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: Petz who wrote (87763)	1/18/2000 12:26:00 AM
From: Bill Jackson	Respond to of 1572100

Petz, As you say a few %, this is a game of small increments and really shows how CPU speeds are plateauing in terms of benchmarks as the mundane I/O drags everything down. Well since the real world is where we operate most businesses that will show up with businesses accepting lower price, and performance solutions leaving the work stations and other demand users to drive the fastest products. A generally leaner prospect, what with Via, Intel as well as AMD fielding commodity CPUs for this segment. Looking at the BMs for the P-III 733 versus the 750 Athlon....there are a few BM where the Athlon will regain the lead and some where it will not. SSE seems to be the factor? Can AMD emulate the SSE like they did with the MMX (3D Ex) or will that stay an Intel alone area? Of course AMD can make their own extensions and provide information to programmers so that the programs can sell with slight variances depending on where they are run and the installation program selects the SSE variation or the AMD variation to install. A fork in the road? will it lead to two roads or will the Intel one become dominant because of SSE and AMD wither? Bill

To: Petz who wrote (87763)	1/18/2000 1:58:00 AM
From: Petz	Read Replies (1) \| Respond to of 1572100

Here's a quick and dirty calculation of benefits of full speed L2 for an Athlon. From tech-report.com, look at the ZD Winbench 99 CPUMark 99 score for 3 different Athlon 800 CPU scores. They all use the same Athlon, only changing the cache divisor CacheRatio...CacheSpeed...CPUMark99...CPUMark/MHz 0.33...............267....................65.6 0.40...............320....................67.0...................0.0263 0.50...............400....................70.2...................0.0400 Average increase in CPUMark99 per MHz of cache speed: 0.042 Predicted CPUMark99 at 1:1 L2 Cache Ratio (800 MHz): 87 % improvement vs. 2/5 cache ratio: 29.9% If we do the analysis based on cache cycle time instead of cache MHz, the table looks like this (Cache cycle time in ns.) CacheCycle...CacheSpeed...CPUMark99...CPUMark/ns 3.75................267....................65.6 3.125..............320....................67.0...............-2.24 2.5..................400...................70.2...............-5.12 Average increase in CPUMark99 per ns. reduction in L2 cache cycle time: 3.68 Predicted CPUMark99 for full speed cache (1.25 ns cycle time): 74.8 11.6% improvement vs. 2/5 cache multiplier) The three point data seems to fit the more optimistic prediction of a 30% performance gain better than the more pessimistic 12% prediction based on access time. Either way its worth at least 100 MHz worth of improved performance, and probably more since on-die cache latency will be lower than off-die. Petz