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: Scot who wrote (43554)	12/16/1998 5:52:00 PM
From: Scumbria	Respond to of 1585098

Scot, Re: "More Katmai ...." Intel Katmai 450MHz - 512K L2 Cache Benchmarks Wintune 98 Integer - 1392.243 MIPS FPU - 727.361 MFLOPS I'm having a very hard time believing that Katmai averages >3 integer instructions per clock, and 2 FPU instructions per clock ;^) How do they do that with only 3 dispatch pipes? ;^)) Scumbria

To: Scot who wrote (43554)	12/16/1998 5:55:00 PM
From: kash johal	Respond to of 1585098

Scot, Interesting that the 450Mhz Katmai is about the performance of a overclocked 300a Celery for current apps. I can't wait for those killer apps in March 99. Regards, Kash

To: Scot who wrote (43554)	12/16/1998 5:55:00 PM
From: Tenchusatsu	Read Replies (3) \| Respond to of 1585098

Thanks for the link to the benchmarks, Scot. It seems that Katmai, even with its larger L1 cache, isn't providing a performance benefit for non-KNI-enabled applications, which should be just about everything that's out there right now. Intel was wise to keep very silent on its Katmai efforts, or else they would have caused people to put off their holiday purchases for a future processor that isn't going to do much for existing applications. The non-relevance of Katmai's larger L1 cache is pretty interesting to me. Does that mean that the oversized 128K L1 cache on the AMD K7 isn't going to show any benefits, either? Tenchusatsu

To: Scot who wrote (43554)	12/16/1998 6:30:00 PM
From: Petz	Respond to of 1585098

K6-2-450 vs. K6-3-450 vs. Katmai 450 on Winbench '98. The benchmark results at dailydemo.com finally give us some common ground to compare these three CPU's. We have K6-2-400 (CXT core) benchmark results from Anand (12/2/98) which gave a benchmark score of 26.8 with 64M RAM and a relatively slow hard disk (Western Digital AC35100). With a very fast hard disk and 128M as used by tomshardware.com, this 26.8 would improve to 30.5. Upping the clock speed to 450 MHz gives benchmark scores of 27.5 (slow 64M) to 31.5 (fast 128M). The K6-3 scores below are based on up-ing the L2 cache speed to 450 MHz and reducing its size to 256K in the CPU model, which estimates the number of L1 cache misses and L2 cache misses and the penalty times for these events. In the table below, actual data is in bold italic and extrapolations based on the model are in normal typeface. The "dailynews" benchmarks use unknown hardware, but I suspect that it is with 128M RAM. Business Winbench 98 Scores .....................Hardware=.................Hardware=...............Hardware= CPU TYPE.....64M Slow Disk.........128M Fast Disk........Unknown K6-2-400........26.8...........................30.5 K6-2-450........27.5...........................31.5 K6-3-450........31.2...........................36.3 Katmai-450.................................................................34.3 In other words, if I am right that the dailydemo benchmarks were run with 128M memory, the K6-3-450 in January will be as fast or faster than the Katmai-450 in business Winstone 98. I wouldn't count on Winstone 99 for bailing out Intel. It favors Intel for two possible reasons -- poor DMA disk drivers from VIA and/or heavy reliance on L2 cache speed. If the former, the Intel advantage in Winstone 99 is likely to be short-lived. If the latter, the K6-3 with 450 MHz L2 cache will smoke the Katmai. Petz