Discussion in 'Videocards - NVIDIA GeForce Drivers Section' started by Alberto_It, Jul 19, 2021.
I just didn't OC to maintain the temps until my WB arrives
Seems like those drivers fixed/reduced driver crashes for me, and also 3DMark scores are back to normal
So any fix for the warzone stutter, after about 3 rounds my game stutters. Rebooting the game fixes it.
Something changed with the way fps limiter are applied in the 471 branch in some games (Generation Zero for exmaple). With the 466.79 drivers, it is working fine (if i limit at 90, fps stay at 90). With all the 471 drivers, fps can drop down to 72 or 75 sometimes in some areas with a lot of stuttering. If you remove the fps cap (you can do this 'on the fly' with RivaTuner) fps are going up to 130 to 150. I tried with and without gsync, vsync, fastsync, MPO on or off, with the Nvidia or the RivaTuner fps limiter with the same results...
But you as the video shows can confirm a decent uptick in performance in game?
It's still crashing in EDO for me, this time after around 30 minutes while streaming to YT, so it's not the MPO's.
Probably need to roll back to previous driver since I don't and shouldn't need to adjust voltages on my graphics card, don't even have any overclocking tool for such installed.
Temperature pretty descent to, cooled with an EK Vector Strix block sitting on a 360.
You can at least monitor the card to see if anything looks suspicious.
I had a few situations in the past where some games crashed when card reached max tdp.
Had GFE's performance monitor with full details running but for some reason Nvidia chose that Shadow Play doesn't record it.... good job Nvidia
Cause I thought that I could see exactly what happens with frequencies etc. just before it crashes when checking the video afterwards.
Not sure if you recorded the game or desktop because during game capture overlay's don't get recorded. Use desktop capture or just use OBS.
Tried with desktop recording, doesn't work, also setting GFE to desktop tanks fps slightly.
Tried setting the powerlimit to 95% and core clock to -15MHz, just had an 1 hour + session with no crash.
Noticed that the GPU boosts beyond it's default turbo boost frequency which is 1830Mhz, think I saw it up at 1860Mhz, so might have been higher when it crashed previously.
If that really is the case it is a complete fail that the BIOS or driver doesn't restrict it better.
Here im to report back, so the driver is ok, its the same in performances as 466.77, strong stable, card is behaving normaly no heat, no turbine noise, no bsods, no desktop return, all fine for now, i tested heavy games for long time period and maxed graphics, pes 2021, bops 3, forza horizon 4, RE VILLAGE, finally a good driver
Thought I was getting crazy, but yeah I saw the same thing with lots of particles in Desitny2 and this game is known to have lots.
I went from 466.77 directly to 471.22 and then saw it, its not a huge fps drop, but it is noticeable (from low 70s into low 60ish fps), out of my head I would say 5-7fps or so.. I keep monitoring fps there so i kinda know how it acts in certain places.
Ingame settings at default high, expect shadows distance v.high.
Spoiler: gfx settings
<cvar name="force_auto_detect" value="0" />
<cvar name="force_enable_multi_threaded_render_submit" value="0" />
<cvar name="window_position_x" value="0" />
<cvar name="window_position_y" value="31" />
<cvar name="window_mode" value="3" />
<cvar name="anti_aliasing_mode" value="2" />
<cvar name="dof_mode" value="2" />
<cvar name="ssao_mode" value="1" />
<cvar name="shadow_quality" value="3" />
<cvar name="environment_detail" value="2" />
<cvar name="character_detail" value="2" />
<cvar name="texture_quality" value="4" />
<cvar name="texture_anisotropy_level" value="2" />
<cvar name="foliage_detail" value="2" />
<cvar name="local_light_shadows" value="1" />
<cvar name="wind_impulse" value="1" />
<cvar name="gamma_control" value="3" />
<cvar name="foliage_shadows_mode" value="2" />
<cvar name="hdr_output" value="1" />
<cvar name="atmosphere_lighting_detail" value="1" />
<cvar name="master" value="0" />
<cvar name="framerate_cap_enabled" value="0" />
<cvar name="low_latency_mode" value="1" />
<cvar name="fullscreen_resolution_width" value="1920" />
<cvar name="fullscreen_resolution_height" value="1080" />
<cvar name="fullscreen_refresh_rate_numerator" value="143981" />
<cvar name="fullscreen_refresh_rate_denominator" value="1000" />
<cvar name="windowed_resolution_width" value="1920" />
<cvar name="windowed_resolution_height" value="1080" />
<cvar name="render_resolution_percentage" value="150" />
<cvar name="framerate_cap" value="144" />
<cvar name="version" value="3" />
Will try 471.11 now, otherwise I will go back to 466.xx
So far so good but my rig is old at this point but working fine on my gtx 1060 6gb. No problems so far played resident evil village for about 4 hours no issues to report. Windows 10 64bit.
You guys mentioned particle effects.
Same for me with parts of Gta5 with lots of mods I installed.
Was having an issue with the last win update from 20H2 to 21H1.
Reformatting fixed that, updated, only applied popular tweaks like, game mode off, CFG off, disabled mitigations using Inspectre, FSE reg tweaks, nagling, system responsiveness 0.
HAGS off, low latency mode nvcpl off.
No more slow down with heavy particle effects.
Might help some peeps out here or out there, but at least somewhere
Oh! I do the spin test as i call it...doing donuts in areas with lots of grass (have it set to ultra) or on vespucci beach in front the condos...when the cinematic cam moves to the wheel wells it was the most noticeable, but now completely smooth.
Fps limit 60, adaptive vsync, using nvcpl., ultra shadows enabled.
I'm not sure why it behaves likes this.
Even if you set a fixed boost freq range still goes 15MHz higher (but only 15) between 45-55C, at least on my card.
I didn't check to see on default settings, despite there is plenty of voltage there, you never know.
Thing is, I saw enough issues from people running default clocks or dynamic boost/oc , but almost none from those who undervolted based on stable tested voltage and fixed clock.
All mine are undervolted and never had an issue like bsod, only crash/recover but then it was because of going 15MHz higher than I set.
Eventually I ended up adding a pinch more voltage and everything was stable since then.
None of the cards reach max TDP also, so that could be another benefit.
Im still on 20h2 though.
I fully agree with your statement, we are all nVidia beta testers...........
Here the same issue with CoD Warzone ,i don't think thats a driver problem, the game is to big in size and loading times are slow even with a Ryzen 9 @4.28 Ghz and 32 GB of Ram @3333 Mhz
Harddisk Samsung nvme 2 980 Pro
Not quite sure why, but the GPU boosts up to 1980Mhz, GPU frequency shown around halfway into the video.
Also got some odd spikes I havn't seen before, like when going into supercruise.
Game needs more optimization, so it's not only the driver.
Getting a bit tired of boost clocks going higher than what stated on the spec sheet, if I wanted higher I would overclock it myself.
Going to test with some other games and see if I'm going to keep this driver.