Some general comments before I get into the main reason of this post.

This is going to be a wall of info, and a few things I'll explain to try to cover all bases.

We all know the r8 through r12 case has bad airflow. Its a case designed more for form over function. At stock settings if you dont care much about fine tuning, the 3080 works great out of the box, and its thermals are well within tolerance. The issue is with the extra heat your fans will ramp up to compensate, and generate more noise. Also at stock the card can draw a lot of power, with an undervolt you can save on this, in my example I explain later I'm about 100w under stock power draw.

All in all, its not much work to get a big improvement in thermals and noise, at a negligible performance hit. Also there is no risk to your card by undervolting. It actually is safer than stock, the worst that can happen is you get a setting that is unstable and you need to tweak it a bit more.

I'll be using timespy to show my results, actual results will vary a little card by card and some may get better numbers, so a little worse, but it should all be relative. Timespy is a common benchmarking took a lot of people use today and for years. Its made by 3dmark, can get it on steam. It's free so you dont have to pay anything to use it, only issue with the free version you can't skip the demo each run. You can pay for packages to unlock more benchmark programs, but for the average user this is a waste. A point of note, is each run with the same exact settings may have a slight variation on the score.

https://store.steampowered.com/app/223850/3DMark/

This youtuber gives a great guide on undervolting the 3080. Also check him out for other info, he is mostly focused on sff side of pc building but all of his videos are great. You will need to use msi afterburner to go further with this.

https://www.youtube.com/watch?v=FqpfYTi43TE

https://www.msi.com/Landing/afterburner

In the video he will show how to undervolt and thresholds that should work across all the 3080s, regardless of model. You may be able to fine tune yours a little better or even a little worse. For example he recommended 850mv for 1850 boost, mine was more stable at 850 for 1840 boost as I did have one crash in timespy at 1850. This is the part that will vary card by card.

There is another benchmarking utility that is good for stability, called Unigine Heaven. The main reason I use this is to just run it on loop to stress the gpu. With it you will see roughly the max temps you'll hit with any given settings if you run it for about 5 to 10 minutes. It also will help find if you have any stability issues.

Now I'm going to give some numbers here from time spy, don't focus too much on the number as yours will be slightly different, but it'll be relative to show my performance at each step. I never stress tested each step. but generally the stress temps will be about 10c above my average temp on the results.

This is my baseline with no tweaks. At stock, out of the box: my graphics score was 17509, I had peak temp of 71c on gpu and average temp of 67c .

https://www.3dmark.com/3dm/56584850?

Ive did a bunch of tests overclocking and undervolting after this, I'm not going to list them all because it just clouds the results, and my final result is what is more important.

This is my final result. In this result I set my curve (as explained in the youtube video) to 850mv and 1840 boost. I kept the fan on auto as well. You might find a different threshold that you enjoy better. but I'm happy with this result and don't think i'll go much further.

Graphics 17531, peak temp 60c on Gpu, Average temp 55c. I did end up stress testing this with Heaven and my stress temps hit 65c after about 10 minutes.

https://www.3dmark.com/3dm/56602733?

So in closing. I had essentially the same performance result and my peak/average temps on the gpu dropped between 18-22%

I spent roughly 12 hours dialing this in yesterday, but realistically if you go quick and dirty you can get the same results in 1 to 2 hours of work.

If you have any questions feel free to ask. I'll do my best to answer them.