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: Dan3 who wrote (52694)	8/27/2001 10:04:12 PM
From: jcholewa	Read Replies (1) \| Respond to of 275872

> The rule of thumb is that if you don't quadruple cache, > don't bother. Could you elaborate on where you heard this? The closest that I've ever heard to this is the concept that the ideal ratio of L2 cache size to L1 cache size in terms of performance and price is 4:1. But the thing that you specifically say here is something that I've never heard before. -JC

To: Dan3 who wrote (52694)	8/27/2001 10:36:37 PM
From: Tenchusatsu	Read Replies (3) \| Respond to of 275872

Dan, <The rule of thumb is that if you don't quadruple cache, don't bother. Taking P4 from 256K to 1 megabyte of cache would be expected to increase performance by 5%. Taking it to 512K would make a 2GHZ perform like a 2.05GHZ> Are you so sure about this, Dan? First, I've never heard of the rule-of-thumb that you just stated. The only rule-of-thumb that comes even close is the one which governs the relative sizes of two-level caches (4:1 ratio). Second, where do you get the notion that a P4 going from 256K to 1M would only increase performance by 5%? After all, a 2.0 GHz P4 is running at a 20:1 processor-to-bus clock ratio. Decreasing the L2 miss rate by even a small amount will have a marked improvement in performance. And third, you'll be unpleasantly surprised at the per-clock performance improvement Northwood has over Willamette. But hey, no one really needs that sort of performance until AMD can sell it for under $100, right? Tenchusatsu

To: Dan3 who wrote (52694)	8/28/2001 4:01:02 AM
From: Joe NYC	Read Replies (2) \| Respond to of 275872

Dan, The rule of thumb is that if you don't quadruple cache, don't bother. Taking P4 from 256K to 1 megabyte of cache would be expected to increase performance by 5%. Taking it to 512K would make a 2GHZ perform like a 2.05GHZ OTOH, P4 crashes so hard when it encounters a stall, that it might be more sensitive to increases in cache size than most processors, so there may well be more of an improvement. On a broad range of applications, you may be right, but lately, everything seems to be turning around performance in benchmarks, and I can think of a few that will see a dramatic impact. One I can think of immediately is SETI@Home. I believe the work unit is about 300K, and it will just fly when run on Northwood processor. Expect to hear A LOT about SETI a few months from now. But IMO, the real goal of the 512K L2 is to soak up extra fab capacity. Intel has a huge surplus of it, and if Intel can somehow force the mainstream .13u processor to be let's say 170 mm^2 rather than 100 mm^2, it will be a victory for Intel, since AMD's capacity will be capped, while Intel will get the rest of the market by default. That's what I would do. Joe