Hi everyone,

I joined milkway@home recently and I have a little problem: All the CPU tasks run fine, the GPU tasks using the application "Milkyway@Home Separation (Modified Fit) 1.36 (opencl_ati_101)" also seem to run properly but all the tasks using the application "Milkyway@Home 1.02 (opencl_amd_ati)" fail after a few seconds and give a computation error.

Also, the GPU application that seem to run properly only take around 20s to complete, is it normal ?

Here is a screenshot: http://i1148.photobucket.com/albums/o575/Dadou148/screenshot_zpsc1ooubha.jpg

I am using Windows 8.1 with the latest ATI driver (catalyst 15.7.1). I have a Phenom II x4 955 CPU, ATI Radeon HD6950 Graphics card and 8 GB of RAM.
Tell me if you needmore information.

Hi everyone,

I joined milkway@home recently and I have a little problem: All the CPU tasks run fine, the GPU tasks using the application "Milkyway@Home Separation (Modified Fit) 1.36 (opencl_ati_101)" also seem to run properly but all the tasks using the application "Milkyway@Home 1.02 (opencl_amd_ati)" fail after a few seconds and give a computation error.

Also, the GPU application that seem to run properly only take around 20s to complete, is it normal ?

Here is a screenshot: http://i1148.photobucket.com/albums/o575/Dadou148/screenshot_zpsc1ooubha.jpg

I am using Windows 8.1 with the latest ATI driver (catalyst 15.7.1). I have a Phenom II x4 955 CPU, ATI Radeon HD6950 Graphics card and 8 GB of RAM.
Tell me if you needmore information.

Yes 20 seconds is normal for the ones that run correctly, but I don't know why you also have bad ones. Are you leaving one cpu core free just to feed the gpu?

I saw this browsing thru some other threads: "That's to say, if I want to run MWv1.02 WUs on my hd6970,I have to use Cat13.1 or older Cat driver?" You can get older drivers here, at the bottom of the page:
http://www.hal6000.com/seti/boinc_ati_gpu_cheat_sheet.htm

Are you leaving one cpu core free just to feed the gpu?

I did not do this previously but I just tried it now and it still fails. I also tried to pause all the other tasks and run it, without success.

To me the only possible explanation is an incompatibility with the newest driver for this card. I could not find any log about the error, is there a way to have more information about what went wrong ?

I really don't want to go back to older drivers because I also use this PC for gaming.

If there is no other solution to get these tasks right, is it possible to ask MW@home to not send me these specific tasks but still send the other GPU tasks ?
Or should I just ignore the errors ? I don't know the real consequences of reporting failed jobs for MW@home, I suppose these tasks are just sent again to an other user ? Is this affecting the global computing efficiency ?

Are you leaving one cpu core free just to feed the gpu?

I did not do this previously but I just tried it now and it still fails. I also tried to pause all the other tasks and run it, without success.

To me the only possible explanation is an incompatibility with the newest driver for this card. I could not find any log about the error, is there a way to have more information about what went wrong ?

I really don't want to go back to older drivers because I also use this PC for gaming.

If there is no other solution to get these tasks right, is it possible to ask MW@home to not send me these specific tasks but still send the other GPU tasks ?
Or should I just ignore the errors ? I don't know the real consequences of reporting failed jobs for MW@home, I suppose these tasks are just sent again to an other user ? Is this affecting the global computing efficiency ?

Most Projects do not test the new drivers before they come out, so you 'gamers' that tend to be upgrading to the latest and greatest drivers all the time ARE in effect the testers. I am sure they are at least trying to figure out why it's happening, but have no clue on if they have found anything or how long a 'fix' could take.

As for specifying the units you want...there used to be a way thru an app-config.xml file, but I have long ago forgotten how to do it.

As for "this affecting the global computing efficiency" I wouldn't worry about it, if you don't care then it's not your problem, either the Project fixes it or comes up with a way to give us more control of the types of units we can crunch. Have you checked under Your Account, "Preferences for this project MilkyWay@Home preferences" to see if you have that control now? I don't think I see it under mine.

I could not find this setting either, I guess the user cannot choose what type of task to work on/not work on.

I browsed my completed tasks and after 2 days, I already have 23 of them reported and I can confirm that they are the only ones failing to compute.

I hope it will be fixed eventually, for now I just ignore the errors...

Thanks for your help.

I could not find this setting either, I guess the user cannot choose what type of task to work on/not work on.

I browsed my completed tasks and after 2 days, I already have 23 of them reported and I can confirm that they are the only ones failing to compute.

I hope it will be fixed eventually, for now I just ignore the errors...

Thanks for your help.

Sorry I couldn't stop the errors!!

I have the same problem, quite frustrating.