Geforce driver 390.65 WHQL Download & Discussion

Yes ,yes and yes.

 

Another major bug here is that these drivers wont install on 1080 sli, dual xeon board, black screen when installing. No such issue with any other previous driver.

 

Yeh even with power monitoring off, I'm still getting some pretty bad stuttering. Gets worse after a long session gaming. By a couple of hours in it's constantly awful. I've reverted to 388.71 and no monitoring at all, and everything was fine. I'll be stepping forward from there to try and pin it down, but I feel like the 390 drivers are basically no good right now. Perhaps they rushed them out the door for the spectre fixes, I don't know...but it aint good.

 

Shocking news are shocking. Does it go away after a system reboot?

I've done a clean driver reinstall with DDU a couple of days ago and so far hasn't run into the same hitching I was seeing.

Edit: aaaand it's back. Definitely a driver issue which seems to be prompted by the resolution / display mode changes.

 

Are we talking about stuttering in ALL games or specific game? and is it only happen id one use AB with power monitor check?

 

In all games. The power monitoring thing is definitely a thing (now confirmed by nvidia and escalated), but there's something else, as well. It's not as severe (not even close) and not as consistent. At this point I'm very early in the stages of pinning it down. I can say, just as the power usage graph had a wave in it in my images above, the VID usage had a similar wave, and that's not normal and I'm not seeing it with 388.71. In fact, with 388.71, everything is AOK. I'm going to have to upgrade to 390, and see if any other monitoring has an effect here, by disabling all monitoring, playing a ton (it can take hours for the stutter to kick in), and one-by-one enabling each monitoring... Honestly though, I'm not sure that we are in control of whatever is triggering this. What I am sure of, is that nvidia are looking into this whole stuttering thing for us.

 

so even if AB monitor isnt enabled it will happen?

Not want i wanted to hear cause i planed to installed this when I get my 1070ti.

 

I didn’t notice *any* stuttering when I tested a Hollow Knight earlier using this driver.

Yes, it’s a 2D platformer but has side scrolling which makes stuttering easy to spot if it’s there. This was my go-to game which I used to test for the stutter fix on the earlier v38x.xx driver. The only way to get rid of it before NVIDIA fixed it in a later driver release was to uncheck Power monitoring in Afterburner. However, I have Power monitoring enabled and the RTSS overlay is on in-game but I haven’t seen any stuttering at all in this game or in Street Fighter V Arcade Edition (this game also features left and right scrolling) which I played extensively this afternoon.

I’m not saying it isn’t there in other games or on specific hardware but, on my system anyway, this issue appears to be far less noticeable than the previous Power monitoring bug.

P.S. Which games are people seeing this stutttering in please so I can check those for myself?

 

Have you changed the power setting in NVCP away from Optimal or not? @Dragondale13 above mentioned that the power profile might have something to do with the stuttering. I'm off to do some testing with the MSIAB power monitoring and NVCP power management. I did notice after a 4 hour session of gaming with my buddy last night the stuttering was worse at the end versus the beginning.

 

No, it is set to Optimal. I will test with it set to Adaptive and Prefer max. Performance.

 

Does sound like the standby memory issue with Windows 10 FCU to me.
Playing perfectly fine at the beginning of a game sessions and after some time it stutters.
Whereas the standby memory should normally be released, it doesn't with the FCU build and the stutters begin.
I can replicate it everytime. Empty the Standby memory with a task and your gaming sessions will be stutter free or update to a newer build. (unless something else is happening).

 

oh definitely not some thing i want to hear i use prefered maximum power by defualt for years last time that was bugged in anyway it took nvidia 6 months to fix it , that bug was it did not drop gpu to idle speeds when gpu was doing nothing and just sitting on the desktop.

then again I am using 1703 build i not bothering with 1709. Are all the people reporting this issue seeing it win 10 1709 build or 1703 build too?

 

While in IRQ-mode I've had a driver crash in Battlefield 1 after a couple of minutes and noticed in the log a "nvlddmkm" related warning that the driver did not respond and needed to be recovered.

But in MSI-mode everything is fine.

 

I have massive stuttering in BF1 using these drivers after alt+tab(1709 build / fullscreen optimization disabled). Vulkan 388.84 is way smoother.

 

^
Imo its best 387 & 388 so far, there was one boost with r387 then it got lost somewhere down the road, this 388.84 has it back or still has it. And its solid as overall, at least by single display scenario, now using dp @ 144HZ 99% of the time.

 

...but we need 390 for spectre fixes.......

 

Do we?

I don't or care about that one bit. :d

 

its not a fix if it reduces performance -in any way shape or form

thats like trying to prevent breaking a finger, by cutting off both your arms

 

C

Can you fill out the driver feedback form below so that we can look into this? Thank you.

https://surveys.nvidia.com/index.jsp?pi=6e7ea6bb4a02641fa8f07694a40f8ac6

 

Too bad if performance does drop. If they purposely leave a known security hole vulnerable to keep up performance, the company can later be held liable for any breaches from same-said vulnerability. Besides, they want you to buy newer cards, so if they have an excuse to make things slower, it'll make their non-vulnerable newer cards even more attractive.

When dealing with corporations, you need to think like one.