Folding hardware question with X2s
edcentric
near Milwaukee, Wisconsin Icrontian
If you are folding on an X2, does each core run one instance, or is the load shared?
I have an X2 box (well it isn't in a case, but you know) that has one instance that occasionally short runs units. Always instance 1. I have erased everything except the .exe and .cfg and reloded.
Instance 2 has never barffed a unit.
The box is mildly OCed, 220x10
Any thoughts????
I have an X2 box (well it isn't in a case, but you know) that has one instance that occasionally short runs units. Always instance 1. I have erased everything except the .exe and .cfg and reloded.
Instance 2 has never barffed a unit.
The box is mildly OCed, 220x10
Any thoughts????
0
Comments
hope that helps
If I run one instance, will one core be loaded 100% and the other 0%? Or will they each run 50%?
Open Task Manager and click on the processes tab. Right-click on the "FAHxxx-Console.exe" Image Name, then left-click on the "Set Affinity..." item. It should have a processor affinity list and your two cores (identified as CPUs) itemized. If they are both checked, then the system can share between them as it wants to. You can tell it to have more affinity for one processor by unchecking one.
Sound slike the best setup is to have two installations of FAH, set up to run as a service. One will have machine ID of 1 in the client.cfg, the other will have machine ID #2. This makes sure FAH runs OK.
Then you can do what he said with task manager, assign the FAH install with machine ID 1 to one core, and ID 2 to the other. In this way, you'll essentially be set up the same way as a dual-core machine. (And that's what you want).
In this setup, both cores will run at 100%, unless some process or another needs a core. then FAH will throttle back on that core, but leave the other one alone. There's a pretty good FAQ over at Stanford's site regarding dual-core systems, it appears that it would apply to dual-core systems as well.
edit: I'm just interpreting from what i read at Stanford's site this morning. Of course, you could always do a dual-core, dual-CPU system and have 4 instances of FAH... <- (that's the heatsink)
yes, but is it one instance of FAH per processor or per core? I know it's not per virtual processor, as Hyperthreaded processors see no gain...but dual-core might benefit the same as a dual-processor?
sweet.
That was my guess, but the Stanford FAQ hasn't been updated yet to account for dual cores, it just talks about the possibilities with 2 processors, so I wasn't sure.
I was talking more about the benefits of multiple instances of FAH, not so much the procedure of getting them going...
Okay then- would you indulge me in a little experiment?
Open Task Manager on your dual core and click on the Processes tab. Select one of the FAHxxx-Console.exe processes and End the process. Now click on the Performance tab- What do you see?
On my PC I see a seemingly random distribution of the remaning FAH task between the cores. When my Affinity is set to both cores (which it is by default on mine)- this is how it behaves. If you add up the percent execution of both cores on each frame, the total execution time is at least 100%.
Now kill your other FAH instance. Restart one and set the Affinity of this one to only whatever core you want and then click back to the Performance tab. What do you see? On mine all of the execution is on one core.
Don't ask me all of why and how the OS (XP) distributes it like this- I assume it is to service interrupts and other processes of each one on a priority basis. It is, however, how it seems to behave on my machine.
as I have never tried, anytime I have had a PC that can support more than 1 instance of F@H I ran 2 instances of F@H
I'm seeing the same thing on a 3.08 P4 w/HT. seems odd. task manager shows FAH pegged at 50%...
~FA
If changing the affinity between two jobs, and one core always stays hotter then the other, I assume its because the IHS is not 100 % connected to the core.
One core transfers heat from the other core to the IHS ?
I also noticed that when running only one WU, the temp differs a lot when executing it on CPU 1 then on CPU 2.
It makes no sense unless there is a bad connection between the core and the IHS.
When I'm gonna install my Opty175, I will know if it behaves exactly like my Opty165.
In the task manager the "CPU Usage" will display 100 % and in the "CPU Usage History" it will show maximum value for both cores.
Then they are both working hard.
But it makes me wonder if Quad core CPU displays 25% in the task manager processes tabulate :smiles:
Running the Core on the HT P4 3.06 increase the ppd by ~20 %, executing two p2414 at the same time.
Note,
I'm using the WinLauncherXP 2.05 to setup the affinity for the work load,
http://www.majorgeeks.com/WinLauncherXP_d870.html
Each core of the Dual Core reports as 50% of total CPU. It looks JUST like HT...
My thoughts were why does XP split the load over both processors for a single threaded task? (I didn't think that was really possible) and thus uses both processors but only attains 50% CPU usage in total still. Wouldn't swapping processors for the same task slow things down?
I realize HT is not that hot of a topic anymore, but I just wanted to set the record straight.
That's strange. I wonder why FAH says that hyperthreading 2 instances 'isn't helpful'? Seems like they should want this, if they get a 125% boost in production per processor.
http://fahwiki.net/index.php/FAH_%26_SMP#Hyper-Threading
EDIT: that link is why I didn't initially do 2 instances on my HT'ers...:confused2