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 (52699)	8/28/2001 1:01:47 AM
From: Elmer	Read Replies (1) \| Respond to of 275872

Ten, regarding your response to Dan's claim: 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 The absurdity of this claim can be easily seen by looking at SPEC scores for 700MHz PIII 256K L2 compared to 700MHz PIII Xeon with 2MB L2. Same frequency yet the Xeon shows a 40% improvement in integer scores with the addition of the larger L2. specbench.org specbench.org

To: Tenchusatsu who wrote (52699)	8/28/2001 9:26:28 AM
From: fyodor_	Read Replies (1) \| Respond to of 275872

Tench: And third, you'll [Dan3] be unpleasantly surprised at the per-clock performance improvement Northwood has over Willamette. Didn't Intel say they would disable "Jackson Technology" in the consumer version of the .13&#181 P4? Is anything else (publicly) on the table, or are you referring to something non-public? ;-) -fyo

To: Tenchusatsu who wrote (52699)	8/28/2001 12:01:22 PM
From: Saturn V	Read Replies (2) \| Respond to of 275872

Ref < Dan 3 and... the only rule-of-thumb that comes even close is the one which governs the relative sizes of two-level caches (4:1 ratio).> The rule of thumb is that L2 must be at least 4X L1, and L3 must at least 4x the size of L2. The miss rate of the cache is highly bench mark specific. However a "general rule of thumb " is that averaged over all benchmarks, the cache miss rate falls inversely as the square root of the cache size. Thus a doubling of cache size will lower the cache miss rate by 30%. Thus the miss rate has a law of diminishing marginal returns wrt to cache size. Dan3 has obviously stretched this fact to suit his view of the world. I concur with Elmer that a L3 does not makes sense. The difference in the speed of L2 and L3 is not enough to warrant a new level of cache, and it makes sense to increase L2 only. And I do not know why Merced used a L3.