to a true 32 bit operating system, made IBM's trauma less painful.] Most of IBM's non-legacy applications could be ported easily to Merced system, further shrinking IBM's role in computing. The Merced bandwagon has so much momentum that, baring a gross Intel fumble,it appears unstoppable. So IBM has little choice but to promote some 'legitimate FUD' about Merced, and try to derail the bandwagon. I have reviewed the IBM 64 bit processsor, and it is an interesting approach. It may even outperform Merced with specially coded server applications.[ The processor to processor bus will take special coding, a la Hypercube processors.] Each new configuration will probably require special coding. And users with legacy applications have no choice but to use this platform. Merced with its much better economies of scale will be a lot cheaper, and most portable big iron applications will indeed migrate to Itanium. So IBM prudently has it foot in the Merced Camp as well, just as it did with the 386. And I expect the 386 history to be repeated.