Oddiity in Seti Beta

[Pappa]

Okay, as Seti Beta is one of my playgrounds. I have a Machine that has been Seti Beta almost forever.
AMD X2 6000 9800GT
http://setiweb.ssl.berkeley.edu/beta/show_host_detail.php?hostid=22789

While it was doing Astropulse 5.05 and Cuda it was cooking just fine. As AP ran out (Eric is looking at that) it started getting the base MB 6.03's (which have real issues on AMD) and in the process Cuda started getting real flaky. From what I see, 6.03's have a Memory Leak and sometimes fail to properly start. That has been discussed with Eric before. The jist of the Phone Con was that the errors that were reported are being sent to Nvidia. Eric asked if he had permission to give my email address which was okay with me. I have a good test case for reproducable errors (lots of errors).

So I wait to see if I get response from inside, it presumes that some people can put in and things are still happening in the background.

[Raistmer]

6.03 has memory leak or 6.08 ?
If 6.03 so why nVidia involved ?

[Pappa]

This it the bare MB 6.03 in Seti Beta that is causing the issue it is affecting the Cuda 6.08.
While the machine was running AP 5.05 (BTW Thank You Joe, quite an Improvement) there were no problems with running the Stock 6.08 Cuda App

So as I finally caught up with Eric and found that AP's were not being split.. and described the issue again on the AMD Platform (this has been a several month battle).

The 6.03 Intiializes and does not properly Run. I have 100% CPU and no progress. The combination of two 6.03's and CUBA keeps driving to the 2Gig I have into the swap file and requiring a reboot about once a day. I even adjusted the % of CPU down to allow Cuda to continue. It appears taht the meory leak is affecting the memory space/CPU necessary and required by Cuda.

The Good thing is that Eric found some errors in the Stock Cuda App Results, that he is sending to the Nvidia contact. There are enough accomulated errors that it is an issue.