Those 479 point WU's made my ppd in fahmon go from ~5300 to ~5100. Takes like 2 1/2 hours each. No idea why your ppd took such a huge hit. :(
Printable View
Those 479 point WU's made my ppd in fahmon go from ~5300 to ~5100. Takes like 2 1/2 hours each. No idea why your ppd took such a huge hit. :(
does my ppd look alright, for some reason i think i was seeing people getting better ppd with an 8800 gt.
anyways heres my fahlog and progress...
That's good, I'm getting little less right now on my 8800GT
Just a note, the Vista X64 177.35 drivers I dl through this thread clocked my 9600gt down to nothing at all and didn't allow ocíng in both rt and atitool. Installing the internation version solved this for me, took me some hours to find it out and maybe this will spare someone else with the same issue the troubles I had. I'm getting just short of 4K with those drivers and 750/1900/900 clocks on my 9600gt, temps around 64c. I'm abit nervous about going to another driver again now I finally got these working but am I missing something, are the new one's that much better?
Thank god for 175.19. I didn't get any 177.x installed on XP x64 and the 174.55 crashed some games. Now it works great, 175.19 rocks!
Hows the ppd and what clocks?Quote:
Originally Posted by jcool
Doesn't show yet.. PPD was around 5000 with the 174.55 so I'm hoping for more. Clocks are a moderate 730/1730/1020 (disabled all Vmods but the vdroop for 24/7 crunching, don't want it to burn out)
They are coming out with new driver sets everyday, I couldnt get 177.35 or 177.37 to work (8800GT 512), but when I got 177.39 to finally work the 177.40 and .4x WHQLs came out... my god. :p:Quote:
Originally Posted by jcool
I really would like the extra ppd of the new drivers but seriously? I have spent so much time modding drivers the past week... its barely worth it. :shakes:
Well, they seem to have changed the PPD with the new 5004 project WUs. Not only are they way longer, they also decreased PPD. I'm getting 3100 right now, was around 5000 before like I said :(
Wow, they have increased my PPD, I was around 3200, now Im around 4000 with a moderate 630/1625/940 OC... :cool:Quote:
Originally Posted by jcool
EDIT: Another question, is it possible to fold 2 separate videocards in the same machine? Non SLi? For instance an 8800GT and an 8600GT folding at the same time?
My PPD only dropped around 100 or so. And Marvin, I was able to OC with RT using the vista x64 driver, That's odd it wouldn't let you. What happened when you tried to OC?
I believe there was a way to trick the program to do that I can't remember how:(Quote:
Originally Posted by [cTx] Raptor22
You can run more then one gpu with the -gpu x flag to the shortcut where 0 is primary and 1 is secondary. You have to disable sli though in the drivers ( I read, don't have sli so no first hand information, and in your example you wouldn't be running sli in the first place ).
@Kevin, it showed rt with just core and mem speedon the oc tab, and when I checked the box to allow ocing, detect now, didn't do anything ( boxed didn't stay ticked, couldn't oc ), with a 'detect on reboot' I had the same issue, nothing stayed ticked, the slider for shader speeds didn't show and I couldn't move the others. Also, my 9600gt stock clock is 650 something something and these drivers clocked it to 500/1200/500 or something close to it. I fixed it like I said with downloading NVIDIADisplayWinVista64(177_35)Int size 81.408.768 bytes directly from nvidia and that made everything work again.
While I couldn't oc with rt I treid Atitool and it kept complaining about not being able to install a low level driver, and yeah uac is of and I run installs as admin. With the new drivers and a reboot ati tool also worked so I'm abit stomped myself.
I running XP32 with an 8800GT
Can I simply download FW 175.19 install, then load GPU client & it will work?
I thought 177.35 where the lowest which supported CUDA? Not sure though. Why not just use those, I can tell you from experience they work :up:
Still waiting for someone who been through em all to determine if it's worth it going to another (newer) driver, it took me some time to get it running properly and I think with 4K gpu and 2K smp it's doing ok.
Marvin,Quote:
Originally Posted by Marvin_The_Martian
Have you tested if this flag is honored on the new version? The last beta ignored the switch - I had two instances running with 2 diff -gpu setting and it ran both at 1/2 speed.
No not personally, but I'm sure it should work since I will link you to the fahforums in a bit showing how to run the client on the secondary adapter, and even better, this guy who post history is such a clear description on what to do ;)Quote:
Originally Posted by angra
I'll edit the other link into this post in a minute.
Edit: though I just noticed I forgot to explain on trying to force the location of the work files with the local flag. If that doesn't work requim has done it diffrently and I have to rethink it but I'm as sure atm as can be without having two cards.
Edit2: http://foldingforum.org/viewtopic.php?f=43&t=3465 towards the bottum of the page. Looks like the same way it was with the ati cards before?
Aha, I think I have it running now. Sad to admit it, but I think I used "-gpu 1" and "-gpu 2" the first time I tried.
