Let me guess, the card in question is EVGA or Gigabyte and the second one is another brand? Most likely thermal throttling is the case, 3080\3090 are known for that, these 2 brands being the worst (in that case only LOL). Note: it's not the core temp that you see in afterburner, it's memory temp, you'd need utility called HWinFO64 to see it. Download it, take a look at your memory jusction temp sensor reading and you'll most likely see 110-112 degrees Celsius. Common fix is to change thermal pads to good quality ones with 12-17 W/mk (I use these,
https://cdn.alzashop.com/ImgW.ashx?fd=f4&cd=CB131w1&i=1.jpg) and do everything possible to improve the ventilation.
Both cards are EVGA FTW3 Ultra 3090's. Not using Afterburner but EVGA's Precision X1 instead. 60c is way too low for thermal throttling when I have the temp target set at 91c lol.
Did you carefully read what I wrote? Did you download the utility and check your memory junction temperatures? You are still talking about your core temperature which has nothing to do with the issue.
My bad, forgot about that. I just used GPU-Z instead as I don't have hwinfo installed on this PC atm and yeah the memory temps are getting seriously high, 110c peak. I've been debating watercooling these cards anyway so perhaps this is a good excuse to get the air coolers off and replace those pads in the process.
EDIT. So I just checked the 2nd card and realised it's not that far behind the first one, sitting at 108c on it's memory yet it's still performing way better.
Commie is right, 110C is where throttling starts. If you keep your monitoring window open (by the way, does GPU-Z show memory junction temps? Didn't know that) and watch it next to miner window you'll see fluctuations in hash rates when temp. goes from 108 to 110c and back. And it may go as high as 112, too. As was already said changing thermal pads is the only solution. If you do it yourself keep in mind that EVGA doesn't put any (sic) between PCB and backplate, don't forget to add them.
PS merit is appreciated
