Folding Forum

good day to everyone. I have a question regarding stats and receiving the the credits for the work units that i am reportedly getting. i have two personal machines i run f@h on. both are win7 64bit pro, one has dual GTX465 video cards and an i7 970, while the other has a single GTX260 and an i7 920. Both machines have 6GB ram and plenty of clean stable power. both machines have been heavily stress tested and return good calculations under basically all conditions.

on each machine i run 7 occurences of the smp cpu client, 1 cpu console client, and one gpu client for each video card on each machine, for a total of 19 clients running basically 24 hrs a day. my smp clients take about 12 hrs to work through on set of data before returning a supposedly successful data set for an average of 100 work units. so by my calculation, i should be getting at minimum 1400 wu per day off my two machines, not even counting the console clients or gpu clients. i check and read the logs twice daily and they consistently show successful returns of data to a server and show as i said before on average 100 wu per send...yet daily my wu total only goes up by a few hundred at best, am i missing something here???

my f@h id is the same as my forum id.

thank you, sorry for the rambling on...

Welcome to the foldingforum, E46_3.0_BMW

I'm not sure what you mean by "7 occurrences of the smp cpu client" but if it's one client with the parameter string -smp 7, that's good. If you have started 7 different copies of the client, that's bad.

According to the official stats, you've have 19 Active clients, and that's not good. Based on what you described, it's reasonable for each machine to have one smp client running 7 threads, one GPU client, and one classic client for a total of 6 clients.

WU per day is probably not the most meaningful goal. Some WUs can be processed very quickly and others are much more difficult. Points are assigned to each project based on the amount of science performed, so a much better measurement is average points per day (PPD).

Be aware that a "stable" system based on something like prime95 may not be stable when running FAH since they exercise different parts of the CPU. Don't worry about that unless you have trouble and if you do, try stresscpu2.

If you don't have a passkey, I recommend that you get one and use it, especially on the smp clients.

Not sure where to post this but its a stats issue.

My stats are not improving, going up.
After completing tasks, my stats are the same.

On Win, When I right click on the F@H taskbar icon and select, Status, User Statistics, I am taken to a web page with my stats.
These have not increased since I stated crunching again, several weeks ago.
I am using a Passkey.

Also when I click certificate, it says 0 points!

@ SKGiven
I'm not sure where the link in the taskbar icon is taking you but you are getting credit for your stats. I did a search and found these links:
http://fah-web.stanford.edu/cgi-bin/mai ... me=SKGiven

http://fah-web.stanford.edu/awards/cach ... en.433.jpg

Yeah, still 197923...

User name issue perhaps? You are using some capitals,
http://fah-web.stanford.edu/awards/cach ... en.433.jpg

... and score still at 197923

Anyone?

SKGiven, could you post client logs so that we can see the completed WUs? Mods will then be able to check the WU database for you.

Well it's now at 198945. Might have been running an extra long task or just wasn't always running?
Anyway now onto another and @550 of a 1500 step task.
How do you get an i7-2600 to run more than one task at a time? It's only using one thread.

Right-click on the taskbar icon and select configure, click on the Advanced tab and add -smp to the additional client parameters(near the bottom).

e.g.
You have to quit and restart the client for this to start using more threads and it might not until you get your next work unit as it might be running a uniprocessor project.

bollix47 wrote:Right-click on the taskbar icon and select configure, click on the Advanced tab and add -smp to the additional client parameters(near the bottom).

e.g.

Code: Select all

-smp -verbosity 9

You have to quit and restart the client for this to start using more threads and it might not until you get your next work unit as it might be running a uniprocessor project.

You do have to restart the client to have the additional client parameters to be applied. You also have to wait until the current assignment finishes. The projects that are assigned to clients that don't have the -smp flag do not automatically become SMP projects when you add the flag, but the next assignment should be for SMP.

Does the -verbosity switch stipulate core/thread number to use?
Will these threads each run a separate task, or collectively run one task at a time?

Thanks,

The verbosity switch tells the client how much info to write to the log(FAHlog.txt) with 9 being the most comprehensive.

The -smp can be used with a number to restrict the number of cores. Without the number all cores will be used to run one task at a time.

e.g.

-smp is equal to -smp 8 with your system provided HTT is turned on.
-smp 6 would be another option. Prime numbers like 7 should be avoided although 3 works and 5 might work but I believe there have been reports where 5 was a problem so if you're not going to use all 8 then I suggest -smp 6.

Using all 8 should be okay unless you're also running a gpu client. The smp client runs at a very low priority so it shouldn't bother your day to day activies.

Thanks. Out of curiosity, can folding be run on a DC?

DC? IF that means a cluster, the answer is no.

A Domain Controller.