• 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 with ht ??

8

84gthatch

[H]ard|Gawd
Joined
Jan 7, 2004
Messages
1,207
I just noticed this between 2 different machines i just installed F@H on which have HT
they never seem to go over 50% on the f@h core?? all the other machines p4 or not run straight up at a 100%

is it that i just have to run 2 instances or what?
 
Due to HT, Windows only reports 50% on both procs, but rest assured that it is being used at 100%. You can run 2 instances on HT enabled PC but one instance is also fine. If the 50% really bothers you, there is an option in the BIOS to disable HT.

 
When I first had my P4 2.8 with windows 2000, I was surprised too.

What happens is that, at 50%, the physical processor is is maxing out, but there is some extra power not being used.

With HT on, you can run Two "console" versions of F@H. This will appear to put the computer at 100% in the task manager, but most people say that you can expect around a 30% increase in Points Per Day. My computer with two consoles puts out about 30% more points per day.

The only thing to note is that each console version must be in a diffrent directory with a diffrent machine ID number (that is set in the configuration of F@H).

~Hope this helps
 
I run 2 instances of F@H on all my P4 HT systems (4 2.8ghz P4 w/HT = 8 instances of F@H). Not sure if it offers a tremendous benefit or not...but it gives me peace of mind that they're being used 100% capacity at all times ;)
 
it is running xp pro
i thought about it and saw a little something at the stanford website about downloading the console version each time you want to run an instance so yes it does make me feel better seeing the task manager say 100 % so i now have it tearing through 2 gromacs;)

1 more quick question how do you get it so that the console version starts as a service so i dont have to start it each time?
 
you can always delete the config file and then run the cmd line version. When it sees no config file, it asks you a list of questions and one of them is if you want it to start as a service in XP.

Of course...someone else may have a better answer. I'm not sure if this results in unfinished units or what...
 
You can also add the -config switch to the console version's shortcut, and then you can reconfigure F@H so that it starts up as a service.

As for running 1 or 2 instances on an HT proc - you'll get answers for each. I run only one instance on my HT comp. HT just makes the proc switch really fast between concurrent processes (or something similar to simulate two procs), and when I run two instances, It takes me about twice as long to finish my QMD cores (38 minutes/frame vs. 18-20). So to get the units to Stanford faster, I just run one, that way it gets the full processor's power.
 
Hmm,

My only HThreading PC is a dual-Xenon.

On those 400-step protiens, 2 instances will step in 12 minutes, while 4 instances will step in 15 minutes. Per hour, this is 2 instances = 10 steps (5 Steps x 2 Instances), vs. 4 instances = 16 step (4 steps x 4 Instances).

AFAIK, the Xenon's shouldn't have an advantage over a regular P4, but rather than argue w/the results, I just fold 4 instances all of the time.
 
Stanford has stated on many occasions that they would rather you not run two instances of folding on an HT machine. Their (Stanford’s) goal is to get work back as fast as possible, two instances works against this process.

The overall gain is more like 15% on most Work Units.
 
BillR said:
Stanford has stated on many occasions that they would rather you not run two instances of folding on an HT machine. Their (Stanford’s) goal is to get work back as fast as possible, two instances works against this process.

The overall gain is more like 15% on most Work Units.
Click to expand...

True, Stanford also states they are happy with any and all results returned.

If your in it for the points, run two instances. If your in this for the science, run one. I've played with running one/two- results vary. If you throw a QMD into the equation, running two slows things way, way down. Some people run one large WU, and one timeless (OC-AMD) on Xeons ect.

Personally, I just run one instance. Stanford does need the results back quickly as the next WU's are based on the previous WU's. However they must not be too concerned with this as they send 600 pointers to P2 450 mzh boxen. Yea, I turned on the large WU option, but Stanford cal still limit which WU's they send to what boxen.
 
I do something a little bit different on my P4 2.8 Ghz Preshott work system w/ 512 MB RAM. It's the only P4 I've ever folded on since all my stuff is AMD (except for a dually PII 266 that needs repairs).

I run one instance with bigpackets turned on that is allowed 100% usage in the console settings. I run a second instance of timeless units that I have set to run 15% usage in the console. This way, the main instance still runs almost as fast as it would by itself and I still get some extra points from the timeless WUs. Yes, it takes forever to get the timeless done since they have 241 and 239 point timeless tinkers now. It's looking to take 18.5 days to finish the 239 pointer it's chewing on on the second instance.

This way I have the best of both worlds in my opinion. Stanford still gets the main WUs done in about the same time while I still get some extra points from the timeless tinkers.

I have set the instance running timeless to 20% in the console which slowed the main instance down a good bit so it was set back to 15%. I have not done exhaustive testing which would be almost impossible since this is my work machine. These are just some quick results I have come up with.

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