what can i do at best to optimize:EINSTEIN program/boinc.

DanNeely
DanNeely
Joined: 4 Sep 05
Posts: 1364
Credit: 3562358667
RAC: 162

Yonah (90.3mm^2) is

Yonah (90.3mm^2) is significantly smaller than merom (143mm^2) which makes them cheaper to produce. In addition to merom's size reducing the maximum number of chips per wafer by 58%, the larger die size will result in a larger percentage of manufacturing defects due to bad silcon. (3 bad spots on an area coresponding to 100merom's is 3% failure, with 158 yonah's it's only 2%).

Yonah is also lower power 31W max vs 34-35 for merom with standard voltage models, 15 vs 17 for low voltage models. 9 vs 10 for ultralow voltage.

Finally, and probably most importantly, upgrading thier existing yonah plants to merom would cost lots of money. Until sales on yonah become low enough to justify it, or they need to convert plants to begin making penryn (core2 45nm) chips converting them would only serve to reduce profits or make the entire line of intel hardware more expensive.

th3
th3
Joined: 24 Aug 06
Posts: 208
Credit: 2208434
RAC: 0

Die size is only 111 or

Die size is only 111 or 113mm² for 2MB versions, much less than Yonah measured in performance per mm² =)

DanNeely
DanNeely
Joined: 4 Sep 05
Posts: 1364
Credit: 3562358667
RAC: 162

Unless you expect the average

Unless you expect the average consumer to buy 20% fewer merons laptops than they do yonah based machines the yonah's are still more efficient to manufacture. For what 95% of consumers are doing both chips are equally fast since niether thier mp3 player, dvd player, nor browser will come close to maxing the CPU load.

th3
th3
Joined: 24 Aug 06
Posts: 208
Credit: 2208434
RAC: 0

Thought i give a quick update

Thought i give a quick update here on the Pentium Dual Core E2140, i now got a box up and running E@H and its performing better than expected. Overclocked to 2.93GHz at stock volts it finishes 452 credit WUs in less than 61,000 seconds.
Results here

Then i compared it with this baby , which i think is a Tigerton? Belongs to Team Intel Corp, so i sure believe so.

Seconds per credit:
~60600/452.35 = ~134 (E2140 2.93GHz)
~85700/634.34 = ~135 (Tigerton 2.93GHz)

So maybe the Einstein app dont care much for large L2 cache. In addition E21x0 got only 4-way associativity on L2.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.