The "-f" flag

[detroit]

I'm using GUIMiner on an ATI 5870 with SDK 2.4.
I've wanted to configure a fallback pool, so I set my primary pools miner to use "-f5" and the backup to have "-f50".  It all seems to run fine, producing a miniscule number of hashes on the backup pool if I leave it alone.  When I try to stop the backup pools miner, though, the display driver seems to crash; the work stops, fan slows down and clicking on things on the desktop becomes ineffective, but I can still move the mouse. About 5 seconds later, the display freezes and the only recourse is to hit the reset button. This will happen when the card isn't overclocked.

Is there some detail of using the -f flag that I haven't heard?  What might be causing this?  Thanks for any insights!

[1714778061]

1714778061

[1714778061]

1714778061

[1714778061]

1714778061

[Basiley]

basically its regulate GPU saturation.
more "-f" number is less it use GPU. in my experience "-f" between 60 and 70 is best ones even for main miner.
about crash: cause can be everything - from non-WHQL-drivers or corrupted OS to software/firmware interference and/or anything else like inadequate power supply.

[Lucidize]

I have the same problem quite regularly, on a 5770. Often the display driver just throws up and error which is ok because I dont need a restart but other times it will freeze and I need to reboot to bring it out of the freeze.

I'm using guiminer, also. Is there any way to search for corrupt drivers?

I overclocked with MSI Afterburner and speeds went up over 200 (unstable tho when it it 99-100% gpu usage). Now I just underclock the memory and it goes to nearly the same speed. Not sure why it has been so unstable, I was thinking of reinstalling drivers and afterburner to see if it helped. Any advice appreciated 

[FreeJAC]

I don't know how to use the "f" flag in the context you describe although I have heard others talk about it so I'm sure someone will chime in. I have been using it to do final tweaking of the hashing of the cards. Basically start at 60 OC the GPU to the sweet spot and then start tweaking the number start at 60, then try 30, 15,10 ,5, 1. As you get closer to 1 your hashes go up slightly but so does your heat. You desktop also becomes less responsive. This allows you to get into that 98% GPU utilization mark. You may not be able to get there because your heat will creep up too much and may have to settle for 94% etc. for the sake of the life of your GPU. 

[Basiley]

yep.
in my experience, GPU saturation between 92%-94% is optimal max for long-term usage.
not only for mining.