New Members Welcome & Advice Thread

[p2501]

Very good, looks like it's working just fine.

When adding GPUs to the mix on everything else but XP you'd be well advised to run WinAFC64 to sort your affinities. Let CPU folding hog everything from thread 0-10 and let the GPU run on the last thread, that way you'll max out production on both.

If you upgraded to 12GB you could try running 4 threads for Windows a3 threads, a 7-thread VMWare VM for bigadv and the last thread as always for GPUs, I'd be interested in seeing what that could produce. (For Folding on virtualized Linux for bigadv always run VMWare, everything else will put penalties on your avg time per %)

[MaximumPig]

i think you answered my question but why does bigadv need to run in a VM? is it because it is linux-only?

[SnW]

i think you answered my question but why does bigadv need to run in a VM? is it because it is linux-only?

Yes .

[1200W]

I got into folding too . Started yesterday and ill try to keep it up and running

[Sparky]

Great!

[penguins]

just breached the 3k+ mark :O got my ps3 quad 771 and my desktop running 1/2 way there

[PoppaGeek]

just breached the 3k+ mark :O got my ps3 quad 771 and my desktop running 1/2 way there

[MaximumPig]

i seem to have angered the folding gods--after about 15 work units with a3 core my last work unit on SMP client is a 59-point WU on fahcore_78

is this because i missed a "preferred" deadline (but still made the final deadline) (wife slept the PC overnight)?

also even at my best, the 980X seemed to max out on <1000 ppd while i know people are getting 2300++ with this cpu, i must be doing something wrong

[=[PULSAR]=]

also even at my best, the 980X seemed to max out on <1000 ppd while i know people are getting 2300++ with this cpu, i must be doing something wrong

Is is overclocked? even at 4ghz should get around 6500ppd.

[MaximumPig]

not OC'd (well mildly, 2 CPU multiplier steps up, so CPU freq is around 3.7ghz )

i'm also using GPU client on my ATI 5870 but that's on 1 core only (using WinAFC) per p2501's suggestions), so 10 or 11 cores are SMP

could be a reporting error, i suppose--i use fahmon and look at the ppd it shows in the main window--is that standard or is there a more reliable way to see PPD?

[p2501]

There must be something wrong if you get only 1k PPD, 11 threads @ 3.7GHz should produce something entirely different, even without bonuses.

Do you have the box folding 24/7 or some hours a day? Checking taskman, is it really always pegged at 100% on all threads? Is it properly cooled (i.e. could it be that it's running into thermal throttling)? Do you run some CPU-heavy apps beside folding?

Besides, the PPD reported by FahMon are pretty reliable.

[Sir_Loin]

I haven't checked to see if fahmon has been updated to properly show the bonus points on a3 WUs, but I do know hfm (hfm.net) does.

[MaximumPig]

lol it would be funny (with my apologies for wasting your time) if this came down to my wife sleeping the PC every night. . . i will have to look into that.
but i note that FahMon lists the GPU at 4500+ PPD even if i run it only a few hours a day, so intermittent folding with the SMP should also show a more or less instant PPD rate, corrrect?

but regardless, CPU was pegged at 100% on all threads except now it is not because of my core_78 work unit (at 12% or so, with 5 of the 10 cores idle). strangely, the 1 thread dedicated to the GPU client is running 100%.

I'm gonna re-check my settings and run SMP alone without GPU for a day or two and see what WUs and PPDs I get.

// edit : thanks Sir Loin, will try hfm instead of fahmon

[p2501]

lol it would be funny if this came down to my wife sleeping the PC every night. . .

No time wasted here. In fact this would be very funny indeed!

strangely, the 1 thread dedicated to the GPU client is running 100%.

This sounds strange, are you sure you set up WinAFC correctly? Do You mind posting your conf?

[MaximumPig]

OK so I restarted the SMP client and it discarded the core_78 WU and I got a massive a3 wu in its place

it turns out if I run SMP on all 12 threads (it maxes all at 100%) HFM gives me a PPD of ~20,000K -- fantastic, if true

so then I add my GPU client on one of the threads, reserving 11 for SMP and 1 for the GPU client.

it seems that adding the GPU freezes up the SMP client--taskman still shows all 12 threads at 100% but there is no progress reported in HFM or in the SMP log file. before I started the GPU, the SMP was racking up 1% progress every 7-8 minutes but with the GPU, 40 minutes go by with no progress, even with 12x100% CPU usage. on top of this, it takes a full reboot to get to a point where the SMP client will run and report any progress, i.e, simply shutting down the GPU client and restarting SMP doesn't work, it just picks up where the last cycle stalled and stays stalled.

