Comparetti is a standup guy. I sent him a "tech-support" query once at 2AM, woke up the next morning and found his response stamped "sent" at just after 7AM.
For a couple months, the program would suddenly grab one of the hyperthreads on my processor and run it up to 100% usage. Definitely a "bug". I sidelined the program until after Xmas.
Then, I send Alfredo my question. I had crashed my system once before putting it on the back burner, and the crash had made my GUI look like Spumoni ice-cream spilled on a hot sidewalk during a Sunbelt summer.
The answer to the question involved keeping the nVidia graphics thermal sensors and stock fan out of the SpeedFan polling loop, and I could've cared less if SpeedFan successfully controlled my video-card cooling fan, anyway. I don't have an ordinary cooling fan on the video card. So I followed directions and inserted /NONVIDIAI2C as a command line parameter in all the SpeedFan shortcuts.
Since then, it has been marvelous. It has been my ASUS P4P800 instead that has fallen short of expectations, since they only implemented full fan-control for the CPU_FAN header despite the potential resources of the Winbond super-IO controller chip on the board.
THIS is why I'm not all that eager for low-rpm (and low-noise) fans if I can find fans that operate in a wide range with low noise levels at least in the middle of the range. Running PRIME95, Hoyle Casino, MS Fright Stimulator 2004 -- just about anything -- SpeedFan 4.20 is rock-stable.
Take a look at the VLSystem Zephyrus and mCubed T-Balancer controllers -- priced at around $60, and take a look at the review of same. They BOTH still have to go through the CPU to do the polling, they both ADD a hardware level to your system, and they BOTH have early-in-the-game software with bugs in it. Why bother?
Send Alfredo a $10 PayPal donation. He deserves it even more for this: When I mentioned that I would eventually send a few bucks his way, he told me "Don't worry -- you don't have to do it . . "