Frequent VPU Recover events w/ 6.23 and Cat 8.12

[LockeOak]

I simultaneously upgraded to the Catalyst 8.12 drivers and accompanying CCC when I upgraded to 6.23. Ever since I'm getting frequent, random VPU recovers. It doesn't crash the system and I don't lose the WU but it's very annoying to lose control and have a dead screen for 20 seconds. They occur 1-2 times an hour. They don't register in the log file.

XP Pro 32-bit
HD4850 stock speeds
AMD Opty 165

Card temps are in the high 80's (this card runs very hot normally, even with the side of the case off). Any help?

[mhouston]

Can you tell us what core and project you are running? How are you monitoring temps?

[LockeOak]

It's happened with Project: 5737 (Run 3, Clone 70, Gen 12) and Project: 5739 (Run 0, Clone 76, Gen 6) and perhaps the ones before that as well but I don't have the log files. Core is 1.18. Temps monitored with CCC.

[LockeOak]

I just saw the 1.22 core update. Downloaded it, I'll see if it helps.

[LockeOak]

Still occurs with the 1.22 core, running Project: 5735 (Run 2, Clone 47, Gen 36).

[EvilAlchemist]

Card temps are in the high 80's (this card runs very hot normally, even with the side of the case off). Any help?

You may need to increase you Video Card's Fan Speed to help drop the temperatures.

I know in the Vista version of Catalyst Control Center (CCC), there is an option to change fan speed under the Overdrive Section.



There may be something similar in XP Catalyst Control Center as well.

[LockeOak]

I have been playing with manual fan control, the event appears to be temperature independent but I'm not sure. Before the update the card was always around 82 degrees and I never had a VPU recover problem.

[toTOW]

You're not alone with this issue ... I'm seeing the same thing on my 4870 ... (and I was seeing it before on the 3870 that was replaced by the 4870)

[LockeOak]

It's someone replicable, usually occurs within 2 minutes of opening the FAH GPU program.

[mhouston]

Can you guys provide more information? I just can't seem to reproduce this on XP or Vista.

[lordglenn]

im having the same problem.

when i had cat 8.11 I had maybe one vpu crash in a very long time. but after updating to 8.12
I get a vpu crash after I run fah in a short time and it happens every time.

also after the first vpu crash if i try and close and relauch fah the vpu recover will happen immediately and i have to reboot.

I am useing fah gpu console client 6.23

my core temps do not affect this problem It happens even when i set fan to 80% and my core stays below 75c

I have a diamond 4850 and i use default clocks.

so im thinking there is some change in the cat drivers in 8.12 that is affecting fah but thats just my opinion.

im useing core 1.18 and im on project 5738 run 3 clone 73 gen 28
odd i have intel e8500 and fah says board type amd... i have a asus p45 chip board...
c2d e8500
4 gb ddr2 800 cas 4
xp 32bit sp2

also ive noticed that fah only uses one of my cpu cores and maybe its only programed to use the gpu and one cpu core..

-G

[Karamiekos]

I haven't paid much attention to it, but I have XP SP3 32 bit, 2 gigs of RAM and a Visiontek 4850. Cat 8.12 core version 1.22 and the 6.23 FAH program.

Sometimes when I have some of the new work units, like this 5736 I have right now, I have random VPU recovers. I have been having this issue for a while since I tried the beta dlls a while back. I had it clocked at 690/1125, and I thought this was the culprit, but I still get them backed down to 650/1100 which should be completely stable, as 690/1125 used to be completely stable. Sometimes the driver has to try to recover twice.

It seems when I am web browsing and go to click on a website or anytime that demands the card to refresh something on the screen it must conflict with the new code which seems to bring down the amount of CPU required( at least in Vista, don't see a difference in XP though due to how it schedules clocks in task manager).

[Karamiekos]

Oh and my temps stay under 70C for the most part I leave the fan on manual @ 55%.

[LockeOak]

I'm going to agree with lordglenn. It was happening with 1.18 and 1.22, but only after updating to Cat. 8.12. Clock speeds or temps on my HD4850 do not seem to matter, will still happen at stock speed and 60% fan speed. Sometimes [it happened again as I was typing this word!] it will silently crash FAH (the progam appears to be "Working 700/10000" for example but without any progress), other times the program continues with the WU.

[toTOW]

also after the first vpu crash if i try and close and relauch fah the vpu recover will happen immediately and i have to reboot.

What happens if you stop the client again, and try to start it a 3rd time ? Does it resume successfully ?

To all people having this issue : could you post the list of other running applications ? That might help to find some common parameters ...