My cc_config file includes the option <use_all_gpus>1</use_all_gpus>
If you have more than one GPU. FYI.
Printable View
My cc_config file includes the option <use_all_gpus>1</use_all_gpus>
If you have more than one GPU. FYI.
cc_config and app_config files are in the proper places.
Yes, installed whole 131mb AMD driver package with the catalyst control panel.
Just restarted and all the GPU WU immediately went to computational error..Weird sheet going on here..LOLOL
Lets start at the beginning, is this exactly what my app_config.xml file should have in it to run 3 GPU on an ati 5870 card and the rest on 2600K cpu?
<app_config>
<app>
<name>hcc1</name>
<gpu_versions>
<gpu_usage>0.333</gpu_usage>
<cpu_usage>1</cpu_usage>
</gpu_versions>
</app>
</app_config>
Quote:
Originally Posted by Movieman
MM, Your app_config file looks OK. It should provide you with 3 ATI_GPU WUs and 5 CPU WUs. Please post your CC_config file as well so we can peak at it and look for that missing start tag.
Open the app_config.xml file for that rig in notepad and re-save it as app_config.xml. Make sure that you use the ALL Files save option. Have Boinc re-read config file and see if that makes a difference. If still no joy, shut down rig completely, leave of for a few minutes and then re-start the OS and BOINC.
Oh yes... beware MM, using your sledge hammer in close quarters, in sleek sheet, could be very messy...!!! :ROTF:
Touche! :rofl:Quote:
Originally Posted by jeanguy2
I just used Old Chaps cc_config file verbatim
<cc_config>
<options>
<report_results_immediately>0</report_results_immediately>
<max_file_xfers>21</max_file_xfers>
<max_file_xfers_per_project>20</max_file_xfers_per_project>
</options>
<log_flags>
</log_flags>
</cc_config>
Yes, files saved as "all files"..Sorry for the delay, I slipped in de sheet swinging da hamma! :rofl:
MM,
Your CC_config looks good to me. The computing error you see in BOINC manager should also show up on the WCG "Result / Error" for that rig online. If they do click on the "Error" location in the listing. Another window should open up listing the name and/or error code encounter. If more than one listing list them all.
Also could post the top portion of your log file up to where the crunching entries start. If you exit and re-start BOINC, you should get a short and fresh log for us to view.
I suggest you try to uninstall BOINC 7.0.59 and installing 7.0.58. If still having problem then I would uninstall AMD CAT 13.1. I would then do an AMD display driver cleanup with either Drive Fusion and/or Drive sweeper 3.2.0 followed by a full re-install of AMD CAT 13.1.
Good luck MM...
Npw I really appreciate the help but damn you really want me to work don't you?Quote:
Originally Posted by jeanguy2
Whay don't I just format the drive while I'm at it JUST to make sure!:rofl:
Seriously, consider this: RIGHT before I made the video card change that machine was working fine on WCG running one GPU WU on a nVidia 460GTX
Also consider that the ATI 5780 has been working fine up to and including friday when I took it out.
I didn't drop it so we should be safe to assume the hardware is good agreed?
Then I copied/pasted the app_info.XML file that was working fine across the network to the machine the ATI card was goiung to..
Bottom line is I can't see what is wrong but my nexct attempt is to uninstall boind totally, make sure nothing is left on it and then yes, I'll try 7.0.58 and see what happens
Did you uninstall the old Nvidia drivers? Sometimes Nvidia and ATI drivers don't play nice together when the Nvidia card has been removed. And FWIW you don't need all that extra stuff from the ATI driver package. Video driver, audio driver and APP SDK runtime is all that's needed. :yepp:
I do hope you meant app_config.xml. :rolleyes:Quote:
Then I copied/pasted the app_info.XML file that was working fine across the network to the machine the ATI card was goiung to.
Yes, I did..:DQuote:
Originally Posted by nanoprobe
app config files from the two machines..Top one is on the 2600K, supposed to do 3 GPU WU and 5 cpu WU. Bottom one is on the Veast doing 12 GPU WU and 20 cpu WU:
http://img194.imageshack.us/img194/7173/appconfig.jpg
cc config files from the two machines..Top one is working fime on the Beast,bottom one is on the 2600k machine:
http://img42.imageshack.us/img42/135/xxconfig.jpg
MM, I'm curious as to why you're running GPU and CPU tasks on the same machine? Are the CPU tasks from another project?