If I get the SMP client running and let it run alone, it continues the 7 min/1% rate of progress, apparently indefinitely.

i am thinking this is either due to a conflict in the way i set up the clients or something is overheating or throttling which locks up the SMP client but apparently does not otherwise affect stability.

realtempGT shows CPU temps in the high 60s, sometimes pushing up to 73C or so, but mostly less than that, which I think should be safe,
and the everest "northbridge" sensor reports <65C at max loads.

[Sir_Loin]

try rerunning the -configonly flag and set your priorities. Since the ATI card won't produce nearly the pointage of the smp (if the numbers from hfm are correct), set the smp client to LOW and the gpu to IDLE. Be certain each has a different machine id as well (1 and 2). If this doesn't help, try unlocking the gpu from a single cpu core.

I've run these two clients simultaneously on a few linux boxen, so I'm thinking you should be able to.

[p2501]

Okay, it seems temps aren't the culprit here.

Did you set -smp 11 and sorted these 11 threads together with one free for gpu or did you just set -smp? I f 12 processes are sorted to 11 threads maybe it results in timing issues between them. Leave out the GPU at first and try to get 11 threads maxed with actual production... then report.

[FoldingFanatic]

New to the forums and an avid Folder. although I fold for another team # we are all on the same team in a sense. Hopefully we can learn from each other and all gain from helping each other.

My many thanks for the advice I have received thus far. And hopefully in the future I can return the favor. I am currently building a portable Folding Farm that will be going out to trade shows and events to raise the awareness of Folding at Home. A project like the one I am currently working on takes allot of time, money and effort. Through this process I have made quite a few contacts that can be quite helpful in this sort of project. I am not telling you this to brag but to enlighten you to the fact that you can get sponsors to help offset the costs of building and running such a setup. FoldingForTheFuture is a non-profit org working towards bring awareness to the cause. If there is any way I can help anyone here just speak up.

Folding For The Future
EG

[Frogeye]

I'd like to be able to run F@H and BOINC at the same time. Anyone care to tell me the correct way to make this happen? I'm using the machine listed in my sig file. Thanks all. Currently I'm running BOINC/WCG @ 100% cpu.

oh shoot noob error UE! I thought Folding was GPU based. I guess it's CPU based so I should just stick with BOINC instead of running both BOINC and Folding right? TY

[Sparky]

Hi Frogeye!

FAH is actually CPU and GPU based - they have a client for each. I run WCG on my CPU and FAH on my GPU right now

I believe there is some little trickery you gotta do to get the radeon 5000 series folding - can't remember off the top of my head (I'm running a 4870) but I'll try to look it up.

[FoldingFanatic]

I believe with the 5000 series you have to use a force flag setup, but i believe at this time about 85% usage of the GPU is what you will get. That is until Standford gets done working on the client.

EG

[BoNg]

hey there everyone just installed fah504 on my machine under linux mint. never folded before but my client keeps getting assigned ip 0.0.0.0
from what i gather im just waiting for an assingment thats right for my machine is this correct?
heres a copy of what im getting

- Connecting to assignment server
[20:49:27] - Successful: assigned to (0.0.0.0).
[20:49:27] + News From Folding@Home: Welcome to Folding@Home
[20:49:27] Work Unit has an invalid address.
[20:49:27] - Error: Attempt #7  to get work failed, and no other work to do.
             Waiting before retry.

[BoNg]

got it working now changed :Allow receipt of work assignments and return of work results greater than
5MB in size (such work units may have large memory demands) (no/yes) [no]? y.... from no to yes and bam it working

now how do i keeps track of progress/ pts/ my username which is different than the one im using in the forums and i did make my team # 36362

[p2501]

Hello there and welcome aboard! For tracking progress locally you can use FahMon, on the net there's eoc for example. Don't find you there though, did you already send in work units on that nick?

[BoNg]

idk lol im just letting it run havent done anything else dunno how.... so far im.. 65000 out of 250000 steps (26%).. and its used about 75 gigs of disk space.... i hope it deletes it when its done

edit: oh and my name in foh program is Savage3306