My memory timings are as tight as can be, stably.
![Very Happy :D](./images/smilies/icon_biggrin.gif)
How much faster is the text based client? Is it?
Thanks!
EDIT: Does an increased molecular draw rate improve performance? I cant find anything on the main F@H website.
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
Yeah, I just noticed that. My 1.4GHz AMD can process almost as much information per second as your 2.2GHz P4 (dont know if it's a Northwood though)... Cool!rpc180 wrote:I don't think its much faster, estimated CPU use time for the graphics portion is supposedly between 1-2% of overall and close to nothing when its minimized.
The difference is that with the text client you can add some of the tags that may force you to optimize methods in case the graphical client decides that it doesn't want to utilize those same methods for whatever reason...
I know you can add the tags to the text client, but some have suggested that you can simply add the same tags to the shortcut for the graphical client and thereby eliminating a major speed factor between the two. In terms of raw efficiency, I don't believe there is more than a marginal difference.
The two most commonly used switches however, are -forceasm and
-advmethods. These are usually used to force CPU algorythms that are optimized for higher end P4's and AMD chips. It also allows a preference for Gromac WU which are optimized for P4 chips. AMD chips work better overall due to their higher processes per clock cycle.
Actually, Athlons have THREE floating point units!Zhentar wrote: I think Athlons have more of an advantage though- I think these things are pretty heavy on floating point calculations, the P4 has 1 FPU, while the Athlon has 2... leads to even larger performance difference.
*cackles*NeilBlanchard wrote:Hello:
Actually, Athlons have THREE floating point units!Zhentar wrote: I think Athlons have more of an advantage though- I think these things are pretty heavy on floating point calculations, the P4 has 1 FPU, while the Athlon has 2... leads to even larger performance difference.
I am running all AMD's and the only tweak I have done is that I've clicked the "Slightly Higher" option on the Advanced Tab. This raised my CPU temp by a few °C so it must be helping a bit (in theory).DaShiv wrote:(Unless some of the top folders have any secrets they'd like to share with us, aside from "brute force"...) Since you're using an AMD though, you've already got a leg up on some of us here.
Happy folding!
Thanks for the correction on that. A supplementary question, is it worth using -advmethods on an Athlon then? Or at least my Thunderbird which doesn't have SSE?DaShiv wrote:Pentiums do better with Gromacs, and terribly with Tinker. Gromacs is newer, and takes better advantage of SSE.
Just checked queue.dat - the current unit is Tinker and 1 out of the previous 9 (total 2 in 10).DaShiv wrote:Have you tried tallying how many Tinkers vs. Gromacs you've gotten?