Now...if only they had a linux client. I would be able to put out a respectable PPD...
Nice! I asked ctx[Merlin] to try and report back but since you already got it running I suspect you'll put up something on how to do it?
May I ask hw exactly one would get the GPU client to be displayed in F@Hmon? Anything I have tried hasn't really worked out for me... :shrug:
Sure...not a walkthrough (no time) but...
What I did is make a copy of the program folder and appended "2" to the name.
Then I made a copy of the working folder in "Documents and Settings"... and appended 2 to the name.
Then I made a new shortcut in my Startup folder from the 2nd copy of the executable.
Then I changed the "Start in" directory on the new copy to point to the new working folder.
Then I started the second client via Startup folder shortcut, and changed the configuration on each running client via the system tray.
...right-click/configure - on advanced tab added "-gpu 0" and "-gpu 1" to "Additional Client Parameters" at the bottom of the advanced tab. Also clicked priority to "slight higher" on each.
This is not necessarily a minimum process, but it is sufficient.
No need to disable any adapters? Good news, means it's an easy fix. Thanks for the how to, hopefully allot of people will be able to use it :)
Quote:
Originally Posted by angra
You sure the -gpu flag works and your not just running 2 clients on 1 GPU? GPU core temps and PPD show you are in fact using a 2nd core? I'd be suprised if it worked, only 2 methods I know of to get multiple cards to work and -gpu flag isn't one of them.
I was thinking the same after reading that this client automaticly binds itself to the primary adapter. Which would mean disable primary, start on secondary and re enable primary again and start the other client.Quote:
Originally Posted by aicjofs
If you know the way, why not share it and not tell us you know how to do it without telling how :p:
My PPD imply that it is working properly. I dont have any core activity or temp meters installed on the machine at work. Is there an easy-to-install core activity monitor for nv chips that i can install remotely (i.e. no reboot)?
Not that I know of. ( edit = gpuz??? beter late then never )
But I trust and value your judgement, and I already suspected them to have kept the flag but for pr reasons treid us to think it was being left out so we wouldn't try this ourself ( and thus mess up their testing things first time table ). Think most will take your word for it if you say you're sure it's working, though imply is a political answer ;)
Marvin,
to be clear, I only mean as much as I say. The PPD imply that it's working, but I can think of scenarios that would deceive me. For example, if I somehow got the data directories confused, they could, conceivably be both working on the same WU or something. I don't know :). All I know is that I did what I said above, and FAHMON reports ~3900 PPD on each of 2 9600GTs running the 479 point WU. When running just 1, I got almost exactly the same PPD estimate.
No one should take my word for any of this! Try it and see. My rigs are at work and I am home for the afternoon waiting on a contractor to come visit my house.
Sorry wasn't implying you said something you didn't ok :)Quote:
Originally Posted by angra
That point estimate sound pefectly right to me, same as my 9600gt ( which clocks are you running? ).
It's just I do recall clearly reading a folding forum staff member post somewhere that the client just attaches itself to the first adapter just as the old ATI clients, so it's great news if it doesn't need the long steps to get it running otherwise.
:wtf:
Marvin,
it's all good - I just wanted to make sure I wasn't accidentally sending the impression that my comments should count for much more than one person's experience :)
I am trying to load up gpuz remotely on my work PC now - will it show loading or just temps?
OK one GPU is @ 68, the other @ 63 degrees. I have no idea whether this backs up or disclaims my "both are working" theory.
ETA:
alright, I took turns pausing the clients. When I pause one client, one GPU core goes down in temp and the other stays constant. When I reenable the first, and pause the second client, the core temps swap (low goes high, high goes low). When I reenable both clients, both core temps go back up.
Sounds like they are both working independently and attached to separate clients!
Great news Angra, on a forum like XS I'm abit dissapointed in the low attention this new client has gotten, especially outside the fah section. The ps3's folding capabilities were more spoken of in the news section I remember ( or think so.. ).
I don't know what issues you can run into though, there must be a reason their making this public especially if it's not that hard to set up. Don't forget to follow up on this, if you're sure you got it working, why not help the rest of xs and expand on your litle roadmap? Though, if you just install the client once, then copy that folder and create shortcuts to each exe they will run from their startup locations so if you use the gpu flag it's really that simple you don't even need to edit files ( instead you create two new shortcuts... well atleast I'm not violating eula's by altering client files ( just seen someone get the boot on the official forum for even mentioning he had it running on dual gpu )...
As to the attention - I have perceived a bizarre thing culturally on XS - there are a lot of people on XS that seem to "root" for one side or the other in tech rivalries (such as ATI/nVidia) as if they were sports teams. Every good thing ATI does is some sort of blow to the nVidia team and v/v. Same goes with AMD/Intel. People actively cheer on shortcomings and failures in the "other" teams' product lines, not seeming to grasp that it actually detracts from their consumer experience in the long run.
