Dabadger84Is anyone else encountering GPUz saying THRM as a PerfCap reason at all on their 3090 FTW3 Hybrid? I'm not certain but I think this has to do with the occasional faulty iCX thermal readouts as reported in the 3090 FTW3 Ultra review by GamersNexus - Occasionally in HWInfo64 I've noticed one of my temps (it's never the same one, always either one of the VRM/PWR readouts or vRAM/Mem) will show a reading in excess of 6500C for a split second. Just curious if others are experiencing this or if I should be concerned my card is going to randomly merc itself. Also, https://www.3dmark.com/3dm/55029822 I'm working on my score (14.5k) :-D
Mathieas Let me install the same program to see. Do you have a sense for how often it happens? Does it happen at idle? Any additional info on your SOP as I would like to mimic your test best I can.
RiffyDivineWhat tools are you using to overclock it? Did you install the new bios from precision? I was trying to find a change log for it with no luck.
Dabadger84RiffyDivineWhat tools are you using to overclock it? Did you install the new bios from precision? I was trying to find a change log for it with no luck. There is no update from Precision X1 for me for the 3090 FTW3 Hybrid - I would assume whatever "firmware update" there was from Precision X1 for 3090 FTW3 Ultra owners or whichever, was shipped already installed on the Hybrid cards. This specific overclock was doing using Precision X1 - closing the app after the settings were inputted because it's a resource hog for benchmarking purposes.When I'm undervolting/overclocking I use Afterburner to set just the Voltage/Clock Curve, Precision still controls the fans etc (if it's open)
RiffyDivine Mine is also the hybrid FTW 3090 and found it weird it wanted a firmware update. Been trying to find out what exactly the firmware does or someone to just tell me the crap I heard about it undervolting the cards is just crap.
Dabadger84Found my max, at least on the core, went one step higher & Port Royal crashed halfway through. All background processes I could close were closed during this run: +210core(+100mV)/+1100vRAM - I have no way of knowing if the power draw was the same as a previous run, but oddly, it did show in that previous run with GPUz open that I was only hitting about 457W? Crazyness. I would assume my max load temp was in the 55C range.
Dabadger84Mathieas Let me install the same program to see. Do you have a sense for how often it happens? Does it happen at idle? Any additional info on your SOP as I would like to mimic your test best I can. It is 100% random when it shows up in HWInfo64, the temp that is, happened at idle, happened in game, in benchmarks, and it's just one readout tick then the temp goes back to what it actually is. With GPUz reading out THRM, it's specifically happening at the end of Port Royal when I'm pushing max voltage/clocks really hard, and I also saw it happen during Time Spy Extreme too, during the test or at the very end of GT1 or 2. It doesn't always show up, and it's just a BLIP of Red/Thermal mixed in with the wall of Green (PWR) - it's never more than single sliver of it. Thanks for checking, really hoping this isn't a cause for concern... I don't think it is, but after seeing someone's 3090 on these forums died after two weeks, my concern level went from green to slightly yellow.
MathieasI actually cannot get mine to the 14643 again. At +180core I did hit 14604 or so. My stock PR run is still 13743 which is about as high as i have seen stock. Are you on the XOC bios? I am using the stock bios on OC setting. Is the +100 voltage different for each card? or is that a static voltage change?
MathieasDabadger84 It is 100% random when it shows up in HWInfo64, the temp that is, happened at idle, happened in game, in benchmarks, and it's just one readout tick then the temp goes back to what it actually is. With GPUz reading out THRM, it's specifically happening at the end of Port Royal when I'm pushing max voltage/clocks really hard, and I also saw it happen during Time Spy Extreme too, during the test or at the very end of GT1 or 2. It doesn't always show up, and it's just a BLIP of Red/Thermal mixed in with the wall of Green (PWR) - it's never more than single sliver of it. Thanks for checking, really hoping this isn't a cause for concern... I don't think it is, but after seeing someone's 3090 on these forums died after two weeks, my concern level went from green to slightly yellow.Do you have a sense for how long it took to get the spike?
Dabadger84 It is 100% random when it shows up in HWInfo64, the temp that is, happened at idle, happened in game, in benchmarks, and it's just one readout tick then the temp goes back to what it actually is. With GPUz reading out THRM, it's specifically happening at the end of Port Royal when I'm pushing max voltage/clocks really hard, and I also saw it happen during Time Spy Extreme too, during the test or at the very end of GT1 or 2. It doesn't always show up, and it's just a BLIP of Red/Thermal mixed in with the wall of Green (PWR) - it's never more than single sliver of it. Thanks for checking, really hoping this isn't a cause for concern... I don't think it is, but after seeing someone's 3090 on these forums died after two weeks, my concern level went from green to slightly yellow.
Dabadger84RiffyDivine Mine is also the hybrid FTW 3090 and found it weird it wanted a firmware update. Been trying to find out what exactly the firmware does or someone to just tell me the crap I heard about it undervolting the cards is just crap. The only Firmware update I'm aware of that was for 3090s/3080s was to fix fan control issues as well as RGB control issues - There was an issue with certain cards clocking too high when the cards first launched causing crashes, but that was fixed with a driver update from NVidia several revisions ago. eVGA cards didn't seem to have the problem at all, so it's doubtful they would release something like a firmware update for it.
eg1122Made a spreadsheet template, see what you guys think. https://docs.google.com/s...PCqM/edit?usp=drivesdk
eg1122So there has been a lot of talk about undervolting. I tried it for the first time and I must say I was blown away at difference in temps (especially this) and performance of my gpu. I am perfectly happy with stock clock speeds and I really have no need for overclocking. I put together my data in a spreadsheet. If anyone wants to add their data feel free to do so.https://docs.google.com/s...nd0I/edit?usp=drivesdkCurrently I have my undervolt set to 925 mV with clocks maxing out at 1995 MHz and I have about a 7C temp reduction and a fps boost on all my games and Port Royal. I was also able to adjust the fan curve to have less fan noise.I just played Assassin's Creed Valhalla at max settings @1440p and max temp after 2 hours was 53C.
https://www.3dmark.com/pr/916858