I'm ready to try anything at this point. Unfortunately I think we need to wait until we're out of S@H v6 work before we try to replace the v6 cuda apps with something that doesn't have dlls.

It'll take months (if not year+) before all the v6 workunits are gone.

Anyway, I mentioned the alternative method just because I noticed the possibility and wanted be thorough.

Releasing 6.08b with DLLs the same way as you do with v7 should (maybe, perhaps, who the hell really knows) work without too much risk in trashing everything.

Did nvidia ever fix the utilization problem it had with its drivers on older GPU's? Perhaps falling back to an older driver set may solve his problem?

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

I'm ready to try anything at this point. Unfortunately I think we need to wait until we're out of S@H v6 work before we try to replace the v6 cuda apps with something that doesn't have dlls.

It'll take months (if not year+) before all the v6 workunits are gone.
...

Perhaps the Sah v6 work should all be cancelled when there are relatively few still unresolved, something like 1000. Then after waiting a reasonable period for hosts to contact the project and get the server abort, running the credit granting script could counter any RAC impact.

                                                                   Joe

With no activity on this thread for 5 days, and no apparent change to the situation (i.e., v7 cuda22/23 tasks still being sent off en masse to almost certain failure), I was about to vent my frustration again when I found this host
http://setiathome.berkeley.edu/results.php?hostid=6989670
with 665 failed cuda22/23 tasks, just on his current list. But then something caught my eye, the two new GPU tasks he just received a short time ago: SETI@home Enhanced v6.11 (cuda), which actually shows up on the Application Details page as: "SETI@home Enhanced 6.11 windows_intelx86 (cuda23)". The next host I looked at for cuda22/23 errors also has recently received 5 similar tasks, with the same name on the task page but "SETI@home Enhanced 6.11 windows_intelx86 (cuda)" on the Application details page.

Could these represent replacements for v6.08 and v6.09, along the lines of Juha's 6.08b suggestion? None of these two hosts have reported back yet for the v6.11 tasks, but this could be an interesting development! I'm certainly pleased to know that at least something seems to be going on, even if it is on the hush-hush. :p

Curiouser and curiouser. I found a couple more hosts with these v6.11 GPU tasks, then checked them again a short time later and found they'd metamorphosed into v6.03 CPU tasks! Not changelings exactly, but they appear to have been resent a couple hours after the original downloads using the same Task ID. Perhaps someone can explain this to me....in the morning. (Yawn! ZZZzzz...)

Hopefully there are also some who've read this thread and followed that same advice (though certainly no one who's mentioned it so far ;-)

Well Jeff, it looks like you've got your first customer!:-D

Was crunching V6 resends exclusively up until (less than) 24hrs ago but finally ran out... Thought it might be fun to crunch the V7 stock apps for a few months (and then try out anonymous), but boy was I wrong.

Thanx for the thread and effort everyone. I'll let you guys know if I run into trouble again. I'm guessing you gurus have all the info you need by now but if there's anything you want me to look at (and relay the info) just let me know.

My one and only little host:
http://setiathome.berkeley.edu/results.php?hostid=5550773&offset=0&show_names=0&state=6&appid=

PS Richard, just out of curiosity, are these the same dlls that caused my ION to run at half speed on stock apps? The ones that 'forced' me to install Lunatics?

PS Richard, just out of curiosity, are these the same dlls that caused my ION to run at half speed on stock apps? The ones that 'forced' me to install Lunatics?

Yes, exactly so.

I am running CUDA22 and CUDA23 jobs on two different machines and all of them go in error.
CUDA42 and CUDA32 NO problem

What can I do to prevent errors or not to recieve CUDA22 and CUDA23 jobs

I am running CUDA22 and CUDA23 jobs on two different machines and all of them go in error.
CUDA42 and CUDA32 NO problem

What can I do to prevent errors or not to recieve CUDA22 and CUDA23 jobs

The easiest way is to run the Lunatics installer.
You can choose which cuda version to run.

---

With each crime and every kindness we birth our future.