• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

P4 Hyper Threading, One or 2 clients?

R

Rockaholic

Limp Gawd
Joined
Jul 7, 2004
Messages
138
I notice when running the console version it olny uses 50% cpu, when set to 100% im guessin this has something to do with the HT feature, anyone done any benchmarking to see if its better to run 2 clients or 1?

 
You only get about a 15% performance gain from HT. It will net you a few more points but Stanford doesn't like it as it effectivly makes both instances use about 66% of teh CPU time so that the Work Units are turned in slower.

 
Tormond said:
You only get about a 15% performance gain from HT. It will net you a few more points but Stanford doesn't like it as it effectivly makes both instances use about 66% of teh CPU time so that the Work Units are turned in slower.

Click to expand...


Well said. Bottem line is just run one client. Science > Points
 
I have mentioned this before. You can run two clients, get a points boost and not slow down regular or bigpacket WU's all that much.

Run two clients, the first with bigpackets if you want to run them and at 100% CPU usage. Then you setup a second client to run timeless units only. This one you set to about 15% CPU usage. It does not slow down regular units much (at least in my limited testing) so Stanford gets them back at the normal times. And the timeless units get you some extra points without much strain on your RAM.

That is how I run two clients on my work P4 w/ Hyperthreading.

 
I've really never had much luck with 2 clients. With the right WU types, yea oyu can get 15% more points. With the wrong WU types I had -30% PPD. Certianly with QMD's run one instance.

As said before, in it for the science run one. In it for the points, run two. Stanford if grateful for every WU folded...

Plus I had some nasty stuck WU's w/o using the -local switch when I started out.
 
marty9876 said:
I've really never had much luck with 2 clients. With the right WU types, yea oyu can get 15% more points. With the wrong WU types I had -30% PPD. Certianly with QMD's run one instance.

As said before, in it for the science run one. In it for the points, run two. Stanford if grateful for every WU folded...

Plus I had some nasty stuck WU's w/o using the -local switch when I started out.
Click to expand...

Oh yeah, if you get QMD's definately run only 1 instance per physical processor (dont know about the new dual-cores though). My dual xeon running 4 QMDs yas taking about 1 hour per frame per instance. Now that its dropped to 2 QMDs (1 per processor) they are down to 25-30 minutes at the most per frame.
 
You must log in or register to reply here.
Back
Top