Since the ATI GPU client has been out for a while, and this is, basically, big news for nVidia GPUs mostly, I think this advancement suffers from this effect. The PS3 client was largely not subject to this situation.
That, and F@H is, to some degree, persona-non-grata among the more respected VIPs of XS. It is what it is.
Anyway, wrt to my psuedo-guide...I will be happy to add clarification. I don't really want to try to write a proper guide - it is not my strong suit. Hopefully someone with more writing patience will follow my lead, verify that it works, and write a more clear description :). Maybe with pictures :).
I would try if I had two cards hehe ;) if you pm me some pics I might still thought you would be lacking the confirmation part which is imho a pretty big part ;)
For now I'm just glad it's doable, and atleast one in the team can inform the rest if they wish to try this.
:wtf:
And sadly you're right about the rest.
This would be great if the new beta supports -gpu flag. I tried on beta 6 and you were right it was disabled, never upgraded to 8 because change log looked boring. I assumed it still was. I will try tonight if this works, it will be better then other methods for sure. ;)
Thanks for verifying temps and PPD, gives a good indication it works.
If you guys want a guide for -gpu flag it's the exact same proceedure for the ATI which I believe is even in the official folding Wiki.
Why am I the only one that took a drastic PPD hit with 175.19 drivers? FAHmon stabilized at 3233 PPD.. when I was getting 5000 before. Now I don't know if it's because of the new WUs or because of the new driver - crap.
I never walked away from the 35's one I got them working and I'm still getting 4K just a lil over atm with project 5003 on my 9600gt. Your gts should do better then that, idk what causes it but you're definetly not under optimal performance atm.
Just joined you guys but it seems like only my CPU is working.
My performance for a 8800GT is only 59.94 Iter sec, with a E8500 at stock
I used the drivers and client that you said here.
Sounds like you got no CUDA enabeld drivers. Try the 177.35 international version directly from nvidia and reboot and restart the client. Iterations should be 2k+.
Yeah I know, I got around 2500 iterations or 4980 PPD with the 174.55 and the old small WUs. Now I'm down to 1700 iterations..Quote:
Originally Posted by Marvin_The_Martian
I tried every 177.x with modded infs but they wouldn't work (installed in like 2 seconds and after reboot no drivers were installed). XP x64... I don't want to go back to 174.55 because I got some problems with games. Sucks :/
Lol I have fired up crysis a couple of times to see the diffrence this sub €100 card makes against my at the time €270 7900gt but that's it. And crysis ( atleast the first to levels ) worked flawlessly on the 35's on vista x64 using the 64bit client ( 32 bit for some reason just hangs on startup, idk could be the drivers? ). I read about the game issues, I just don't game enough to have noticed something. Crysis,Sins and Vegas2 work, idk if they will crap out after a long session but my short one's haven't shown instability. Guess I'm lucky uh :)
I installed with the modded inf the first time, just got 60 iterations and rt showed my clocks were brought down to 500/800/500 or something and it didn't allow overclocking nor showed the shaders slider under the overclocking tab. I treid allot off 177.35's linked from various forum threads but none solved the issue untill I downloaded the international version from the nvidia site. Now I can oc again and it's chugging along at 750/1875/950 and 67c on 50% fan speed.
Did all of that and still the same Iterations, i also noticed that i have 1 CPU Core at 100%.Quote:
Originally Posted by Marvin_The_Martian
Installed also the Beta8 for F@H.
Here is a SS
http://img53.imageshack.us/img53/303/cudaissuesaj3.jpg[/URL]
[URL=http://g.imageshack.us/g.php?h=53&i=cudaissuesaj3.jpg]http://img53.imageshack.us/img53/303...c6bc440ca1.jpg
Do you have a taskman shot? Could help, taskman and gpuz. The problem I had was that my card was severly underclocked by the drivers I treid first, and I only noticed when rt showed a 500/800/500 setting.
After installing the international download x64 vista drivers rt showed correct stock clocks and allowed oc'ing again, but I treid 3 diffrent downloads before that. The one which worked for me was NVIDIADisplayWinVista64(177_35)Int.exe size 81.408.768 bytes. The other two downloads both gave the same problem. Beta6 here though still btw idk why. If it aint broken don't fix it?
Edit: news belongs in this thread, but I don't want to make a big fuzz of it so this edit is all there will be.
http://www.xcpus.com/forums/folding-...e-nv-gpus.html ;)
Ok maybe for some this will deserve it's right for a seperate post, if you're repeatedly having issues and haven't updated to the new 1.06 core, manually pause or stop the client ( idk for sure, I'd quit it entirely ) and delete the fahcore file from the work folder. It will download the new version from the server and continue your wu, and for some it has been reported to help a whole lot.
Source
Would the same work for 2 9800GX2's?Quote:
You can run more then one gpu with the -gpu x flag to the shortcut where 0 is primary and 1 is secondary.
