• 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.

SMP client

F

flint338

Gawd
Joined
Oct 26, 2005
Messages
1,005
Hi all.. trying to get the smp client to work with my c2d machine, running XP pro.

Do the install, then run install.bat, it says MPI is installed (twice, of course), then I try to run FAH.exe .. it asks me all the normal questions, downloads a core (FahCore_a1.exe).. then says "+ Working ..." and just stops there.

No units working, no cpu being used at all.

Any ideas? :)
 
flint338 said:
Hi all.. trying to get the smp client to work with my c2d machine, running XP pro.

Do the install, then run install.bat, it says MPI is installed (twice, of course), then I try to run FAH.exe .. it asks me all the normal questions, downloads a core (FahCore_a1.exe).. then says "+ Working ..." and just stops there.

No units working, no cpu being used at all.

Any ideas? :)
Click to expand...

It will take a bit for it to complete the first step. I think you just have to be patient with it. If you have not made a step in 30 minutes or so then you should worry. You should see th CPU being used though. The SMP client does take a bit to engage.
 
All it took was me complaining about the thing not working.. and unsetting the 'read only' flag on the work directory (why it was set is beyond me)

Seems to be running 4 threads.. not showing any status, but it's at 100% cpu on both cores.

The last messages it has on there are..

[01:38:50] Entering M.D.
[01:38:56] Rejecting checkpoint
[01:38:56] Core Shutdown: FILE_IO_ERROR
[01:38:56] Finalizing output

But I guess it's working as the CPU's are cooking.
 
flint338 said:
Seems to be running 4 threads.. not showing any status, but it's at 100% cpu on both cores.

The last messages it has on there are..

[01:38:50] Entering M.D.
[01:38:56] Rejecting checkpoint
[01:38:56] Core Shutdown: FILE_IO_ERROR
[01:38:56] Finalizing output

But I guess it's working as the CPU's are cooking.
Click to expand...
That error statement is usually indicative of a problem somewhere in my experience. Did the client DL another WU and currently processing it?
 
Whats the Protien (Run, Gen, Clone) numbers ?

There are a few bad protiens still in the wild and you may have got one from that error message.
Any I find, I'll report to the support forums so I only get it 2-3 times before Stanford pulls it.

Luck ........... :D
 
flint338 said:
All it took was me complaining about the thing not working.. and unsetting the 'read only' flag on the work directory (why it was set is beyond me)

Seems to be running 4 threads.. not showing any status, but it's at 100% cpu on both cores.

The last messages it has on there are..

[01:38:50] Entering M.D.
[01:38:56] Rejecting checkpoint
[01:38:56] Core Shutdown: FILE_IO_ERROR
[01:38:56] Finalizing output

But I guess it's working as the CPU's are cooking.
Click to expand...

Okay.. left it going overnight, no change, no packets uploaded. Seems the only machines that are actually working are the linux ones! :)

[01:38:50] Project: 2653 (Run 3, Clone 173, Gen 60)

I have 15.1m of data in the work/ folder. Going to try running it again..

edit:
This time:
[11:28:52] Project: 2653 (Run 3, Clone 173, Gen 60)
[11:28:52]
[11:28:52] Entering M.D.
[11:28:59] Couldn't open Go file
[11:28:59]
[11:28:59] Folding@home Core Shutdown: FILE_IO_ERROR
[11:28:59] Finalizing output

Still running 4 threads at 100% cpu, but doesn't seem to actually be processing a WU.
 
Make sure the folder you have the client installed to has read and write permissions set. This should be fine by default under XP Pro, but hey you never know
 
schenksmill said:
Make sure the folder you have the client installed to has read and write permissions set. This should be fine by default under XP Pro, but hey you never know
Click to expand...

yeah, that was the first thing I checked too..

the thing that seems to be constant is this "Can't find go file" error

it also seems that the SMP client requires some sort of loopback networking. It appears to be working correctly with windows (I can ping 127.0.0.1, localhost, longitude and my actual network IP). The only thing kind of funny with this machine is that I'm using 2 NICs, one to 192.168.0.0/24 and one to 192.168.1.0/24. I think I'll try disabling the second nic and see if that makes a difference.. I can't figure this one out for the life of me.
 
arg! new update:

installed vmware, built XP pro, enabled both procs in it and just installed net 2.0 and the smp client.. ran it and the damn thing works.

