NVIDIA GeForce 528.24 WHQL driver download & Discussion

B450? your profile says 550.

 

Mostly likely a typo from when I first signed up.
Im running MSI B450-A PRO.

Ill keep testing, but If I get even 1 error or sniff any sign of instability with 517.48 ill rma the card.
528.24 is not the only driver i get these events, i got them on 526 branch and 522 branch. Again, never crashes, just events on the viewer..
on the 516 branch I also got them on some versions...

but all of this started with 511.79 + 512.15 (just google it)...It got triggered originally by setting digital vibrance on nvcp over 50%, I know that because it was reported at the time by alot of people and acknowledged by NVidia (and fixed) on future drivers

I wonder if digital vibrance is STILL some kind of problem since I use digital vibrance on every driver (1st thing I do after install actually).

I'm puzzled, because 517.48 is ROCK SOLID. I gamed for hundreds of hours on warzone 2 with it, a game that famously unstable, never crashed once. Never got one single nlddmkm event. Never a hint a instability. but only on 517.48... Drivers before 511.79 also seem to work well.

 

BTW do you have any tip for RAM testing? (except Memtest of course).

 

Warzone 2 will make your day miserable with unstable OC's or even unstable xmp.
Other than that, I like hci mem test for overnight tests (very slow) but I still think this is the best to actually find errors. I can be stable with memtest5, primem, karthus and it WILL fail with hci memtest.
So i learned to test with testmem5 first for quick tests and test with hci overnight (I aim for 1000%).

And for my "issue":

I just ran asus realbench for a couple of hours without a single event (517.48 driver).

I then went ahead and tested prime 95 small ftts, with avx and hyperthreading with my stock cpu settings and xmp while I ran 3d mark stress test 20 loops. Bit of overkill I know but at this point Im curious.

Im actually trying to get the dammed driver to error out but it wont.

Not a single problem.

Well, I guess ill stay with 517.48 forever? Until nvidia fixes their crap...

temps:

84c max mem temp
74 max gpu temp.
90c hotspot...

temps dont seem bad considering it pushes 300 watts.

 

TM5 (TestMem5), then find and download 1usmus_v3.cfg, then edit that using Notepad and change Cycles under [Main Section] from 3 cycles to 60 cycles.

Run it for around 10 hours for it to finish.

I would highly recommend your CPU not being overclocked first when trying to find a stable memory OC. Because TM5 1usmus can pickup CPU instability and you might assume it's RAM timings instead.

 

lawood: 517.48 is very good driver, I am still using it with my 3080. Tried some of these latest drivers and have unstable frametimes and weird spikes with all Vulkan/OpenGL API games. So I always returned to 517.48 and everything was back to normal with flat frametime charts. What is the point of these new branches I don't know - didn't even saw any performance improvements in anything I played.

 

True, it literally triggers triple fault exceptions on my 1680v2 if i don't have atleast 1.306v at full load.

 

Yeah, even the performance seems better.
Not FPS related , but frametimes stability overall. I feel this more on apex legends specifically.

It just works.

 

It blew me away when always between 45 and 50 cycles I was getting 1 WHEA error, like bloody hell come earlier took me over a week to iron out the final steps of getting my PBO2 and final secondary and tertiary timings tightened to the max.

Then ran 60 cycles just to make sure all is well. This was months ago. I should actually run another 60 cycles now that its summer just for good measure.

 

Dead Space 2023 runs very good with this driver after playing an hour at the highest in game settings.

 

plus, prime95 blend test for at least 30min, specially large ffts to test imc and ram together communicating.
I would also recommend GPU stress test on OCCT to heat up the case during TM5 test parallelly.. that will uncover heat related ram issues faster (tRFC oc).

all these should be enough. it did for me when OCing to max. and yes RAM, IMC instability is very sneaky and bothersome (driver crashes out of nowhere, even corruptions etc. and you wont think its the ram/cache).

 

I've been running 517.48 for many months and just a couple days ago decided to try this 528.24 out on a 1050Ti PC I have before I try it on my gaming rig with 3070.
Point to note is that the 1050Ti machine runs almost 24x7 except when it's put to sleep and restarted for windows updates. It's been running just fine with 0 issues/BSODs ever since.

Initially the driver ran fine but today when the system was resumed from sleep of 8+ hrs, I immediately tried to watch a YT video which ran for about 10 seconds before triggering a BSOD with DPC_WATCHDOG_VIOLATION code. The screen was also turned into green boxed and lines.
I'm not running any overclocks/underclocks on either CPU (Xeon E3-1280 v2) or GPU (1050Ti) or RAM. It was a one off hopefully. If it happens again, will revert to 517.48 and report here.

 

@Astyanax Nice ninja delete but I appreciate your response. To clarify to others, it's not only caused by hardware failures.
https://learn.microsoft.com/en-us/w...bugger/bug-check-0x133-dpc-watchdog-violation

To quote directly from Microsoft's documentation...

 

Wccftech reports : Enabling DirectStorage in Forspoken will deliver Faster Loading times At the Cost of lower FPS.

I have 4 Samsung 980Pro M2 2TB pieces with gen 4 enabled in the Mother board bios, supported with Samsung magician 7.2.1 and my games loading lightning fast, can it be faster with DirectStorage.?

Can anyone confirm this who played Forspoken.? thanks in advance


Read more about it here : Wccftech (@wccftech) • Instagram photos and videos

 

My brain saw the Watchdog, but not the DPC, jumped to Clock Watchdog Timeout.


Deleted because i currently have a a DPC Watchdog Timeout bug open for UsbDK

 

Less than a 1s on my setup (also fast NVME). It can be as low as 0.4s
Never saw this fast loading times in my life.

In menu - click Continue then bum You're there.... less than 1s - let it sink.
Note:
Download Demo on Steam & see...

 

But according to the the article up to 10% loss of performance on a 4090, I think I would rather have 1 sec more load time an 10% better performance.

 

No, the performance in Forspoken with DirectStorage enabled isn't worse, it's actually better. However, performance is roughly the same no matter which kind of SSD you use in with DirectStorage, so a faster SSD only cuts the loading times down more.

The person at PCGH who's responsible for the misconception that you performance is worse with faster SSDs using DirectStorage simply made the mistake to calculate/record the average framerate including the mostly black loading screens that naturally have a very high framerate. The longer these loading screens were displayed (the slower the drive, the longer the screen is displayed, if Direct Storage isn't enabled, then the loading screen is there even longer, etc.), the higher the average framerate got due to it, since it's an average.

 

I recently done a fresh install of win 11 22H2 and I am seeing weird issues with scaling and older games. For instance dosbox, if I use any display option apart from overlay, exit the game and the start menu becomes a small window in the top left so I cant navigate it, to fix I need to reboot or sign out. Then if I play a game using DLDSR then exit my icons are messed up, a desktop refresh fixes this. I then see issues with apps like GOG galaxy and Steam having messed up scaling where the UI would be small or zoomed in. I never had this problem pre install so I do not know if this is a driver or Microsoft issue, its annoying but I can workaround it.

My display is set 4K 120hz set at 200% scaling. I also noticed the recommended scaling has changed from 200% to 300% so something has changed.

 

Ok, thanks for posting, that makes sense.