How to start folding

I already had -smp in

 

The core you downloaded was for the standard client, as you can see it isn't eligible for variable processors. You have the standard client installed.

You need to download and install the SMP client.

This is the page you require for every client BUT the standard one http://folding.stanford.edu/English/DownloadWinOther It is at the bottom.

 

I can't make it out in the picture, but HFM.NET is probably showing 'Standard' and a project number that is for a standard/classic client.

In future the flag -verbosity 9 is a good one to include as it writes more at the critical parts of the log file. You should also set the WU size as 'big'; most SMP jobs are 'big' so without it you may not get any WUs to fold.

Other than -verbosity 9 and 'big' WUs, this is where you went wrong - note the time to put it into context

Code:

################################################## #############################

Launch directory: C:\Users\Michael\FAH\FAH6
Executable: C:\Users\Michael\FAH\FAH6\fah6.exe

[18:48:55] - Ask before connecting: No
[18:48:55] - User name: meandmymouth (Team 35947)
[18:48:55] - User ID: 6EB4FEC75CCF620E
[18:48:55] - Machine ID: 1
[18:48:55]

You had not set the -smp flag as as an additional option when previously configuring, and you then started the client without the flag -smp, so it defaulted to the standard client and downloaded the appropriate core. The recommended settings are to put -smp and -verbosity 9 as additional flags when you are running the configuration, since you then don't have to remember them afterwards.

You do NOT have to uninstall and reinstall (unless you want to).
From here you can let the current WU finish and return the result, the -smp flag will kick in when the new WU starts (but you will still need to configure for 'big' WUs); this would be Stanford's preferred option or
you can stop the client, delete everything, and reinstall from scratch; that would be the cleanest option or
you can stop the client, delete the work folder, the log files and the core; that is the dirtiest and most likely to fail option.

Look again at the Stanford instructions, always a better option than second or third generation instructions from relative newbies.

Good luck

 

That is the exact one I downloaded, twice!
I'll try it again tomorrow

 

opening post edited: made clearer and included links

The opening post does say to make work unit size 'big' and does say to read fah instructions '3 times'

 

If you don't configure it right, it defaults to the standard client.

 

Well well, I tried again today and as far as I could see did everything exactly the same but now it works...

However, now my GPU has stopped producing 8000PPD and gone to half that....

 

Can we take this one step at a time, otherwise confusion happens (at my end at least)?

You now think you have the SMP client running on your i3, so you should be seeing 4 threads, mentioned at the beginning of the log file, and also shown in Task Manager. Is that right?

 

The SMP client is definitely working. 4 threads and just over 4k ppd on the i3 which i believe is about right for it.

My GPU however has slowed down dramatically. A couple of days ago it was producing almost 8k where as now it's just below 4k ppd. Could it just be a rubbish work unit that it's running?

 

Lol. So yeah, is smp working as it should, fahcore A3 in task mgr, 100% load and reaching expected ppd?
If yes and your sure then....

Did you change gpu client priority to 'slightly higher' in the config?

EDIT: Damn slow typing on phone lol

 

Slightly higher is checked, as always, and almost at full too

 

We now enter the realms where only someone with exactly the same hardware and software will be able to give you definite answers - so usually you just have to do the experiments yourself.

You ask if you have a rubbish WU - but since you don't tell us what project it is, we can't answer that question. But you can go and check the relevant project page, and search the Stanford Forums, to see if others have reported problems.

Also you don't tell us which GPU client you are running, GPU2 or GPU3. Some would say that with a GTX 275 you will get better PPD with GPU2, others would say using GPU3 won't make much difference, but may reduce the PPD you are getting from the SMP client. Try them both.

You will need to configure the GPU client as 'slightly higher' if you are using the systray version (assuming you configured the SMP client at the default setting of 'idle'.)

You should also check what happens to the PPD of both SMP and GPU clients if you run the SMP client with the flag -smp 3 or -smp 4. Enjoy.

I am assuming you have FahMon/HFM.NET configured to calculate its figures using the 'last 3 frames' option, and you are waiting long enough for the changes to show up. Note that changes in the configuration only take place if you restart the client, or at the start of the next WU.

 

Ah, now I did install the GPU3 client, that may well be part of the issue. I'll have a check as to what client is actually running, sheck what work unit it's on and look on the projects page.

Thanks for the help guys, if I can get it higher again I shall be back with my noobish questions!

 

@Chris sort of a side note: how do you learn about WU's being good or bad as far as ppd goes? Is it a general consensus + experience thing, or is there more to it?

Cheers

EDIT: just found the 'projects' page @ Stanford, will have a proper look tomorrow as phone + browsing = doing my head in! Lol

 

Noooooo!! Just checked stats page, says I will overtake you soon if your ppd stays low lol

Good luck getting it sorted

 

Ah ha! All working again and producing 12k!! My highest yet!

Now i just need to find an extra folding card or 2...

 

........ lol nice one.

What was it then ???

 

I'm not 100% sure

I made sure I was actually running the GPU3 client by unistalling my GPU client and downloading it again. Seems to have done the job.

 

Nice work guys. My mate gets his flat next week and he has a corei3 540 and a gigabyte 1GB gtx460SOC so hopefully should get him onboard.

 

Very nice, that's what we like to see!