SMP F@H client woes

[legoman666]

Hi.

I've recently switched all of my dual core machines from 2x single core client to the SMP client due to the fact that it gets ~4-5x more PPD. However, it is not stable. I have it on 3 of my machines.

It'll randomly freeze
It will not autostart after a reboot
often gets to the end of a WU and sits there and does nothing
Uploads WU results and then doesn't get a new WU
shutting down the client in the middle of a WU seems to have ~50% chance of nuking my progress so far.

and other things. The most annoying is the random freezing, it happens on all 3 of my machines. Any ideas for these problems?

 

[Insidious]

Probably the autostart attempts are leading to a lot of this.

All attempts to run SMP client(s) automatically at start up I tried (and lots of other people too) didn't work.

There are some processes that have to get going at startup (.Net framework stuff and others) for SMP to run right If you try to autostart, they race and SMP usually looses with very unpredictable operation.

I'd begin the troubleshooting with waiting until Windows is up and then use a shortcut (don't forget those flags.... ) to start your client

-Sid

 

[Gravity]

Assuming you followed the directions?

installed the .net fix
run the exe file
run install bat
run the fah.exe

Installed the affinity changer (first)

Are you over clocked? If so, is your ram pretty good? I experienced the same trouble with an overclocked box. I am running it stock now and it's not producing as many errors. Putting better ram in seems to solve the problem.

Stick with it,

Gravity

 

[legoman666]

Originally posted by: Insidious
Probably the autostart attempts are leading to a lot of this.

All attempts to run SMP client(s) automatically at start up I tried (and lots of other people too) didn't work.

There are some processes that have to get going at startup (.Net framework stuff and others) for SMP to run right If you try to autostart, they race and SMP usually looses with very unpredictable operation.

I'd begin the troubleshooting with waiting until Windows is up and then use a shortcut (don't forget those flags.... ) to start your client

-Sid

Yea I invariably end up killing the fah.exe instances that autostarted with windows and then starting it via command line. Glad to know that isn't just me.

Originally posted by: Gravity
Assuming you followed the directions?

installed the .net fix
run the exe file
run install bat
run the fah.exe

Installed the affinity changer (first)

Are you over clocked? If so, is your ram pretty good? I experienced the same trouble with an overclocked box. I am running it stock now and it's not producing as many errors. Putting better ram in seems to solve the problem.

Stick with it,

Gravity

2 of the machines are only dual core (3rd is quad) so I don't have the affinity changer installed. I was under the impression that it only helped with 4 or more cores? And yes, I followed those install directions.

2/3 machines are overclocked and I seem to have the most trouble with the machine that isn't. I am not at home right now so I'll investigate later.

 

[GLeeM]

When Random freezing happens does the window stay open but if you look in Task Manager the four "FahCore_a1.exe" processes are gone?

That is what happens when a network change/interuption happens.

Have you checked out Sid's guide thread?

 

[legoman666]

Originally posted by: GLeeM
When Random freezing happens does the window stay open but if you look in Task Manager the four "FahCore_a1.exe" processes are gone?

That is what happens when a network change/interuption happens.

Have you checked out Sid's guide thread?

That is exactly what happens. All of my pc's on my home network are set up with static IP's.

 

[GLeeM]

The following is from here. (warning their "weeks" have stretched to months!)

March 08, 2008
New Windows client/core development (SMP and classic clients)
The Windows SMP client is a pretty important new client for us for two reasons. First, multi-core CPUs are becoming more and more common, and second, Windows clients are the most common in FAH by far. The SMP client appears to be working quite well on linux and OSX, and this is due to the fact that these machines are unix-based, and so multicore libraries (eg MPI) are much more robust on these platforms. The issues we've been seeing on Windows can be directly attributed to MPI issues. We're the first distributed computing platform to roll out MPI calculations (typically reserved for supercomputers), so we are dealing with some major growing pains issues there.

To help fix this, Peter Kasson has been working hard on a new approach to improve the Windows SMP core stability. The trick was to use an improved MPI library. So far, it looks like it is indeed a significant improvement, but we're still testing it to be sure. If the Windows SMP core is not working well for you, we hope that the new fixes will help.

For now, if the SMP client isn't working well for you, one can always run multiple regular Windows clients and still make major contributions to FAH (and lots of extra points than a single client!). You can see more info on our Windows console FAQ.

Finally, we're working to reintegrate and revamp our Windows clients in general. We've revamped the graphics, made it much more stable (and better looking), and we've set up a single client to serve as both regular and SMP clients (with GPU functionality on the way). We're excited about these new developments and donors will be seeing these advances in the next few weeks, if all goes well in our beta tests.