I think so, but gpu 0 and 1 are on card one and 2 and 3 on card two. Report back if you get it to work using the link I posted above :)
Why is this?Quote:
Originally Posted by angra
There is a DC project which is supported more vigerously? No that's not true..
Ok let me ask you this... look at this forum sections leadership, UnG and Pythagoras. Have you seen em? I haven't in a long while. I like them, I appriciate their opinion, but I don't feel lead/helped whatever not because their less then lets take Movieman, but rather their absence. When I read the wcg forum, there isn't a thread I don't feel Movieman's pressence in. UnG had a good stretch to but gone belly under without notion just when I felt he was getting everything under control.
Idk what causes this to happen, nor do I know what to do about it. Ask them to come back? Think I've done that already to no avail. So, I don't know what to do, and you?
This client, enabling distrubuted computing on GRAPHICS cards, used to bench here all the time, and not one top bencher has been persuaded to do something about it? Missed chanche for the project and distributed computing/maybe even overclocking because a mutual combined effort would make all involved parties win. I know, the ati client was first but this was so damn buggy in my experience. Maybe when the multi gpu client will be availbe ( Vijah posted days rather then weeks ) and the ppd output fluctation will seize with a stable client, benchers will also pick it up as a nice way off competition/flexing muscle as well as helping a noble cause.
Oh, it's just less popular and has less 'prime movers'. I thought he meant people had something against it or something like that. :confused:
Finally got it all sorted out using XP32
- 177.35 outperforms 175.19
- 175.19 installs without anything special..but a performance hit
- F@H Mon 2.3.2B works ok...When using this go to:
Documents & Settings / Applications /folding at home
- Rivatuner works great, ATI Tool doesn't interact with 177.35 so well
- My 8800GT's both get 5,000+ PPD at 680 Core on Q6600 at 3.2Ghz
Thanks to all for helping...
:up::up::up::party::party::party::party3::banana:: banana::YIPPIE::YIPPIE:
Let's say allot off people I know at one time had a run in on the official forums. As a tech savy team, and fah allowing a level of tinkering, their been accusations in the past of cheating but that wasn't specific to xs as with that other dc disaster ( majestic 12 I think ). Some mods there have a real bad name here, though atm the folding forums I think have a friendly atmosphere ( mind you my expirience there is now maybe 12 posts and one day as of this moment ).Quote:
Originally Posted by bowman
It's just that it's the official forums, and they do not officially support tinkering with flags ect so allot of people I knew who wanted some help were turned down at the time on principal.
Also overclocking and fah... Allot off eue questions were sufficiently answerd with 'is it oc'd?' no matter if you're 48h prime stable or have folded at those clocks for months already.
Maybe the new look made them abit more flexible, and I do know that if you keep things in pm you can get more done then if you do it publicly in these cases.
Edit:
Good news DAK, are you still running a cpu client? If you are, 7im posted my experience about vista allowing my smp to still get 97% of cpu cycles seems to be normal of the os ( vista x64 ). So even two faster cards would take maybe ten percent a piece still leaving an smp capable off 1K+ easy ( low estimate ).
Oh, right, I get ya. I had one person mouth me off for 'ruining the project' by running 177.35 drivers on an officially unsupported card (because I believe it's all a load of marketing bahookey from Nvidia, a way to sell their newer cards even though G80 supports the same features). That was one person, though, and he wasn't staff or anything. Just a troll to me. :yepp:
all of the history is (last I checked), still on this board. there are links to discussions on official forums as well as people stating their own feelings directly. I don't want to add my own commentary except to say...if you want to read about it, look back through the posting history and find the big gap in posting times...just before the forum went dead for a long period.
holy crap... I jus noticed Angra's sig... broke 11mil today, and cleared the top 50 overal user rank.
http://folding.extremeoverclocking.c...=Angra&t=36362
http://www.scootplace.net/smiles/original/218.gif
Nobody buys new cards for newer drivers, and NVIDIA would be retarded to think that people would.Quote:
Originally Posted by bowman
The fact is it takes time to qualify all the drivers that go out, and that means running a lot of software on every single GPU product that the drivers support. That's a LOT of time and resources, and sometimes new product launches wont wait, so they launch the drivers to support the minimum amount of products possible.. only the ones that have been tested (the new ones). Full drivers that support all products usually follow soon after so there's no reason for that kind of slanderous accusation.
Is this OK for you? i dont know what you mean by taskman.Quote:
Originally Posted by Marvin_The_Martian
http://i28.tinypic.com/29usjkz.jpg
I've finished some WUs but all on the CPU
does the ram memory on the card make much if any difference?
has anyone tried the 1gb 8800 gts cards?
