Why is 537.58 still the best performing driver for me?

Same, everything is running better than 54x.xx and 55x.xx for me as well. Thank goodness I'm only playing older games right now and have always used Firefox. HAGS and ReBAR have been enabled since I had my 3090, so a couple years or so now.

I'm 8 or 9 chapters into RE4 Remake and that is a large part of why I'm just staying on 537.58 for the foreseeable.

This is good to know. Hopefully by the time I'm required to update drivers we will have a 555.55 or above that is a keeper. Do you happen to be playing FH5, RE4, Fortnite, Stardew, Gears Tactics, Starfield or Mechwarrior 5? 537.58 is the smoothest driver I've used for these games that I'm actively playing.

 

I think I saw the HAGS microstutter in Talos 2 under those few problematic driver versions. Although I also updated drivers regurarly, so my limited experience might be random (it's possible that simply the reboot after disabling HAGS was the fix).

 

Well, I guess technically, it may block the CPU, which then blocks the GPU.
The power sensor on GeForces usually has higher latency than others, so it's a prime suspect in case there is stutter. I don't know how high the values for power sensor (or GPU category as a whole in hwinfo) need to be to be a clear cause of stutter, I guess we only can fiddle with knobs and see what happens...
But generally, I'd rather not poll the power sensor during regular gaming, just to be sure.

 

can someone send clean version of it please, i could not find toaster's mega archive cus it was empty

 

None of those but am playing Division and RE Village atm, and played RE4 remake earlier. Far as I can tell things are more or less the same and performance is the same but I haven't really monitored actively.
It is entirely possible 537.58 would be better still.

I did do a brief run of Control, Days Gone and Metro Exodus Enhanced Edition to verify performance was on par with 537.58 using 551.46 and it was, far as I can tell.

Also I turned HAGS off for a bit with 551.46 and saw no difference in smoothness in games or desktop so based on that things are good for me, atm.

 

RTX 4080, 5800x3d, 32 gb RAM, driver version 537.58, monitor lg27gp850, gsync, vsync activated with -3 fps from my max refresh rate, HAGS disabled, rebar enabled, hpet default in windows, enabled in bios, tested with and without hpet, 0 differences in games. 5800x3d undervolted, PBO -30, AMD CBS > NBIO > SMU > CPPC Enabled, AMD CBS > NBIO > SMU > CPPC Preferred Cores Disabled, AMD CBS > CPU > Global C-State Control Enabled, PPT - 122, TDC - 82, EDC - 124. I tested it with undervolt settings and default settings, no difference towards reducing stuttering.
I have stuttering in games like god of war 2018, witcher 3 dx11 huge stuttering (almost 0 in dx12), some stuttering in death stranding (not terrible), battlefield 5 enormous stuttering in dx12, almost 0 in dx11, stuttering in edlen ring (but i guess it's just the game), manage to reduce stuttering almost to 0 just disabling anticheat. Latest drivers gave me worse performance compare to 537.58. I did reinstall drivers multiple times with DDU and through nvcleanstall, 537.58 always better. I have no idea who to blame here, nvidia, microsoft, my config, or anything else, honestly i just tired of this. Not a lot of people talking about this issues, either they don't have it or just don't care at all.

 

For me, the 551.xx is smooth and just as good as previous ones.

I hope all reporters of stuttering are well aware of the initial phase caused by shader compilation, occurring alongside any new driver install/update. It is expected for any (most) modern game(s) to stutter when you first load it and load a certain area/map right after switching drivers. Especially if in between the driver swaps you do more thorough clean-ups via DDU; but I think most of the time each new driver causes the shader cache to be renewed and will exhibit this phase.
Then, to correctly assess problematic stuttering, and performance improvement/worsening, one would need to rule out this initial source, and I wonder how many of those who report it take this into consideration and have the patience to run their regular test runs and scenarios a few times at least, to have all involved shaders compiled, and also don't have any other problems with their shader cache.

In short, this is very wrong assessment method: 1. Install/update new driver 2. Start your game 3. Stuttering encountered in first 10 minutes. You might even load multiple levels 4. Conclude/judge driver is bad, stuttering mess

 

Reading this pique my interest as I had been playing MechWarrior 5 every now and then and had found it to be a little bit stuttery, not bad stutter... but it is there.

Installed 537.58 and yep, absolutely the game is noticeably smoother

 

I don't think anyone has smooth gameplay in all games, if so, please let us know your secret But I do agree that these drivers worked better with games in general, however some games simply stutter by nature for different reasons.

 

Dead Island Riptide~

 

I've noticed pegging chrome to prefer maximum power in NVCP and dumbing VSR (if you have it on) to 2 gets me no hitches in YT content.

 

You're making me wanna downgrade (again) too lol

 

smells like cognitive bias in here.

 

MechWarrior 5 is actually smooth now compared to before, call it what you will... I'm just happy with how the game is playing now.

 

Thank you for your pointless contribution to the discussion

 

The latest drivers are pretty good for me. No issues.

 

Thanks for the tip, unfortunately it didn't help I'll try the latest driver now.

Edit : 552.12 fixed it, yay
Or it didn't. Damn, there's something running in the background messing with Chrome. What the heck could it be.

 

Yeah, not real sure if it is because the game is older and no longer updated, but this driver is definitely smoother in Mech5 with a set pack of mods that my friends and I play with. Pretty simple to compare in my case.

Uninstall 537.58
Clean install 552.12
Play ~3 hours
Uninstall 552.12
Clean install 537.58
Play ~3 hours

The second half of that was much more smooth. Higher %lows, fewer shader stutters. Nothing changed between the 2 tests. Same mods, same graphical settings etc.

Don't get me wrong here, I'm not a GPU driver conspiracy theorist spouting "THESE ARE THE LAST DRIVERS BEFORE THEY PUT THE BUGS IN THE CODE!!!" or crazy things like that. I've just done this same thing with my last 4 GPU's and it works well. Establish a set of games I currently intend to play, find a driver that runs them well, stay on that driver until something comes along that absolutely requires me to upgrade. Probably Gears 6 at this point.

 

I'm sure they don't put bugs in the code, but at some point they stop optimising for older games, and by extension, older architectures. Granted, this was nearly 20 years ago now, but there was a time when performance on older cards in some games would decline in newer driver revisions, sometimes significantly so.

PhilsComputerLab over on Youtube did a retrospective on this, benching some old cards and games across multiple driver revisions, and his findings correlate with my recollection of those days. I'll try to find the video link.

Edit: here it is

 

Haven't tested the last 2 drivers that have released. But agree, 537.58 is the last driver without frame pacing issues for my system as well. Immortals Aveum really suffers from it. That's always my first game to test, fluid and perfect with the 537 driver, anything above, it's a nightmare.

Might try this HAG trick and see if that helps, I only use DLSS 2, frame generation for me is always a miss, either bad IQ or other bugs. Just not a fan. Wish it was more stable, I'd gladly take the extra frames.