FAH: Linux vs Windoze clients

The Windows and Linux cores are all made from the same code. In fact, they're all compiled on the same machine running Linux using GCC.

The speed difference mainly comes from the fact that the Linux clients tend to get more experimental things such as cores that are not available to the other platforms yet.

The performance difference is probably less than 5% and could probably be easily overcome by running the Linux client on a minimal, F@H only, install. In the sticky I reccomend the Linux client for machines that run Linux and I stand by that assertion. :)

-BTI

P.S. The WUs available to OS X clients have gotten much better of late. I used to see ~70 ppd from my 1.8 G5s now I see around 200. :)

I haven't seen a Tinker in some time.
 
While we are on the subject of clients, should I be using 5.02 or 5.04 beta? At present I'm running 02 everywhere. :o
 
It doesn't matter. 504 has a few nice things that won't matter if you've already configured the client, mainly you can tell it to lie about how much memory your system has and you can set -advmethods without hacking the registry.
 
Last edited:
OK I think that has answered all my questions - Oh except I have a centrino lappie I think it is 1.7ghz with what I believe is a Pentium M in it, and yet it doesn't seem that great at crunching - I thought the Mobiles were supposed to be good? It has a 1mb cache? Was running linux on it though which was part of my reasoning behind the original question here..
 
Well my PM 1.5 running Ubuntu 5.10 tore through Wus at about 200 ppd average. It was able to hit 500+ ppd on 364 pointers.

the answer might lie in the fact that there are a lot of crap WUs out there at the moment. It might also be some other problem like another process stealing cycles or that you're getting crappy FahCore_7a.exe WUs.
 
Well, it is currently running its first WU (should that make a difference?) a p2409_Ribo_lysine_sidechain, and it must have been running it for about 24hrs on and off in total, and it has only reached 2%... didn't seem right to me, so stopped it. Process definately running at 100% though in top.
 
Hmmm just removed cpudyn and it seems to be doing a bit better now. Odd though since I didn't think cpudyn would do anything when the power is plugged in.
 
Back
Top Bottom