Collatz app 2.0x problems

[NeoGen]

If you get problems with the newest app from Collatz for CUDA, just reset the project or do a detach/attach, because there's one of the files that boinc misinterprets as already present when it should override it with a new version, thus causing instant crashes as soon as you start the WUs.

I had that same problem and had to dig around in the projects boards for the answer. I hope it helps you guys. :)

[vaughan]

OK, been there done that. Now its working but for some strange reason it now estimates that a task takes 147 hours on my 8600GT card. In fact this card does a task in about 23 or 24 hours.

How do I "fix" BOINC so it will get the estimated time right and d/l more than 1 CUDA task for Collatz?

[liuqyn]

don't think you can, but it will correct itself over time.

[AMDave]

OK, been there done that. Now its working but for some strange reason it now estimates that a task takes 147 hours on my 8600GT card. In fact this card does a task in about 23 or 24 hours.

How do I "fix" BOINC so it will get the estimated time right and d/l more than 1 CUDA task for Collatz?

First I will refer you to this link becuase it is a well informed response from a BOINC admin group:
http://einstein.phys.uwm.edu/forum_thread.php?id=7183

Then I will state that "it is incomplete".
To enforce the change (having stopped BOINC altogether on that machine), you also need to apply the same change to the "client_state_prev.xml" file otherwise the client may reject the change and fetch the old value from the "prev" file and initiate a perplexing sense of confusion.

Either way, as the posts say you can lose your entire cache if the boinc client rejects your change. I have tested this to my satisfaction. It is true.

Given the time period issue that you are looking at I would definitely give this a go and be prepared to wear the consequences if it fails. If it were me. (folder snapshot backups are the beez-neezs :icon_mrgreen:)

"Caveat emptor" and all that. :icon_wink: