• 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.

Interesting way to get less PPD?

M

marty9876

[H]ard|DCer of the Month - February 2006
Joined
Jun 11, 2003
Messages
4,906
Ok, per this thread over at OC Forums by ChasR

For the lazy-
I set instance 1 -adv/big
Instance 2 big
Instance 3 big, but disable the service most of the time
Instance 4 client type 2.

Mostly I just let it run. However right now you can't always count getting a big WU, so if I don't get one on instance 2, I'll put -oneunit in the registry for instance 2 and enable instance 3 to try and pick up a big WU. There will be a few days where you don't do any better than 2 QMDs but for 9 day out of 10 you'll be getting 100ppd more than two QMDs.

When running 3 instances there may be some benefit to setting the system environment variable omp_num_threads=4. Doing so will let the semi SMP aware QMD core use 2 threads while the other instances get one thread each. There is a slight (3% maybe a bit more) boost in doing this.

I don't have any p4s folding two instances. There just isn't much to be gained. My experimenting came up with a 5% boost folding two QMDs compared to 1 QMD, One QMD and one RIBO (600pt) got 2% fewer ppd than one QMD. One QMD and a regular Gromac got a lot fewer ppd. I'm sure one QMD and one DGromac would out point one QMD by a significant but haven't gotten a dgromac to test it on. I guess I'll have to sneaker net one off one of my XPs to see what happens.
Click to expand...

How to set per ChasR- start,control panel,system,advanced,environment variables

I used variable omp_num_threads=2 and here is what I got-

This boxen is just a P4 3.2 @ 3.something doing QMD's 24/7. Only one instance running. Day in day out when I leave the machine alone per EM3 it does 536 PPD.

The task manager always shows 50% CPU usage ad all HT intel chips do.

However...

When I did this change, the CPU would be at 50% most of the time, but spike every ~15 seconds for about 2-3 seconds for a total usage of 97-99%. Per the processes it's the folding client spiking it.

It's kinda neat seeing the graphs peg (I know, it's using 100% even thou it only reports 50% under previous normal config).

However--- I let it run this way all day, got home and was doing 513 PPD per EM3. :confused:

Other problem- I was going to post neat screen snap shots showing the usage, took the ones without this change, then was all set to take them with the changes and paint won't load... :confused: :confused:

I dunno. Disclaimer: me SMP moron
 
I think 536->513 (4.4% difference) isn't statistically valid. That's not enough difference to matter, IOW. Granted, it may hold true, but I'm not buying it 'till it's been repeated a few times.

 
You must log in or register to reply here.
Back
Top