At the moment this problem of the OC and temperatures is causing me great problems. there are already two RIGs, the rig of the RTX there is a card, which is more in the background that even reaches 87 and turn off the miner in security.
But I start observing the FAns in GPU, and I see that suddenly I'm at 84 and fans are lowered to 65% because it has picked up the temperature of another card, according to my rules at 84 degrees it would go to 90% fan, and repende is lowered to 65%, because that card reaches the maximum temperature of 87 and turns off the miner.
I hope you have a solution or I will have to drive MSI aftherburner again. But I really like this native solution. You should only take the value of the hottest card in the RIG and do not go changing card reading
At the same time I'm going to have to leave the fixed fans something I do not want, but of course it's a big mistake that I take the temperature data from different cards, so it can not be efficient.
Please, if you are not going to fix it, tell me already and I will change to the old system with Aftherburner or a fixed fan.
One of the two is wrong. Right now, like this one, you do not look at just one card, you look at all of them. Then a card, the hottest one that should be sent, is activated at 90% fan at 80 degrees, but now it turns out that another card of the same rig reaches 75 degrees and I lower the fans to 70% .. Is it more important to reduce 75 degrees or reduce 80 degrees?
Therefore the concept of temperature for a rig, should take into account only 1 card, the hottest. Because the change of speed is applied to ALL THE RIG. If the speed of the fans was independent by card, his vision would be accurate, but as it is not so, it is wrong.
I suppose I could use Atherburner only for the fans, its solution to put rules that jump at different temperatures is not valid. It can not be that I have a card at 84 degrees that would correspond to 100% of FAN, and suddenly because another card of the rig reaches 75 degrees, I lower all the rig to 75% of Fan. Obviously the card that is in 84 degrees will rise and to 86 I turn the miner to avoid problems of breakage by temperature.
Here there are only 3 solutions:
1- that the rules affect each card independently, it would be very good.
2 - Always take the value of the hottest card, this would be the optimal point, that always in the rules send the FAN the card with the highest temperature. This would be the best solution and faster in programming.
3.- re-use Aftherbuner at the same time as the native, only to control the fans.
You understand that as it is now, even with rules, if I have not activated the heat protection rule, I could have damaged cards. I like your native OC a lot, but the temperature should send the card with more heat, whatever it is inside the rig, and just follow that. If suddenly another card in the same rig exceeds the previous one in temperature, then that new card is the one that would take control of the entire RIG. TAl as it is now, yesterday I skipped the rigs every few minutes and they stopped. I could not give credit to my eyes, to see cards at 84 degrees to 100% and suddenly went down to 65% because another card had reached the goal of a rule and applied to the whole rig, that does not matter as you explain it But it's not good.
You give a greater control of fan as aftherburner or you guide by the card of more temperature of the rig, because if not, cards of those people who do not have rules of stopping the miner can be broken if it reaches 86 degrees as I have.
This kind of situations did not happen with the complete solution of aftherburner, although the connection with the main program is very bad, so I will use the native and I will see that I think so, use a fan speed curve in aftherburner, but this is already ahead of time It will be a problem for many as soon as the heat arrives.
The Device Temperature trigger is currently looking for the GPU with the highest temperature. If your trigger condition is 80 degrees, it will trigger as soon one of the GPU's hit this level. Any action will be performed on the miner itself.
If you have multiple Device Temperature triggers and two of them have met the conditions at the same time - there are no concept where Awesome Miner makes a guess which one of you actions that would make most sense to run - it will simply run the defined actions.
I think what you are asking for is basically a more specialized way to setup the clocking based on conditions, and the rules are very generic purpose way of doing operations and may not be refined enough for this level of per-device control.
As pointed out earlier, making the rules more device-aware could be a way forward in the future.
Update: Please also see my answer below that is a bit more solution-oriented.
You understand but half. If AM only takes the value of the hottest card, whatever is inside the rig, problem solved. You know it but you do not say it.
But right now I do not care anymore. The fan control system in AM does not work well and works perfectly Aftherburner, I do not change anymore. I stay with Aftherburner for the fans, since it takes into account each card by independent and can be curved. I tried to emulate the curve with rules, but as any card triggers contrary rules, it is a disaster.
