EoN optimization thread

[jasong]

What's the best bang for buck in Eon?

Is RAM especially important? What about cache size?

Which is the better Eon cruncher in general, AMD or Intel?

Bartons are supposed to be fabulous at Find-a-Drug, what chip kicks butt in Eon?

(These questions are to stir the pot, btw, if I need personalized help, I'll post a new thread.)

[jasong]

Come ON!!!

Surely, SOMEONE's done some experimentation.

Also, I'm still waiting for the upgrade that allows priority setting changes. I WILL run this project when that happens.
----------------------------------
Edit: Proth_sieve at Low and client.exe at Below_Normal seems to do the trick.

Still waiting for a benchmark utility.
--------------------------------
Edit 2(2-3 hours later): DCers love DCing. And some of them are willing to put in significant cash to get high scores in a project. But they need to know where to focus. Especially in a project where the optimal specifications for getting points could change. Could that be this project? Possibly. A benchmark utility and a thread that discusses what client.exe "likes" when it comes to different scenarios.

I made a couple requests, Priority management and a benchmark utility.

The first one might not be wanted by a lot of people, but might attract people who are anal about maxing out their cpus.

The second was a benchmark utility, something I'm interested in, but don't know how difficult it would be to implement. I know some people would absolutely LOVE IT if you could give them the ability to tweak their machines to optimize their output. I don't know how yours might work, if you had one, but I'm guessing someone there is familiar with things like how often RAM might be accessed, how much L1 and L2 cache is utilized(or useful) and things like that. Would it be possible to determine, within each individual process of the computer, the amount of clock cycles between RAM accesses and whether or not there's any cpu time spent waiting for RAM to respond? If you could give some statistics, I'm sure someone could come up with either a formula or a randomizer program to simulate waiting times. That would be fabulous.