Folding@Home - Who's using what?

zodac

Daegerty wrote: »

Will there ever be a linux GPU client?

Client v7 has a native Linux client, though there aren't any GPU WUs on the Linux core at the moment. They are aplnned though, so it's only a matter of time.

Lethal_Bullet wrote: »

Hmm, fan control profile isnt working now....well within limits but I'm really not happy having it that hot. 5Hrs a unit now.

roast

Using a -bigadv client for my Core i7 920 @4.2Ghz.
I have the SMP client running on a C2D E7200 @3.8Ghz
I have two GTX 285's (oc'd slightly) and they're both folding with the GPU2 client.
Finally, a 9800GT (again, oc'd slightly) running the GPU2 client.

Was going to throw together another machine with three 8800GT's that I have lying around, but I have no spare PSU to put into it, and cant justify spending more money on F@H!

zodac

:eek:

That's a nice collection of rigs. The i7 alone beats my entire 19k output. :P

roast

zodac wrote: »

:eek:

That's a nice collection of rigs. The i7 alone beats my entire 19k output. :P

And also probably exceeds your power usage too! :pac:
The only thing thats really stopping me from folding 24/7 these days is the ridiculous ESB bills I've been getting. Although, its quite handy that I never have to turn on a heater in the apartment.

zodac

God, I know. Been running full pelt these last five days, as high an OC as I could manage (until my GPU started dying ), and I am starting to melt here...

I'll be semi-glad when this is over. :P

roast

zodac wrote: »

God, I know. Been running full pelt these last five days, as high an OC as I could manage (until my GPU started dying ), and I am starting to melt here...

I'll be semi-glad when this is over. :P

Ugh.... I've tortured so many GPU's to the point of death.
Ah well! All for a good cause!

zodac

The best.

ED E

Card keeps dropping its clock back to 400/900 from 900/1200....any ideas?

[Not heat]

zodac

nVidia Control Panel> Manage 3D Settings> Change "Power management mode" to "Prefer maximum performance".

ED E

That'd be great.....but I'm not on the green team.

UPDATE: I had suspected powerplay, but there wasnt much on FAH searches about it, so I assumed the cores are deemed a 3D app by the card. Turns out it was powerplay. Youtube(flash) is 2D, and once I open chrome the clocks dive, once I close it they restore:p I can change an afterburner key to lock out powerplay though so all is good

zodac

Lethal_Bullet wrote: »

That'd be great.....but I'm not on the green team.

UPDATE: I had suspected powerplay, but there wasnt much on FAH searches about it, so I assumed the cores are deemed a 3D app by the card. Turns out it was powerplay. Youtube(flash) is 2D, and once I open chrome the clocks dive, once I close it they restore:p I can change an afterburner key to lock out powerplay though so all is good

Stopped reading after that. :pac:

cian1500ww

Looks like I got a whole bunch of points that I was due, I was wondering why my points were so poor lately!! http://folding.extremeoverclocking.com/user_summary.php?s=&u=541418

zodac

Nice points drop.

cian1500ww

zodac wrote: »

Nice points drop.

I wonder is the problem on my end

zodac

It's possible... you got a copy of the log?

cian1500ww

zodac wrote: »

It's possible... you got a copy of the log?

I have four machines folding but my main machine pulls in the bulk of the points. Here's it's log:

[12:32:36] Called DecompressByteArray: compressed_data_size=24866578 data_size=30796293, decompressed_data_size=30796293 diff=0
[12:32:36] - Digital signature verified
[12:32:36] 
[12:32:36] Project: 6900 (Run 26, Clone 7, Gen 47)
[12:32:36] 
[12:32:36] Assembly optimizations on if available.
[12:32:36] Entering M.D.
[12:32:42] Mapping NT from 8 to 8 
[12:32:45] Completed 0 out of 250000 steps  (0%)
[13:05:29] Completed 2500 out of 250000 steps  (1%)
[13:42:46] Completed 5000 out of 250000 steps  (2%)
[14:18:42] Completed 7500 out of 250000 steps  (3%)
[14:58:50] Completed 10000 out of 250000 steps  (4%)
[15:02:48] - Autosending finished units... [May 19 15:02:48 UTC]
[15:02:49] Trying to send all finished work units
[15:02:49] + No unsent completed units remaining.
[15:02:49] - Autosend completed
[15:30:02] Completed 12500 out of 250000 steps  (5%)
[15:59:05] Completed 15000 out of 250000 steps  (6%)
[16:26:30] CoreStatus = 89 (137)
[16:26:30] Client-core communications error: ERROR 0x89
[16:26:30] Deleting current work unit & continuing...
[16:39:05] Trying to send all finished work units
[16:39:05] + No unsent completed units remaining.
[16:39:05] - Preparing to get new work unit...
[16:39:05] Cleaning up work directory
[16:43:09] + Attempting to get work packet
[16:43:09] Passkey found
[16:43:09] - Will indicate memory of 3938 MB
[16:43:09] - Connecting to assignment server
[16:43:09] Connecting to http://assign.stanford.edu:8080/
[16:43:11] Posted data.
[16:43:11] Initial: ED82; - Successful: assigned to (130.237.232.141).
[16:43:11] + News From Folding@Home: Welcome to Folding@Home
[16:43:11] Loaded queue successfully.
[16:43:11] Sent data
[16:43:11] Connecting to http://130.237.232.141:8080/
[16:43:17] Posted data.
[16:43:17] Initial: 0000; - Receiving payload (expected size: 24869118)
[16:45:29] - Downloaded at ~192 kB/s
[16:45:29] - Averaged speed for that direction ~202 kB/s
[16:45:29] + Received work.
[16:45:29] + Closed connections

It's running the Linux v6 SMP client, I think I'll try the beta version of the v7 clients and see if that works. That error about the client core communication reoccurs a lot in the log.

zodac

Don't bother with v7 unless you have an AMD GPU; very buggy.

As for the log... do you have anything from when your last WU completed? Then we can see if it's an issue with the servers just crediting the WU late, or your client not uploading.

cian1500ww

I'm going to switch back to windows on that machine until they've released the new clients for Linux because I have a GTX260 that I can't fold with in Linux

zodac

Why not run a VMware of Linux for -bigadv, and then Windows GPU? Would still get more PPD than native Windows.

cian1500ww

zodac wrote: »

Why not run a VMware of Linux for -bigadv, and then Windows GPU? Would still get more PPD than native Windows.

Just got that setup, it'll be interesting to see how it performs

zodac

Should be a nice improvement. Seeing 15-20k increases on high end i7s, and even 10-15k on X6s and i5 2500ks.