"Completed 0 out of 5000000 steps"

Now I wonder why the thing won't work on my host computer.
 
Happy you have it working, but I'm puzzled...
Are you saying that you installed vmware on XP, then created a virtual machine of XP, and that folding is now working within the vm?

 
Celerator said:
Happy you have it working, but I'm puzzled...
Are you saying that you installed vmware on XP, then created a virtual machine of XP, and that folding is now working within the vm?
Click to expand...

Looks like it. Weird. I wonder how performance is with that setup?


 
flint338 said:
All it took was me complaining about the thing not working.. and unsetting the 'read only' flag on the work directory (why it was set is beyond me)

Seems to be running 4 threads.. not showing any status, but it's at 100% cpu on both cores.

The last messages it has on there are..

[01:38:50] Entering M.D.
[01:38:56] Rejecting checkpoint
[01:38:56] Core Shutdown: FILE_IO_ERROR
[01:38:56] Finalizing output

But I guess it's working as the CPU's are cooking.
Click to expand...

Take a look at this software if you are using NOD32. It can cause that error.
 
aldamon said:
Looks like it. Weird. I wonder how performance is with that setup?
Click to expand...

Yup, XP within XP. Figured I might as well see if I could replicate the problem.. and honestly, performance wasn't bad. I was getting (according to FahMon) around 1300ppd. I *think* I figured out the problem, but it makes no sense: if I accept "big" WU (the 5meg question), it fails. If I don't accept them, it works fine on the host machine, pulling around 1400ppd right now.

However, I think that's not a good thing as I'm supposed to be doing big WU's.

Also, I de-clocked the machine back to 2.4ghz stock/memory stock, just in case that was fubaring it. So I'd be really interested to see how quick this thing goes at 3.0ghz..

And funny part, ALL of my other machines combined are doing 1100. This one machine beats them all with the smp client, provided it's actually able to process a WU. :D

Mayhem33 said:
Take a look at this software if you are using NOD32. It can cause that error.
Click to expand...

I *am* running nod32. Any links on how to bypass it with folding? I rather like nod.
 
flint338 said:
it also seems that the SMP client requires some sort of loopback networking. It appears to be working correctly with windows (I can ping 127.0.0.1, localhost, longitude and my actual network IP). The only thing kind of funny with this machine is that I'm using 2 NICs, one to 192.168.0.0/24 and one to 192.168.1.0/24. I think I'll try disabling the second nic and see if that makes a difference.. I can't figure this one out for the life of me.
Click to expand...
I have 3 NICs on two of my machines each and at least two NICs on most of my other machines. No issues related to the NICs as long as they're not defective. I had one bad NIC that was randomly disconnecting and I replaced it with a working one.

aldamon said:
Looks like it. Weird. I wonder how performance is with that setup?
Click to expand...
Can't be any faster than running it on the host (native), I'm certain about that.

flint338 said:
Yup, XP within XP. Figured I might as well see if I could replicate the problem.. and honestly, performance wasn't bad. I was getting (according to FahMon) around 1300ppd. I *think* I figured out the problem, but it makes no sense: if I accept "big" WU (the 5meg question), it fails. If I don't accept them, it works fine on the host machine, pulling around 1400ppd right now.
Click to expand...
Since you're using VMWare why don't you install Linux and see if your PPD will increase? I figure you will actually see better performance anyway and you can maintain your host XP installation without the associated headaches you're experiencing and keeping your preferred configuration intact as well.

 
APOLLO said:
Since you're using VMWare why don't you install Linux and see if your PPD will increase? I figure you will actually see better performance anyway and you can maintain your host XP installation without the associated headaches you're experiencing and keeping your preferred configuration intact as well.
Click to expand...


I'm actually not running vmware, got the native smp client to work. I could actually run a pretty stripped down version of linux with vmware -- the big problem with running windows under vmware is the amazing amount of memory it requires. I wonder how much it would need to allocate to run the linux smp client effectively..