Memory doesnt matter much for F@H, Overclock the core a moderate amount and go all out on shader OC for maximum PPD. :up:Quote:
Originally Posted by Sickapotamus
good to know. 8800gt ordered ;)
Yeah it will do thanks, I can see your gpu2 core isn't running though is it able to get work units at all? How did you configure it?Quote:
Originally Posted by metro.cl
The itterations are way to low to, I'm well over 2K with a lesser card, but it looks like you got the right driver version I'm just not sure about the localization playing a role. Like I said I had to download the international version before it would work as it should.
Btw, keep track off new beta's, allot of the issues already talked about are resolved so having the newest beta will shield you from allot of known issues.
Edit: if you're not able to connect or anything, or never have seen the message change from 'Core is not running', post your fahlog ( between code blocks to not have my scroll wheel die prematurely.
I saw you said you got your setup all ironed out and working. I'm getting some strange behavior on mine with huge variations between frame times. Did you install any modified INF file with the 177.35 drivers?Quote:
Originally Posted by DAK1640
Here's a sample of my log file...
Quote:
[07:15:48] Project: 5214 (Run 9, Clone 97, Gen 0)
[07:15:48]
[07:15:48] Assembly optimizations on if available.
[07:15:48] Entering M.D.
[07:15:55] Working on p5214_supervillin_e1
[07:15:56] Client config found, loading data.
[07:15:56] Starting GUI Server
[07:17:22] Completed 1%
[07:18:48] Completed 2%
[07:20:14] Completed 3%
[07:21:41] Completed 4%
[07:23:21] Completed 5%
[07:25:22] Completed 6%
[07:27:41] Completed 7%
[07:29:44] Completed 8%
[07:31:17] Completed 9%
[07:32:44] Completed 10%
[07:34:10] Completed 11%
[07:35:37] Completed 12%
[07:37:03] Completed 13%
[07:38:29] Completed 14%
[07:39:56] Completed 15%
[07:41:22] Completed 16%
[07:44:29] Completed 17%
[07:47:57] Completed 18%
[07:51:24] Completed 19%
[07:54:41] Completed 20%
[07:57:46] Completed 21%
[08:00:14] Completed 22%
[08:03:20] Completed 23%
[08:06:27] Completed 24%
[08:09:23] Completed 25%
[08:12:15] Completed 26%
[08:15:01] Completed 27%
Are you a beta team member?
Edit: The inconsistent frame times are normal and could point to heavier gpu use by another processor or another process keeping a cpu core bussy so it can't feed the gpu consistently.
I have no idea what in the world you are referring to, a beta team member of what product?Quote:
Originally Posted by Marvin_The_Martian
This computer is a dedicated cruncher so there are no other processes running. Overnight it settled at a consistant 2500ppd which is half of what Dak was getting. Since this particular computer is using the same OS, video card and driver he was using I was hoping to figure out why the difference.
OK, the SMP client was conflicting even though I had the GPU client set to a higher priority. So now I have 1 core running only the GPU client and 3 other cores running the SMP client. Dang, I need to get a copy of Vista for these XP machines...
For you guys running headless computers, what do you use to maintain/monitor your systems being that remote desktop crashes the GPU client?
I'm about to give up on the GPU folding. I can't get the client to run properly no matter what I do.
OS...XP 32bit SP 3.Code:
--- Opening Log file [July 8 23:53:52]
# Windows Console GPU Edition #################################################
###############################################################################
Folding@Home Client Version 6.12beta6
http://folding.stanford.edu
###############################################################################
###############################################################################
Launch directory: C:\Program Files\Folding@home\Folding@home-gpu
[23:54:12] - Ask before connecting: No
[23:54:12] - User name: freeloader (Team 36362)
[23:54:12] - User ID: 3B379E125871B1B9
[23:54:12] - Machine ID: 2
[23:54:12]
[23:54:12] Work directory not found. Creating...
[23:54:12] Could not open work queue, generating new queue...
[23:54:12] Initialization complete
[23:54:12] + Benchmarking ...
[23:54:12] - Preparing to get new work unit...
[23:54:12] + Attempting to get work packet
[23:54:12] - Connecting to assignment server
[23:54:13] - Successful: assigned to (171.64.65.20).
[23:54:13] + News From Folding@Home: GPU folding beta
[23:54:13] Loaded queue successfully.
[23:54:14] + Closed connections
[23:54:14]
[23:54:14] + Processing work unit
[23:54:14] Core required: FahCore_11.exe
[23:54:14] Core not found.
[23:54:14] - Core is not present or corrupted.
[23:54:14] - Attempting to download new core...
