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: Paul Engel who wrote (89588)	10/6/1999 3:13:00 PM
From: Charles R	Read Replies (2) \| Respond to of 186894

Paul, <Count me in.> If the rumors from JC's have any merit, you have one hell of a long shot on your hand. jc-news.com "According to Daiki, here are the scores of the Coppermine (at 600MHz, presumably with 133MHz bus and PC800 DRDRAM) in spec95: specint_base specfp_base 29 25 For comparison, these are some unexpectedly sweet scores. I might wait for the numbers to officially appear at spec.org, as they're pretty striking -- unless I see the full configuration, I may have to assume that these scores are partially obtained with modifications to the compiler. There's nothing wrong with that, except that if, for instance, Intel changed their compiler to use SSE prefetching, then this would result in performance jumps of all PIII subfamily processors as well as AMD's K7 family processors (thereby making the current Athlon spec scores obsolete). As a quick note to the uninitiated, SSE (or 3DNow! for that matter) cannot handle the double precision floating point calculations required for spec (at least, not in the vast majority of the benchmark), so it'd be "illegal" to use the floating point portion of SSE in spec95. Anyway, if these scores are obtained without any unseen changes to the system, they are phenominal improvements, representing a 20% improvement in specint and an impossible sounding 57% boost in specfp. This would imply that PIII with lower latency 256K L2 by far outwits a PIII Xeon of much higher clock with 2MB (higher latency, albeit) L2. That's, well, why I suspect that either the report is in error or some sort of interesting optimizations are being used. Of course, this might just be me being an AMD apologist and making up excuses as to why these scores are significantly higher than that of the K7-600. Oh, and for reference, that c't page earlier today said that Coppermine is instead 12% faster in specint and 20% faster in specfp (than Katmai), which is actually just about exactly what I've been predicting (three cheers for my big ego!) over the past several months. BTW, I checked with a guy at AMD who said that the K7-700/266 with optimized compiler score is not with increased or sped up cache. (as a reminder, this yielded projected scores of 37 specint and 42 specfp). No idea how much the Ultra would get over this" Chuck