To me now if it fixes or I do not care, I will not use it anymore. Fan control only by aftherburner.
I appreciate the feedback, but there is no need for this attitude.
I'm very well aware of that if rules in Awesome Miner execute the actions on a miner level, not for a specific device. That was exactly my point above - to explain how the rules works today and why it's not a perfect fit for this scenario. So I do acknowledge that for the scenario you describe, the rules concepts are not yet good enough.
This was also why I referred to my second post that included a proposed solution to be developed.
I just tried a rig with native OC + aftherburner for the fans and it's another world. The much lower temperatures.
The temperature change is such that now I can increase the OC, because the temperatures are very low.
It has to improve the temperature system of native OC. I already happened to Aftherburner to control the fans. We are talking about almost 10 degrees less per card with a car curve in Aftherburner.
Is this with the "Automatic fan control properties" in the Properties dialog of MSI Afterburner? I think this is a pure software periodic check for temperature that set a fan speed based on the curve - without being a feature in the nVidia drivers itself. Defining a high fan speed here will of course result in a lower temperature.
Awesome Miner can probably be extended do something similar. The question is where it should be applied. If it isn't a feature of the nVidia drivers, Awesome Miner would have to check for the temperature all the time and compare it what you defined in your profile. A specific GPU is however not linked to a clocking profile, as you can apply any clocking profile to a miner while it's running. So there needs to be a new concept of last applied profile. But what if you apply a clocking profile to a miner where the new profile no longer have any fan/temperature curve? It's getting complex quite fast here.
I see that MSI Afterburner only do this fan control on a global level, completely outside their profile concept. That's a quite easy approach but may not be flexible enough if you run a mix of GPU's where you want different fan/temperature settings.
A solution somewhere in between would be to introduce a new rule action that you can schedule to run every 15 seconds, where the action will apply different fan speed settings per GPU depending on a fan/temp curve (or table) you define in the rule. That would be a solution outside the clocking profile concept, but maybe easier to both implement and configure.
Please do not say those things. With aftherburner takes the independent temperature of each card and increases or decreases the fan for each graphics card. So if one is hotter the fan will go faster but if one is cooler, the fan does not turn up. In the end it is less extress for the fans, less noise and a much more exhaustive control. It does not have to go faster
The problem in AM is that a card reaches 83 it gets ALL the RIG to 100% of fans although it is not necessary, but then suddenly one of the cards reaches 75, or 70 degrees and makes the speed drop to 70% but there is a card at 84 degrees that will reach 86 and stop the rig.
Aftherburner does not make the fans go faster, gives each card the amount of fan it needs, independently of one another within the same rig, something that AM can not do and I do not care as I do or do not. I think Aftherburner gives a control precision that is going to be very difficult for you to equalize.
I always appreciate all your work and I am always grateful to you, but I think you should recommend Aftherburner to the users and leave the fan control for aftherburner. And I repeat, the fans are not faster, and in some cards I have dropped to 10 degrees, in others the normal is from 4 to 7 degrees, and each card at a speed automatically dispends according to my curve. Do not try to discredit Aftherburner by saying that it will go faster, it goes fast on the card you need, on the others you do not.
The future solution I was considering for Awesome Miner above was the following (not yet supported): "...apply different fan speed settings per GPU depending on a fan/temp curve (or table) you define in the rule". Isn't that exactly what MSI Afterburner do? Per-GPU control of fan speed based on monitoring of the temperature for each individual GPU.
If Awesome Miner would apply per-GPU control of fan speed based on a fan/temperature curve/table like MSI Afterburner do, the result should be the same. Some GPU's will have lower fan speed while other GPU's will have higher fan speed - depending on temperature. Just to make it very clear again, Awesome Miner cannot do this today - all I'm discussing here is a new feature to address your initial request.
If a GPU runs at a specific clock, voltage, power limit and fan speed - shouldn't the temperature be the same no matter if Awesome Miner or MSI Afterburner set those parameters? If the mining software/algorithm, clock, voltage and power limit remains the same, isn't it mainly the fan speed that will impact the temperature of a GPU? So reducing GPU temperature is a matter of fan speed control (= higher fan speed for some GPU's) at this point.
So let's work together on making the new per-GPU control of fan speed based on temperature (using a curve/table) a good new feature! Thanks!