 

[Wolfsraider]

Awesome to see such a well informed team at the head of things and them communicating with the end users 🙂

Thank you Gleem!

 

[legoman666]

just lost 2 more completed SMP WU's because they froze at 100% and wouldn't upload. When I restarted the clients, they rejected the checkpoints. I eagerly await the new beta 😛

 

[Insidious]

I think it's time to re-group.

I think you should :

Re-boot to stop the clients
use Add/Remove Programs to uninstall Folding at Home SMP windows client
delete the working directories you were using
download a fresh copy of the SMP client installer
Follow these directions:

Installing 2 Folding@Home SMP clients on a Quad core WindowsXP or Vista PC

First thing is to make sure that you are using a passworded Administrator account
Make two directories: c:\fah1 and C:fah2
Run the downloaded installation program and make sure you tell it to install in C:\fah1
Next run the downloaded installation program again and make sure to tell it to install in C:\fah2
Go to c:\fah1 and run the file 'install.bat' (Vista: right-click and use 'run as administrator')
It will ask for the user name (windows account user name)
It will ask for your password (windows account user password)
You should see it display MPI is working twice if it is has worked
Then hit a key to exit that window
Right-click on fah.exe and select create shortcut.
Right-click on the shortcut and select 'properties'
In the target line of the shortcut, add these command line switches to the far right end of the existing text:
-local -forceasm -advmethods -verbosity 9 (no quotes around these commands)
Under Vista, be sure to go to the advanced button of the shortcut properties window and then select run as
administrator
Now double-click the shortcut to run the SMP client for the first time. You should enter the configuration
questions (remember we are team 198)
The important questions are:
Your name
Your team (198)
Tell it you do NOT want it installed to start automatically as a service
Tell it NO to using IE settings
Tell it YES to accepting work units >5MB
Tell it NOT to disable highly optimized assembly code
Tell it YES to changing advanced options
When it asks for machine ID set it to 1
When you finish the configuration it should exit
Now, do all the above for the installation in the c:\fah2 folder. Everything is exactly the same EXCEPT BE
SURE TO SET THIS CLIENT TO A DIFFERENT Machine ID (2)
So now they're both installed and you can start them using the shortcuts you created (this gets the switches
put in each time)

If you are on a wireless network, follow these directions:

Note: Before you start any of this, re-start Windows to make sure your F@H clients are not runninig. Also as you go along, remember what your original settings are in each place I'm going to have you change things so you can get back if you choose to later.

first check your router settings to see what is the starting address of DHCP (it's normally in the basic setup section.) For instance, mine says the starting address is 192.168.1.100

in one of the next entries it should tell you how many is the maximum number of DHCP users is. mine says 50, so that means that addresses ranging from 192.168.1.100 to 192.168.1.150 are DHCP assigned addresses which will be renewed periodically and can even be changed automatically by the router (this borks up the SMP clients).

from the above, I know that any address from 192.168.1.2 to 192.168.1.149 and addresses from 192.168.1.151 to 192.168.1.254 are all free to have clients with static addresses. (ones that won't change)

From this I decided to give my wireless PC the address of 192.168.1.160 (I just picked one)

Next, we have to set lup your PC with the static address.

First, right-click on the my network places icon and select properties from the drop-down menu.

double-click the wireless connection entry to pop up the local area status window. Select the support tab

write down the entries you see for subnet mask and default gateway. now click on the details tab and copy down the addresses for DNS servers. now close the status window and right click the wireless connection entry in my network places and select properties

in the window that comes up, highlight the internet protocol (TCP/IP) entry and then click the properties button.

we are going to select the 'Use the following IP address' button (which clears the 'Obtain an IP address automatically' button)

enter the IP address you chose to be the static address of your PC in the IP address section. Then enter the addresses you copied earlier into the subnet mask and Default gateway entries.

enter the DNS server addresses you copied earlier too.

You're done. re-boot windows and make sure your internet is working then start F@H

-Sid

edit: For dual core PCs only follow the direction pertaining to the first folder and one client install... don't attempt running 2 SMP clients on a dual core PC

 

[legoman666]

will do. cheers!

 

[GLeeM]

Good luck legoman666 🙂

After re-installing my SMP has been pretty good, only an occasional hiccup.

 

[legoman666]

OK. one of my comps was constantly giving me probs with the SMP client. Mostly File IO errors, but some times it wouldnt upload a finished WU. I was about to throw the fucking machine out the window I was getting so fed up with it. Then I disabled Nod32 and now it's working fine.

Le sigh.