Hysteresis / Response Time zones for graph curves

[roltairtheprotogen]

There may be a better / already existing way to do this, if so please tell me and sorry for wasting your time.

My suggestion is to add hysteresis zones for graph curves, so for example if a GPU sensor is between 65C - 100C, the response time down should be lets say 2 seconds so the fans aren't running too high for too long,
But, between 50C - 64C, the response time down could be 120 seconds, this would allow the GPU to reach a lower temperature before the fans shut off. More importantly, this also means that if your temperatures tend to creep up slowly you can cool it down for a longer period of time before the fans shut off instead of just turning off and on again every few seconds.

This is particularly GPU focused, due to them having a minimum fan speed, and having them turn off and on every few seconds can get annoying, but could apply to the CPU aswell.

[Rem0o]

You can kind of do this effect it with by adding a mix and a trigger. You mix the trigger + your existing graph.

Set the trigger like so:

This will make it so the mix will output the trigger function at 40% if the gpu got to 45deg or + for 120 second, and hold this until the gpu goes back down to 35 for at least 120 second.

You'll only get this long lasting effect thus if the GPU had a sustained load were it will get some heat soak, and it will get that long 120 second cool-off minimum.

[roltairtheprotogen]

Oh that appears to work!
Thank you so much for the help, but maybe the hysteresis zones would still be a useful suggestion? It's not a priority as this method works, but this seems a bit overly complex for the same outcome.

[Rem0o]

You are the first person to raise that idea up tbh so I've never really though about it. There might be some other methods of combining stuff I haven't thought about (yet). This is a very niche feature so I don't think making it part of the "usual" single graph workflow to be the way forward, as it would add complexity and visual clutter to the 99% of people who won't need/use it.

[roltairtheprotogen]

Understandable, I can see how it may confuse users, and I did somewhat think this was a niche issue.
There most likely is a better method of getting this outcome already, but maybe trying to find a better curve suggestion to deal with creeping temps / heat saturation should be researched as I can see it being of use for users who want to "purge" their heatsinks in a way if their temps repeatedly creep up to and hang around at just where fans get introduced.

[Rem0o]

Well the asymmetric hysteresis by itself does the trick. Put a long "down" value and you get that effect. If you were gaming for an hour with your GPU maxing its curve, having it for 2 more minutes after your session at max ain't much of a problem if you weren't bothered by it for an hour straight.

Now that I think of it, some user suggested something similar, but was "per fan focused". So you would set a "residual" % and time value to your control card before it could go to 0%, whatever the curve/input is on that control.

[roltairtheprotogen]

Yeah.. it does make more sense for gaming. My focus was on dealing with creeping idle temps while having it respond to actual loads in 1-2 seconds both up and down but yeah you could just have it run for an extra 2 minutes the entire time.
Some kind of "residual" system would probably work where if the curve starts the fans it then has to reach under lets say 40c and stay there for 120 seconds before the fans can stop.

I'll leave this discussion open for the time being to see if anyone else has any better method suggestions for this.
Again, thank you for the help!