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: Ali Chen who wrote (74138)	3/10/2002 5:36:29 AM
From: pgerassi	Read Replies (2) \| Respond to of 275872

Dear Ali: 2.0 to 2.2 is not a large sample difference. The noise alone makes projections to 3.0GHz very difficult. Small reporting errors in either sample could make huge differences in the projection. A simple MB swap or BIOS change could alter the results by 5% or more between the samples swamping any meaningful projection by +/- 50% at 3GHz. This is a classic error that is commonly made. And you are assuming linearity rather than a curve in that projection. At least three points are required and more are desired. That is the other classical error. I tried five different 2000 to 2200 P4 SPECint estimates for a P4 2800, they were 903, 850, 915, 870 and 930. A variation of 80 units around 892 or about a 9% difference. Yet when I looked at lower speed estimates of known higher speed CPUs, the estimates were always high. And that estimator never looked at what the estimate would be if it used a second or third order estimator. Those estimates come 25% lower and much closer to reality when used with Williamette (the only one we have data enough for the 4 or 5 points required) but, they are still higher on average. All in all, those estimates of yours are probably high. To get to the real numbers take the square root of the ratios (1.32 ^ 0.5 = 1.148 or about 15% not so good wrt a 50% clock increase). This is why many want the faster FSB and lower latency memory for P4. Pete