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: Charles R who wrote (59292)	5/22/1999 2:29:00 PM
From: Tenchusatsu	Respond to of 1572938

<I should clarify/correct myself - what I meant was the challenge of tuning the compiler to a new CPU. i.e., not the function of the CPU itself but getting software to work on it. Would that change your response?> I don't know what exactly you're asking, but I'll take a shot anyway. One of the big challenges in architecting the EPIC instruction set is allowing for generational changes in processors, such as Merced to McKinley. There are hint bits that are tacked onto every three instruction packet telling the processor which instructions can be executed in parallel and which are dependent on previous ones. It's then up to the processor to figure out how many instructions it can gobble at a time. This way, the compiler can schedule the maximum amount of ILP (Instruction Level Parallelism) possible without regard to the ILP capabilities of the processor. Code compiled for Merced should be able to run on McKinley without any recompile necessary. And vice-versa, so the best case in performance is to compile to the ILP capabilities of McKinley, then let Merced figure out how to digest the ILP from there. Tenchusatsu

To: Charles R who wrote (59292)	5/22/1999 4:35:00 PM
From: Scumbria	Respond to of 1572938

Chuck, I should clarify/correct myself - what I meant was the challenge of tuning the compiler to a new CPU. i.e., not the function of the CPU itself but getting software to work on it. Would that change your response? I don't see the basic IA64 compilers as being particularly difficult to write. The difficulty with them lies in the code optimizations, which is an enormously complex problem. Compiler optimizations will not prevent the CPU from shipping however. The problems I see with IA64 software will be created by hack programmers who think they can do a better job than the compiler. Improperly written IA64 machine code will be a nightmare to debug. Scumbria