He has 32 cores to play with and a 7870 on one machine and 8 cores and a 5870 on another machine. on the former there are 12 cores dedicated to gpu just now and on the latter 3 cores on gpu
Gotta keep the others busy so they work HCC as well
MM,
I know that my previously suggested corrective actions involved a bit of work on you part. Here is my reasoning for my previous suggestions. They were listed from least work to most work involved.
1- Re-save or re-create app_config file - I have had one experienced where an app_config.xml file got corrupted transferring it from one rig to another. After many frustating problem solving hours re-creating it corrected the situation.
2- Switch from BOINC 7.0.59 to 7.0.58 . BOINC 7.0.59 is an unknown entity for me I am successfully using 7.0.58 and I know that it works.
3- Re-install CAT 13.1 after full uninstall and Nvidia and ATI display driver sweep with Drive Fusion and/or Drive Sweeper. As stated by Nano, switching a rig from Nvidia to ATI/AMD video drivers or vice versa can leave many entries in your file folders and OS registry that will cause you problem(s) after the new install. This is specially true when dealing with older video drivers and OEM un-installing applications.
In summary MM, by switching from an Nvidia to AMD video card & driver you have made a major switch to that crunching rig that could have left troublesome remnants of the previous Nvidia video driver behind. This may require you to do a full set of video driver sweep (Nvidia + ATI/AMD) to rectify the situation.
For your consideration,
Jean-Guy
MM,
Also suggest that you continue your trouble shooting in the "app_config" thread located here The "Working app_info" thread you are using is dated and no longer valid. It must be removed from our stickies and replaced with a new App_config thread. The app_info sticky is providing first time GPU cruncher with invalid info and much duress specially if we keep it alive with new "app_config" related posts.
Thanks guys! I will figure it out.,.:D
That's the attitude. A little trick I figured out for modifying the app_config file is make a Shortcut of the file and move the Shortcut to your Desktop. That way, while you're modifying the file, trying to figure out what is the best settings for your computer is, you don't have go looking for the app_config file. Just click on the Shortcut on your Desktop. It saved me time and frustration.Quote:
Originally Posted by Movieman
Quote:
Originally Posted by Gandalf
I find it easier to just toss the machine out the front door and start over! :rofl:
What is your address. I'll be out front to catch it. Especially if it's that 32 core beast. I can always use another cruncher.Quote:
Originally Posted by Movieman
Beast is fine and that sucker is screaming with 12 GPU WU and 20 cpu WU..I'm thinking of going to 16 GPU WU as my cpu load never shows more than 81%Quote:
Originally Posted by Gandalf
A yardstick that I use to determine if I have a good setting in the app_config file is how much time does it take to complete a GPU task? If you increase the number of GPU task running but it takes longer to run each task, then have you really gained anything? And, of course, if it starts causing errors, then you know you've gone to far. Been there, done that.
Agreed.. Hey, can you guys send back work right now? I can't get through. I have a ton completed and a huge amount on hand so I'm not worried about running out yet..Quote:
Originally Posted by Gandalf
can't access any of the WCG pages inc the forum.
OK here, but saw something similar reported ....since then things are good again... http://www.worldcommunitygrid.org/fo...e,yes#lastpost
(Maybe you can see that when you get connected again) :rofl:
If you intend to change the number of GPU wu's running I will make a suggestion:
<app_config>
<app>
<name>hcc1</name>
<gpu_versions>
<gpu_usage>0.0625</gpu_usage>
<cpu_usage>0.5</cpu_usage>
</gpu_versions>
</app>
</app_config>
This one runs 16 wu's on the card but only uses 8 cores to do so so you free up 4 cores to run cpu.
an alternative might be:
<app_config>
<app>
<name>hcc1</name>
<gpu_versions>
<gpu_usage>0.083333</gpu_usage>
<cpu_usage>0.666666</cpu_usage>
</gpu_versions>
</app>
</app_config>
This one is running 12 on the card and using 8 cores to do so.
Works for me, too. Maybe they just don't want to talk to you... :shocked:
works now..Just saw my totals on the Beast go up by 12,000 BOINC...:D
edited above to show change to GPU wu's running