• 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.

Only using 50% of CPU??

T

Time2Kill

[H]ard|Gawd
2FA
Joined
Jul 10, 2005
Messages
1,240
I just started folding today, got everything going great on two computers (FX-53 and P4 2.4ghz), but I'm having problems on another one (2.6ghz). Task manager says it is only using 50% of the CPU. The other 50% is going to System Idle and I can't figure out why. I'm using the no nonsense winxp client. The machine is a Dell 9100 if it matters. Anybody have any ideas?
 
http://www.hardfolding.com/index.php?go=52

Those are the instructions for setting up multiple sessions. Stanford prefers you not run F@H this way as they prefer the results as soon as possible. If you do choose to run multiple clients on an HT P4, you'll only see about a 15% increase in the amount of work done.

 
Simple solution.

Get two boxen running at 50%, add them up and you have 100%!

It's running wide open, trust me.
 
I think it may be the only way that the Task List knows how to show it. Since windows sees the HT as an additional CPU, so in theory at least, with 50% its really running 100%, and then another Client on the HT you get the 15% more production.. so it would be 115% maybe?
 
KodiakStar said:
I think it may be the only way that the Task List knows how to show it. Since windows sees the HT as an additional CPU, so in theory at least, with 50% its really running 100%, and then another Client on the HT you get the 15% more production.. so it would be 115% maybe?
Click to expand...

This is what I had assumed... because windows sees it as two CPU's, its saying that one CPU running at 100% = 50% of total CPU power. To bad that doesnt mean two instances at 100% for each CPU = 200% total. That would be sweet... :D
 
Viper87227 said:
I have actually wondered on this myself... It shows 50%, but is it still using 100%... not include HT?
Click to expand...

Taking an extra minute to think about it. what you could do is benchmark it..

one with HT off, with a WU
another with HT on (same WU or as close as you can get...) but only one client running...

then compare the times.

Or... A HT Supported chip @ x.xGhz, and a non-HT chip @ the same Ghz...
 
KodiakStar said:
Taking an extra minute to think about it. what you could do is benchmark it..

one with HT off, with a WU
another with HT on (same WU or as close as you can get...) but only one client running...

then compare the times.

Or... A HT Supported chip @ x.xGhz, and a non-HT chip @ the same Ghz...
Click to expand...

How would I get the same WU again? Not like you can chose them? As far as testing two different chips...no dice there. At home the only HT chip ive got is this 2.4C. Everything else is cellys. The cache, while I kknow means little when it comes to F@H, could throw results off...
 
An HT P4 still uses the same pipeline to execute the instructions, however, with HT on ,two different threads can occupy various stages. The L2 cache is also split 50-50 between the cores. This explains why the performance boost is nowhere near the 90%+ that true SMP will realize, as a pipeline flush on the first thread will affect the execution of the second thread, etc. On a true dual CPU or dual core system, the threads proceed entirely independently. Probably the best advantage of having HT is that the impact of your general use will have a smaller impact on Folding - if a renegade flash ad, application hang, etc hijacks all the CPU time it can get, then folding would halt on a Celeron/Athlon XP/Athlon 64, but it would keep going on a HT P4 :cool:
 
Stop being so damn technical :)

But good job :D

The only intels I have my hands on are from 1999 :p

Oh well.
 
Simply put, HT is good for light multitasking, makes the experience alot smoother, but it's all in the details of how the CPU fools things into seeing a 2nd processor. Still full tilt, just not as reported. It is a saviour when you have an app go on a bender though.

That's why I like my HT box at work, it shreds through QMD's, but with the HT, everything else is as smooth as it gets. :cool:

 
Thanks for clearing that up!

On another note, should I be changing the machine ID on each machine? Right now each PC has a machine ID of 1.
 
Time2Kill said:
Thanks for clearing that up!

On another note, should I be changing the machine ID on each machine? Right now each PC has a machine ID of 1.
Click to expand...

You only need change machine ID when you're running more than one session on a PC. Otherwise sticking with the default 1 for separate machines is just fine. :D


 
Each PC can have the same ID = 1..

However if you run two or more clients on the same computer, each client should be 1, 2, 3 etc
 
Ok guys I have the same issue with my four processor server.

I can only get the processors to fold at 20% not at 50 or above percentage and I am running a form of 2003 server.

I better ask this now or I'll never get an answer to it.
 
Do any of them have HT?

If not you would think each would run at 25% (x4 = 100)

Is there anything else listed that is using up any CPU Cycles?
 
No my processors are not Ht, they are the old P3 Xeon processors.

I can set the affinity to the folding to the processors, but looking at the task manager it shows four graphs for processors and all four never get above 50%.
 
Do you have to manually assign the processor affinity? I was under the impression that having a separate folder for each processor, using a different machineID, and using the -local flag was all you needed to run on a multiproc system.
 
I vote you just keep adding clients till it says 100%.

Some OS's, things get wierd on. I've done a 2K3 server with one instance where it says 50% with one core running with local flag, another gou did one and it said 100% on one ore and 0 on the next.
 
Mohonri said:
Do you have to manually assign the processor affinity? I was under the impression that having a separate folder for each processor, using a different machineID, and using the -local flag was all you needed to run on a multiproc system.
Click to expand...

That's the way it's intended to work, but like marty said, there's times where it can misbehave. I would suggest maybe asking at the official forums as well.


 
marty9876 said:
I vote you just keep adding clients till it says 100%.

Some OS's, things get wierd on. I've done a 2K3 server with one instance where it says 50% with one core running with local flag, another gou did one and it said 100% on one ore and 0 on the next.
Click to expand...

wholly crap I can't type
 
Viper87227 said:
How would I get the same WU again? Not like you can chose them? As far as testing two different chips...no dice there. At home the only HT chip ive got is this 2.4C. Everything else is cellys. The cache, while I kknow means little when it comes to F@H, could throw results off...
Click to expand...

Here's how I benchmarked Folding when I was running a bunch of tests to prove that oc'd ram on a64's does little:

Have a wu folding.....which you should :) Shutdown fah, go to windows explorer/my computer and copy your folding folder to another folder/drive,whereever. Start fah back up and let it crunch 5-10 frames depending on if it's a little 69 pointer or a 600. Write down all the times for each completed frame. Shut down fah, then copy/move the contents of the backup folding folder into the original folder. Repeat process.....compare times :D

My first batch of (uncontrolled) tests resulted in a 1sec/1mhz decrease in crunch time. But by uncontrolled , I was browsing the forums :D , so that probably skewed results. The next batch of time trials acctually showed the slower ram being 1-2 faster in crunch speed than the oc'd ram (200mhz stock / 243mhz oc'd) These tests were run with no background programs running, and me sitting idly on the couch wondering wtf to do for an hour with no
computer to use for 30 minutes.
 
You must log in or register to reply here.
Back
Top