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.

Technology Stocks : Intel Corporation (INTC) -- Ignore unavailable to you. Want to Upgrade?

To: Elmer who wrote (142005)	8/20/2001 12:10:21 PM
From: Charles Gryba	Respond to of 186894

EP, I am not sure how familiar you are with programming but you can make a compiler optimize specific chunks of code easily. All you have to look for is patterns and substitute hand-tuned assembly code instead of generically created code. The way to do it is to run SPEC through a profiler first and see where most of the time is spent and concentrate on optimizing for those parts. Still, not a trivial task. I don't know if Intel would hire a team of assembler code experts to do this. Constantine p.s. This can be proven/disproven easily if someone has access to the compiler source code.

To: Elmer who wrote (142005)	8/20/2001 2:58:29 PM
From: pgerassi	Read Replies (1) \| Respond to of 186894

Dear Elmer: Yet this, highly regarded by you, compiler fails to compile C code, that is accepted by all other commercial and the freeware GNU C compilers, from the Scimark benchmark. You can talk to the creator Tim Wilkins on the inability of Intel to successfully compile for over 9 months three C programs that implement three different scientific applications. This is a compiler that would not be acceptable in any production environment. Its SPEC scores should be disallowed on that basis alone. Using those commercial compilers, P4 runs slower than a Duron whose clock is less than 66% that of the P4. Thus, a TBird at 1.33G outruns a 2G P4. From tests, Palominos run about 10% faster than Tbirds on this benchmark. Thus, a 1.2G Palomino outruns a 2G P4 on real code by real compilers using best algorithms and best libraries. Yes, SPECint and SPECfp have outlived any real usefullness for technical people. Anand's database test benchmark and Tim's Scimark are far better at predicting the real performance of a system in their respective work areas than the SPEC scores. Pete