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 : Advanced Micro Devices - Moderated (AMD) -- Ignore unavailable to you. Want to Upgrade?

To: Tenchusatsu who wrote (45330)	6/26/2001 9:40:31 PM
From: Dan3	Read Replies (1) \| Respond to of 275872

Re: You might find that Itanium achieves some very respectable TPC-C numbers despite the low SPECint Could be. SUN SPARC64 is the leader in TPC-C right now (now that Alpha is dead), and their SPECint is mediocre at best. 1 GHz AlphaServer GS320. The GS320 supports 32 processors, and the latest 1 GHz Alphas are the fastest yet. The system achieves 230,533 tpmC with a price/performance rating of $47.90. This result is the highest submitted so far for the non-clustered TPC-C benchmark, beating the previous top score of 222,772 tpmC for the PRIMEPOWER 2000, which is based on 48 SPARC64 aceshardware.com

To: Tenchusatsu who wrote (45330)	6/27/2001 2:13:40 PM
From: Petz	Read Replies (1) \| Respond to of 275872

SPECfp per $ when computed properly is a valid comparison yardstick for scientific applications or workstation applications. Why? Because, for scientific applications, more and more of them are using a NUMA architecture with many CPU's as the preferred computing methodology. Roughly speaking, you double the number of CPU's and almost double the computational throughput. But $ cost of the CPU alone is the wrong measure. You have to take CPU + motherboard cost + memory cost. Athlon MP and Pentium III are the only game in town on that basis. A single Athlon MP 1.2 or Athlon 1.4 achieves a CFP2000 Rate of ~5.3 at a node cost of $200+$140+60 or $400 (256M DDR). The Itanium platform only achieves 7.22 with the "cheap" (and most cost-effective) version of Itanium. The platform cost for that has got to be over $3,000 since the CPU is $2,000 and the motherboard and power requirements are extreme. At least 7X the cost for 36% more performance. A pair of Athlon MP's beats the single Itanium performance by 21%! (8.72 to 7.22). The MP node cost would be less than $1,000 compared to $3,000 for an Itanium node. NO THANKS. For workstations, Itanium has a niche market for it's 4 -way and up platforms where 8.72 CFP2000 Rate is just not good enough. But 4 and 8-way PIII Tualatin systems will be much better buys. only Duron would do well in SPECfp/$ department, yet you don't see any Duron-based systems in high-end workstations Yes, because the definition of "$" for a workstation includes CPU, memory, motherboard and drives. You might find that Itanium achieves some very respectable TPC-C numbers despite the low SPECint scores This statement intrigues me. Is there some reason that database transaction code is easily parallelized into 3 instructions per cycle, but SPECint isn't? Petz