[23:54:14] + Downloading new core: FahCore_11.exe
[23:54:14] + 10240 bytes downloaded
[23:54:14] + 20480 bytes downloaded
[23:54:14] + 30720 bytes downloaded
[23:54:14] + 40960 bytes downloaded
[23:54:14] + 51200 bytes downloaded
[23:54:14] + 61440 bytes downloaded
[23:54:14] + 71680 bytes downloaded
[23:54:14] + 81920 bytes downloaded
[23:54:14] + 92160 bytes downloaded
[23:54:15] + 102400 bytes downloaded
[23:54:15] + 112640 bytes downloaded
[23:54:15] + 122880 bytes downloaded
[23:54:15] + 133120 bytes downloaded
[23:54:15] + 143360 bytes downloaded
[23:54:15] + 153600 bytes downloaded
[23:54:15] + 163840 bytes downloaded
[23:54:15] + 174080 bytes downloaded
[23:54:15] + 184320 bytes downloaded
[23:54:15] + 194560 bytes downloaded
[23:54:15] + 204800 bytes downloaded
[23:54:15] + 215040 bytes downloaded
[23:54:15] + 225280 bytes downloaded
[23:54:15] + 235520 bytes downloaded
[23:54:15] + 245760 bytes downloaded
[23:54:15] + 256000 bytes downloaded
[23:54:15] + 266240 bytes downloaded
[23:54:15] + 276480 bytes downloaded
[23:54:15] + 286720 bytes downloaded
[23:54:15] + 296960 bytes downloaded
[23:54:15] + 307200 bytes downloaded
[23:54:15] + 317440 bytes downloaded
[23:54:15] + 327680 bytes downloaded
[23:54:15] + 337920 bytes downloaded
[23:54:15] + 348160 bytes downloaded
[23:54:15] + 358400 bytes downloaded
[23:54:15] + 368640 bytes downloaded
[23:54:15] + 378880 bytes downloaded
[23:54:15] + 389120 bytes downloaded
[23:54:15] + 399360 bytes downloaded
[23:54:15] + 409600 bytes downloaded
[23:54:15] + 419840 bytes downloaded
[23:54:15] + 430080 bytes downloaded
[23:54:15] + 440320 bytes downloaded
[23:54:15] + 450560 bytes downloaded
[23:54:15] + 460800 bytes downloaded
[23:54:15] + 471040 bytes downloaded
[23:54:15] + 481280 bytes downloaded
[23:54:15] + 491520 bytes downloaded
[23:54:15] + 501760 bytes downloaded
[23:54:15] + 512000 bytes downloaded
[23:54:15] + 522240 bytes downloaded
[23:54:15] + 532480 bytes downloaded
[23:54:16] + 542720 bytes downloaded
[23:54:16] + 552960 bytes downloaded
[23:54:16] + 563200 bytes downloaded
[23:54:16] + 573440 bytes downloaded
[23:54:16] + 583680 bytes downloaded
[23:54:16] + 593920 bytes downloaded
[23:54:16] + 604160 bytes downloaded
[23:54:16] + 614400 bytes downloaded
[23:54:16] + 624640 bytes downloaded
[23:54:16] + 634880 bytes downloaded
[23:54:16] + 639082 bytes downloaded
[23:54:16] Verifying core Core_11.fah...
[23:54:16] Signature is VALID
[23:54:16]
[23:54:16] Trying to unzip core FahCore_11.exe
[23:54:16] Decompressed FahCore_11.exe (1822720 bytes) successfully
[23:54:21] + Core successfully engaged
[23:54:26]
[23:54:26] + Processing work unit
[23:54:26] Core required: FahCore_11.exe
[23:54:26] Core found.
[23:54:26] Working on Unit 01 [July 8 23:54:26]
[23:54:26] + Working ...
[23:54:26]
[23:54:26] *------------------------------*
[23:54:26] Folding@Home GPU Core - Beta
[23:54:26] Version 1.06 (Mon Jun 23 10:53:13 PDT 2008)
[23:54:26]
[23:54:26] Compiler :
[23:54:26] Build host: amoeba
[23:54:26] Preparing to commence simulation
[23:54:26] - Looking at optimizations...
[23:54:26] - Created dyn
[23:54:26] - Files status OK
[23:54:26] - Expanded 40329 -> 244433 (decompressed 606.0 percent)
[23:54:26] Called DecompressByteArray: compressed_data_size=40329 data_size=244433, decompressed_data_size=244433 diff=0
[23:54:26] - Digital signature verified
[23:54:26]
[23:54:26] Project: 5006 (Run 6, Clone 15, Gen 5)
[23:54:26]
[23:54:26] Assembly optimizations on if available.
[23:54:26] Entering M.D.
[23:54:33] Working on 576 p5005_supervillin_e1
[23:54:33] Client config found, loading data.
[23:54:33] Starting GUI Server
[23:56:34] Completed 1%
[23:59:55] Completed 2%
[00:01:33] Completed 3%
[00:03:10] Completed 4%
Quadcore Q6600 (stock speed).
2gb Ram.
Asus P5N-D mobo.
8800GT
What's the problem? That ss can be shot during the up or download when the core isn't working but your log file looks good. Try fahmon instead of the viewer ( as it shows it's currently done with the unit or waiting for a download.. )Code:[23:54:26] Project: 5006 (Run 6, Clone 15, Gen 5)
[23:54:26]
[23:54:26] Assembly optimizations on if available.
