Slow down a F@H slot

[Pete_In_Syr]

I recently added a second GPU to my system and and am happy with the results.

Question arises about my legacy GPU. It is currently running its little old heart out, but it is now reporting temp right around 100 deg C and fan at 100%. I'd like to allow the card to continue to run, taking the approach of leaving the card configured as it is but possibly adjusting something in the F@H configuration to lower the work being done. Is that possible? I don't necessarily want to disable it.

Thank you

[Neil-B]

You cant as such slow it down via fah as gpu scheduling in os doesnt work that way but you could either reduce the power setting to the gpu or underclock it - the first might be the easiest ... the issue is probably less about it overworking but more about lack of cooling/airflow in you case ... the extra heat from the new gpu will mean the ambient temp in the case is higher and the legacy gpu isn't getting the cooling it used to ... something like gputweak or whatever works for your legacy card should hopefully allow you to reduce the power limit for the card ... I would suggest also look into seeing what you can do to improve cooling as if the old gpu is being impacted the cpu and actually the new gpu may well be running hotter than necessary too

[Pete_In_Syr]

You cant as such slow it down via fah as gpu scheduling in os doesnt work that way but you could either reduce the power setting to the gpu or underclock it - the first might be the easiest ... the issue is probably less about it overworking but more about lack of cooling/airflow in you case ... the extra heat from the new gpu will mean the ambient temp in the case is higher and the legacy gpu isn't getting the cooling it used to ... something like gputweak or whatever works for your legacy card should hopefully allow you to reduce the power limit for the card ... I would suggest also look into seeing what you can do to improve cooling as if the old gpu is being impacted the cpu and actually the new gpu may well be running hotter than necessary too

Thanks, Neil:
Case has plenty of cooling. A full size tower, CPU cooler is a monster, 1 120 fan on input, 3 on output (using the PC to keep my room warm in these cold winter days! ) I'm just not sure of myself when is comes to adjusting the settings on a GPU. I see

[iero]

As Neil said, try to lower the power target all together, since excessive tuning to the settings, may lead to lost WUs.
As an example, earlier today I tried an -32% Power Tuning which led to ~ -8/11% with a decrease of 51W on avg and a drop in temps to go along with it.

As for the process itself, you cant damage the gpu through the software. Worst case scenario, you get a crash and the settings revert back to normal on their own.

If you need more help, let as know more about your system.

[Joe_H]

As described, your setup has not enough fans inputting air, I would suggest adding at least one more.

Adding the GPU is probably blocking airflow to the older GPU, a side case fan directing air towards the GPUs would probably help.

Basically, your fan setup sounds backwards. In general you want more airflow from the fans into the case than out. This helps keep dust out, and any excess air will flow out of the case anyways.

[bruce]

I agree with what has been said. Also, you might try figuring out how to stop that GPU from getting flooded with hot air. A long time ago I had a device that filled an empty slot between two devices that pushed the hot air from one if the devices out the back-plane. Just cleaning up the cables can even remove unintended obstructions but adding intake fans is clearly a good idea.

BTW, I just noticed that the fan on my power supply is not turning.

[Pete_In_Syr]

Thanks, all for the replies. Due to time and priority restraints, I'm not going to be able to do much with the case/fans, etc. or otherwise messing with hardware.
I originally just put this old PC into service to try and contribute to Covid cure effort. I added the second GPU as a means of additional contribution.
The case is what it is, a fairly high end Antec from years gone by. I'm sure the new GPU is blocking some airflow from the old to the way air is pulled out.
I was just looking for the quickest fix, but my hardware finesse just isn't there.
The software that came with the card (an EVGA GTX 650Ti) seems to have some controls I don't know how to get at. They (power target and temp target) are greyed out. The only ones I seem to be able to do much with are GPU and MEM clock offsets, which don't seem to effect much. There's a voltage setting, which is apparently in millivolts

[Neil-B]

BTW, I just noticed that the fan on my power supply is not turning.

... Not Good

having said that I grabbed a psu for the build I have just done that only spins up the fans when needed ... since fan has yet to hit temps where it needs it I have only seen it spin up under test !!

[Neil-B]

The software that came with the card (an EVGA GTX 650Ti) seems to have some controls I don't know how to get at. They (power target and temp target) are greyed out. The only ones I seem to be able to do much with are GPU and MEM clock offsets, which don't seem to effect much. There's a voltage setting, which is apparently in millivolts

Quick google - it appears that the power on that gpu may not be adjustable ... someone will suggest what is best to do with GPU and Mem offsets (I guess lower them as much as possible ... if you can make the case at least two fans in that really should help but understand layout might be problematic ... the other option might be to switch the two gpus around? ... The older card should generate less heat overall so should preheat the newer one less - and the newer one may well have better fans/cooling/thermal paste/pads and therefore be more manageable from a heat perspective ... Obviously only do what you are comfortable with doing.

Hope you get some improvements somehow ... Good Luck

[Pete_In_Syr]

The software that came with the card (an EVGA GTX 650Ti) seems to have some controls I don't know how to get at. They (power target and temp target) are greyed out. The only ones I seem to be able to do much with are GPU and MEM clock offsets, which don't seem to effect much. There's a voltage setting, which is apparently in millivolts

Quick google - it appears that the power on that gpu may not be adjustable ... someone will suggest what is best to do with GPU and Mem offsets (I guess lower them as much as possible ... if you can make the case at least two fans in that really should help but understand layout might be problematic ... the other option might be to switch the two gpus around? ... The older card should generate less heat overall so should preheat the newer one less - and the newer one may well have better fans/cooling/thermal paste/pads and therefore be more manageable from a heat perspective ... Obviously only do what you are comfortable with doing.

Hope you get some improvements somehow ... Good Luck

Great minds think (and seem to come to the same conclusion!). Low end card, very likely not adjustable. Will keep an eye on it so as not to let it become dangerous.

[Neil-B]

It you have the time/inclination you might also want to run just the new gpu for a while and get a feel for what ppd it puts out .. then run both slots and see if the extra heat in the case from the older card is causing the newer one to throttle at all .. it might not do this directly as the temps
might slow the cpu a bit which can limit faster gpus .. it will also give you a feel for what temps the new gpu can achieve on its own

[iero]

Some suggestions:

1)Put the newer card on the top PCIe slot, and the older one down lower - unless this is the current setup that leads to the 100°C situation.
2)When you get the chance you could try to rearrange the fans, more input/less outpout, the change would be as simple as putting them on the front after a 180° degree flip.
3)As for the GTX650Ti, it has a max rated Temp of 105°C, so it being near 100°C, it must be throttling heavily to save itself . While a burst of workload hitting such a temp, would be fine, a machine running at 100°C 24/7 must be really stressing the core and VRMs of the card. A target of 75~85°C would be better for the longevity/efficiency/productivity of the card.
If the card hasn't been cleaned in years, I think it is time . If you are not comfortable-knowable to open the card and change the paste too, just take a can of air to the cooler to break loose any debris.
4)As for the software side, you should try downloading MSI Afterburner. There,on the main window, you should see:
-Core Voltage(mV)
-Power Limit(%)
-Temp. Limit(°C)
-Core Clock(MHz)
-Memory Clock(MHz)
-Fan Speed(%)

Of the above, Core Voltage should be the only one greyed out.
Power Limit shouldn't be greyed out. If it indeed is adjustable, try lowering it 10~15%.
If it not, take a not of your fans and if they are lower than 60~65%, press the gear icon(just below fan speed) and manually drag the bar to 60~65%. More is better for the temperature of the card, but excessive speeds (>70~75%) lead to premature wear to the fan's bearing.

*Excuse my poor English.*