That, and I'm not sure if vmware supports a 64-bit guest on a 32-bit host.
 
I am successfully running 4 vmware guest w/ ubuntu 64bit on a Windows XP 32 bit host.

I have given each vm 750mb of RAM and they seem to be running pretty well.
 
Wharf_Rat said:
I am successfully running 4 vmware guest w/ ubuntu 64bit on a Windows XP 32 bit host.

I have given each vm 750mb of RAM and they seem to be running pretty well.
Click to expand...
So, how much total RAM in that configuration does the motherboard have installed?
 
It has 4GB total, but since it is 32 bit host, device manager only reports 3.325GB
 
Well, if it was 64bit probably, unless the vms somehow have direct access to the RAM...

I dont mind how sluggish the host is, since all I have this thing doing is folding, for now anyway. It is a dual 2.0 xeon quad, not sure of the chip model number, 53XX something.

I have it running 4 vms, with each one having a set affinity to 2 cores. When I get in on monday, Ill take a look and see how long it is taking to finish a frame.

I would bet if this thing had to be used, other than just the vms folding, it would be god-awfully painful...since it might be only seeing 325mb of RAM.
 
Thanks for all the info Wharf_Rat. Pretty much what I was expecting. Too bad I don't have several spare 1GB FB-DIMM modules lying around. They're not cheap either.
 
Yeah those FB-DIMMs can really put a hurting on things $$$ wise.

Thankfully, I only have 4 of these boxes, only one is folding at the moment. Once I put them into production, I will probably only be able to get away with one client per box though :(
 
I logged in to check it out and I am glad I did! The host restarted and VMware server never did. Anyways, it is averaging ~19/20 mins a frame per 2 cores.

I grabbed some screens while I was in there:

1 vm running: not bad memory usage...

1vm.jpg


4 vms running: LOL, suck the life outta that thing!

4vm.jpg


So to answer our questions, it leaves the host virtually nothing with my current config. This also says the vms are using the 750 they are allocated. These ubuntu 6.10 installs are full blown installs though, no trimming at all. I didnt want to put forth any effort, since this is a temporary configuration.
 
I use a slightly striped down install of CentOS and my VM's only use ~500 Meg.
If you use "notfreds" CD installer you can get it down to ~350 meg per VM.
So in theory 4x "notfreds" VM's + Windoze will run inside 2 Gigs of memory but it will be tight.

Luck ............ :D
 
Tigerbiten said:
I use a slightly striped down install of CentOS and my VM's only use ~500 Meg.
If you use "notfreds" CD installer you can get it down to ~350 meg per VM.
So in theory 4x "notfreds" VM's + Windoze will run inside 2 Gigs of memory but it will be tight.
Click to expand...
Thanks a bunch for the additional info. I guess the oldest Windows version one may use in these scenarios is XP, correct?
 
I've got Xubuntu running SMP at 350 megs per VM....I allocate 512 per, but you could drop that to 400 safe and fit 4 within 1.6 gigs I guess, however I wouldn't recommend it....if using XP Pro, I'd say try and find even 512 megs to get you up to 2.5, even if the performance isn't the best. That being said, I would recommend getting 3 gigs. Any OS should be fine though. I've got 2x in Vista Business x64 and 1x in XP x64 myself.

 
schenksmill said:
I've got Xubuntu running SMP at 350 megs per VM....I allocate 512 per, but you could drop that to 400 safe and fit 4 within 1.6 gigs I guess, however I wouldn't recommend it....if using XP Pro, I'd say try and find even 512 megs to get you up to 2.5, even if the performance isn't the best. That being said, I would recommend getting 3 gigs. Any OS should be fine though. I've got 2x in Vista Business x64 and 1x in XP x64 myself.
Click to expand...
In any case, I was already considering the possibility of acquiring more modules in the future just to enable quad channel bandwidth for a slight speed improvement overall. From what you guys are saying, I can get away with adding two 512MB modules for 3GB total.
 
You must log in or register to reply here.
Back
Top