[23:54:26] Entering M.D.
[23:54:33] Working on 576 p5005_supervillin_e1
[23:54:33] Client config found, loading data.
[23:54:33] Starting GUI Server
[23:56:34] Completed 1%
[23:59:55] Completed 2%
[00:01:33] Completed 3%
[00:03:10] Completed 4%
I got it figured out. When I ran the Folding@Home installer, it installed the dll files in documents/application data/folding@home. I copied those files over to C:/pf/folding@home and used the -local flag. Works perfect now. Averaging 4500 PPD on the current Protein. :) I guess the F@H client couldn't find the cudart.dll file until I placed it in the local folder.Quote:
Originally Posted by Marvin_The_Martian
Why didn't you just leave it where it was? I know it's a strange location to most but it's actually where ms wants it to go and it makes debugging allot easier :)
You're just a couple off hundred points above me, 500x are all the same size so should be the same ppd and I'm pulling 4K with a 1:43m frametime average ( with just half your shaders ( depending which gts you got )).
Barnacle...Hope you get it sorted out...Yes, I used the modded INF drivers...The 5,ooo+ PPD are my 512MB 8800gt's at approx 10% OCQuote:
Originally Posted by The_Barnacle
The 4700+ PPD are my 256MB 8800GT's at approx 10% OC...I do push the shaders a little higher, but not much.
I'm just trying to find out how good the error checking routine is. Asked about it on the forum, waiting for an official reply. If it's 100%, I'll push it higher since I'm pretty lowish on oc atm, just above a xfx XXX model (750 vs 740 shaders linked giving 1875 iirc ).
Btw, those wu's only use 86mb from your gpu's ram, where is the speed diffrence coming from then :S
When I left it in it's stock location, the client wouldn't start up. I kept getting a "serious error has ocurred, client will now terminate".Quote:
Originally Posted by Marvin_The_Martian
I have an Alpha Dog 8800GT (512mb) with G92 core.
Simply re-start your comp, s/b fine upon reboot
Freeloader if you want to pm or post your fahlogs I'm very interested, maybe it's something that needs to be fixed before the client is taken out off beta?
@DAK1640 are you sure? Shouldn't have to reboot, should run from installment, but the fahlog will tell more about the problem :)
Yeah, it's working now. But I'm not running the modded INF drivers, what exactly do they do? Right now I just manually set the affinity for all of the SMP client threads to cores 0,1,2 and have the GPU client affinity set to 3.Quote:
Originally Posted by DAK1640
Q6600 @ 2.4
GB 8800gt-256MB
OS -WinXp 32
SMP - 1905ppd
GPU - 4712ppd
Modded inf is only for the installer, and only because the drivers originally linked was for the 260 and 280 only. If you just download the official 177.35's you never have to worry about the inf, some people just had used it to install the first version.
OK, cool. I'm pretty happy with 4700+ ppd from a $77 investment :D
add that on to current 6000ppd.........:shocked: http://i58.photobucket.com/albums/g2...illy_nilly.gifQuote:
Originally Posted by The_Barnacle
That is sounding better and better all the time actually. Was thinking I should drop one in my single core box fold6! Except, I don't think the PSU is up to the task :( *edit* just checked the PSU, it is a 450W that has 20A on the 12V rail. Is that enough to run an 8800 with a single core A64, 2 sticks DDR, and a hard drive?
You're outrunning my more expensive ( I hate the local market here ) 9600gt, and you'll start noticing that the other 9600gt's will get a much tougher time in the future compared to the g92's and up.
Though I might run better if you look at watts used per point per day, idk really though but 8800gt has to pull allot more with twice the shader count.
Well done Barnacle...Quote:
Originally Posted by The_Barnacle
I simply run the GPU along with the SMP version..I do NOT set any switches or assign affinity's. It all works well together. :up:
i just set up gpu2 and it seems to go alot faster when fah viewer is minimized/off, is that normal?
Yes. The viewer takes a lot of resources. :)Quote:
Originally Posted by grimREEFER
so, if you're running smp and gpu client on a quad core, what cap is optimal for the smp client?
75% (3 cores for smp)?
No smp all out, with higher priority on the gpu client so it will take whatever it need. No need to worry about things other then that. If you limit yourself to 3 cores you will get less gpu2 points.
ps, reason I actually openend this thread is because I haven't seen any mention of the -forcegpu flag, and I think it should be in here ( though this isn't really a good setup thread ).
forcegpu tells the client to skip the internall gpu test, and for nvidia and ati there are specific core codes ( r600 and g80 for all current cards for both brands ), so if you know your card is folding capable but you get the message that your card is not supported then this could help getting it running :up:
With the Linux Wine gpu setup you have to use the -forcegpu flag from what I was reading but I did not think it was really necessary for the straight Windows gpu client.
