Folding Forum

Is it possible to set FAH to pause if the CPU temperature rises above a given level? There are utilities that show CPU temperature, e.g. RealTemp, so it can be read on at least some machines. If not currently possible, please take this as an enhancement request. Thanks!

It's not supported by the standard client at the moment. Although it would be a pretty sweet app to have I can see how it would lead to some pretty serious cycling issues, depending on how fast your cycle it. That is, if you set it at 80 degrees and it tripped, your folding would stop for however long you set the delay, say five minutes, and then immediately restart (because as the system is idling its pretty likely to be below 80 degrees), at which point it would probably quite quickly go over temp again and shutdown again, as the underlying contributers to the high heat (warm ambient temps, ventilation etc) are unresolved.

If you wanted to do it yourself you could write a script that monitored your CPU temperatures and then sent a command to the client to shut down the offending slot, perhaps you would want to include an increasing counter, i.e. the first shutdown was five minutes, but the second within half an hour increases to 10 minutes and so on... Most basically you could write a simple bat file that sends the stop command to the client, which realtemp or coretemp execute when the temps hit a certain point... add a wait and then unpause & quit the bat file.

Modern CPUs already downclock at certain temps to prevent damage. Also check your bios as some have custom temp options there also. This feature is unlikely to get implimented, as it's too difficult to program across 3 OS types and thousands of motherboard/cpu combinations. Please note this type of feature has been a long standing request, so if it ever becomes possible, it's on the list. Other 3rd party apps are the way to go for now.

https://fah-web.stanford.edu/projects/F ... ticket/634 is the ticket for monitoring temperatures.

Thanks! Re: OS types, if this worked for just one OS it'd be a big help. E.g. for Windows, integrating with a 3rd-party app like http://www.techpowerup.com/realtemp/ (if they agree to provide an API or a file output of current temp) might be an option.

Of course, the real solution is to get a laptop fan

But beside CPU also GPU temp would be interesting to measure and to include in the system control. That can get complex/messy to build a generic solution. One of the many reasons I prefer Linux/BSD based systems. With the 3rd party API/telnet and a little script it's quick done.

RealTemp measures GPU as well as CPU temperature.

I don't always monitor my hardware temperatures, but when I do, I prefer Speccy.
Speccy provides temperature readings of CPU, GPU, motherboard, HDD, and a whole lot of other information about your hardware.
http://www.piriform.com/speccy

One fact that needs to be considered: Constantly cycling from above some temperature to "pretty cool" and back is probably harder on the hardware than being several degrees warmer than some arbitrary temperature that you choose. The real probem is that there's no straightforward way to MODULATE what resources FAH uses other than progressively reducing the number of CPUs one at a time. That takes monitoring temperatures under varying conditions and making an intelligent decision that works on your system.

This same issue is commonly discussed for GPUs, where it's even more difficult to modulate anything other than the fan speed.

Isn't there an option to specify what % of a CPU FAH should use?

If you have a CPU Slot which uses only a single CPU, then you can use the CPU Slider. However, if you have a CPU Slot that uses more than one CPU simultaneously, then the best option is to reduce the number of CPUs as bruce stated. Each reduction of CPU will drop a "step" in the percentage, for example:
2 CPUs -> 50% per CPU
3 CPUs -> 33.33% per CPU
...
8 CPUs -> 12.5% per CPU
12 CPUs -> 8.33% per CPU

Changing the number of CPUs dedicated to F@H can be done via the Advanced Control (AKA FAHControl).

There was an option to specify how much CPU is used and it worked fine for uniprocessor CPUs, but if is essentialy unuseable because it modulates only one of them while the other (N-1) still run full blast. There also was an attempt to modulate GPUs by running full-blast for M seconds and then allowing the GPU to be idle for N seconds. It resulted in an undesireable cycling of the temperature.

Generally, the slider can be used to reduce performance to the point that the fans can take care of the heat. (A well designed computer should always be able to keep itself cool enough unless you overclock and then you're taking full responsibility for it).

I personally use power profiles to control cpu clock speed during different folding senarios to limit heat.