Guys,
sorry, not much time atm. Thing is: I got a machine here that is running dual HD3870X2 in CF, and I need to really test it for stability for like a week or so. Now what would be better than the GPU client running on all 4 GPUs :D
Can anyone give me a quick rundown on how to set up the client? Running on Vista (not sure whether it's 32 or 64, gotta check later).
Thanks
Here this should help:
GPU2 configuration guide
Its for Vista and XP, but is/seems a little mixed up. It would probably be easier if you just follow the guide and then we can help you through any specific issues.....:up:
HTH
Assuming you have a Quad Core...Quote:
Originally Posted by grimREEFER
If you have 2 GPU's then:
2 GPU's & 1 SMP yields most ppd
If you only have 1 GPU then...
1-2 SMP's & 1 GPU yield most ppd
Guys, it doesn't work :(
I followed the Stanford website's guide on how to set up the multi GPU client, made 4 shortcuts for the two HD3870X2's, started up the 1st one (- gpu 0 flag) and that one runs fine. FAHmon shows 2400PPD which is about ok for one HD3870 core I guess (Catalyst 8,8).
But when I try to start the modded shortcut for GPU 2 (- GPU 1 flag) it just says my GFX card doesn't support GPU folding and I should use the standard client (lol). The same holds true for GPUs 3 and 4 :(
Running on WinXP 32bit, can't use the Vista install because it's the customer's :wasntme:
I have removed the CF bridge, disabled CF manually (CF even worked without bridge thanks to the Asus Maximus Extreme mobo and its :banana::banana::banana::banana:ing Crosslinx chip) and clicked every "expand desktop on this screen" box I found (8 of them lol, some told me it couldn't do that because that one was reserved fro CF and physics).
Help? :(
Quote:
Originally Posted by jcool
Does it show something like this:
http://i50.photobucket.com/albums/f3...eenShot008.jpg
if it does, use -forcegpu r600 as extra flag.
Sorry Naja just had to finish your sentence for you :up:
Sorry, but that wasn't the finish of my sentence. Right now, I'd just like to know if I'm on the same page......:up:
Lol ok but let's assume you are since well I think you are :fact: :rofl:
Ya, that's it, I think.. stupid ass program, can't they create a :banana::banana::banana::banana:ing installer that maybe says "insert number of GPUs here" and does the rest for itself?Quote:
Originally Posted by Naja002
:rolleyes:
Will try that -r600 flag (even tho it's not an r600? lol) and report back
I know it isn't, doesn't matter. For nvidia it's G80 even if you have a gtx280 ;)
Well, that didn't work. The error message is gone, but I get all kind of weird effects when starting GPUs 2-4. Some start then hang, some start and then exit straight away. Damn beta software...
Ok that's not that good. Can you post the log files from those clients? Though I think it's known that second gpu's sometimes have to be started more then once to get it running, so that's not really uncommon.
The straight exits is something I think we can fix, it might have something to do with the desktop extending that you did idk but maybe the logfiles might help!
And btw, which version are you using?
Quote:
Originally Posted by jcool
I've just been through this stuff and this is what I would do:
Go to Device Manager and verify that each gpu is recognized, working properly and no conflicts. On the drivers tab--update the driver on each card. I don't remember the exact steps, but do it manually and direct it to the driver that you are using.
Reboot
Uninstall the GPU2 client and reinstall it (reboot in between if ya want)
That should get rid of the popup about the driver. You'll probably have one or more other issues before all is said an done. Let us know and we'll go from there......:up:
^^^If the above does not work--try removing the -gpu 0, -gpu 1, etc. flags for now.
Check out this thread when You get time:
Need Multi-GPU Help......56K!
Long, lotsa screenshots, but may help out......:up:
EDIT: Marvin has a Good Point about Extending the desktop. Its is necessary in Vista, so You will need to do it. You will mostly likely need a VGA Dummy, so here ya go:
DVI to VGA Dummy.....56K!
Very easy to make and cost effective. You will need one for each gpu. Also checkout the Multi-GPU thread about extending the desktop.
The clients all start now, I can set them up, then they download the FAH 1.1 core and then they exit saying they encountered a grave error. This happens for GPUs 2-4 consistently. GPU 1 (aka 0) is crunching fine but only giving 2300ppd according to FAHmon.
All 4 cards are listed in the device manager, all have the current driver installed (well naturally) and I'm on Win XP so I don't need VGA dummies.
Sorry guys, that's it then, I don't have hours and days to try this, either it works or it doesn't. Clearly, it doesn't. Seriously they should make like a working installer for this stuff, as I did nothing wrong - double checked all the steps, it's a nogo :(
You did extend your desktop onto the other cards right you have to do that for both Vista and WinXP just in Vista you have to trick it into thinking there is a monitor attached with the VGA dummy or it won't activate the card for use.