3080 ftw3 new card - install new driver clean mode, black screen crashes pc, spike in cpu temps

markm75

Gawd
Joined
Sep 12, 2008
Messages
760
I just received the new 3080 ftw3, i am assuming since it just came its one of the newer ones without the cap issues? Unsure how to tell.

At any rate, i had 456.55 already installed from the 1080ti hybrid. Ran some quick tests to verify the new card worked (noisy at 70c but fps increase was great).

I decided to install 457.09 and as a custom and clean install.
Each time i try the pc crashes, black screen bios warning. I also tried default bios clock settings no change.

Im running an 850 watt seasonic focus psu the MSI Unify z490 Meg motherboard and 10900k cpu.

Anyone else ran into this?
 

arestavo

[H]ard|Gawd
Joined
Mar 25, 2013
Messages
1,630
There never was a cap issue, it was drivers. There was a German dude who resoldered an all SPCAP board to all MLCC and the only difference was around 30MHz more OC potential.

So, this card worked with 456.55? Try them again, or try the 456.71 (or hotfix 456.98 based on them).

Edit: running DDU in safe mode to uninstall the drivers cleanly, and then before rebooting unplug your PC from the internet to install the drivers of your choice is also good practice (unplugging prevents Microshaft from installing old drivers before you can install the drivers you want).
 
Last edited:

markm75

Gawd
Joined
Sep 12, 2008
Messages
760
There never was a cap issue, it was drivers. There was a German dude who resoldered an all SPCAP board to all MLCC and the only difference was around 30MHz more OC potential.

So, this card worked with 456.55? Try them again, or try the 456.71 (or hotfix 456.98 based on them).

Edit: running DDU in safe mode to uninstall the drivers cleanly, and then before rebooting unplug your PC from the internet to install the drivers of your choice is also good practice (unplugging prevents Microshaft from installing old drivers before you can install the drivers you want).
Thanks, that was the ticket though i didnt end up using ddu this time:


What i did do was booted to safe mode, uninstalled the nvidia driver from apps and features.
Then rebooted back into safe mode, ran the latest installer and no black out / crash this time.

The odd thing is, since i changed to the 3080 from 1080 hybrid, now my cpu is overheating compared to before, about 10c higher, cant figure out why, perhaps the hybrids fan as an exhaust on the rear did a lot more than the 120 fan i put in place of it (exhaust as well).
 

Denpepe

[H]ard|Gawd
Joined
Oct 26, 2015
Messages
1,717
Well you said your fps went up so I assume your cpu needs work harder to keep it fed hence higher temps?
 

markm75

Gawd
Joined
Sep 12, 2008
Messages
760
Well you said your fps went up so I assume your cpu needs work harder to keep it fed hence higher temps?
Well the fps went up 13 in msfs2020.

But the cpu temp during say cinebench or other avx stress tests now are up 10 to 12c. I guess it could indeed be related to the faster gpu
 

Falkentyne

[H]ard|Gawd
Joined
Jul 19, 2000
Messages
1,679
Well the fps went up 13 in msfs2020.

But the cpu temp during say cinebench or other avx stress tests now are up 10 to 12c. I guess it could indeed be related to the faster gpu

It's because of the video card dumping hot air into your case.
The CPU is running the same as before. It's not pulling any more watts in Cinebench.
What's more obvious to most people are the RAM temps (assuming you have RAM With temp sensors, like Gskill). The 3090 and 3080 FE cards (especially the 3090) is literally positioned right next to the RAM slots and the fan blows hot air right over them (the other fan exhausts part of the air).
 

arestavo

[H]ard|Gawd
Joined
Mar 25, 2013
Messages
1,630
So, your 1080 hybrid exhausts just about all the hot air out of the case. Your 3080 does not, as it's not a blower style card.

You'll need to add case fans, or fine tune your case airflow to better exhaust all of the heat being dumped into your case. Personally, I'm a fan of positive pressure - all fans as intake except the fans on the top of the case.

An easy check is to remove the side of your case. If temps improve, that's the cause.

Edit: someone beat me to it by moments. But I'm glad you got the bigger problem solved.
 

markm75

Gawd
Joined
Sep 12, 2008
Messages
760
It's because of the video card dumping hot air into your case.
The CPU is running the same as before. It's not pulling any more watts in Cinebench.
What's more obvious to most people are the RAM temps (assuming you have RAM With temp sensors, like Gskill). The 3090 and 3080 FE cards (especially the 3090) is literally positioned right next to the RAM slots and the fan blows hot air right over them (the other fan exhausts part of the air).

Well, using hwinfo, ram temps dont exceed 48.5C. I took the side panel off, on my 5ghz test (max load 1.308), i still reach 91.5C with cinebench, it was 92C with the case side on. At this load level before the temps were still at least 10C lower with the old card.

I have the thermaltake v51 case.
The x73 kraken blows up through the radiator out the top. I have two 120mm fans pulling air in on the front and one 120mm fan pushing air out the back (replacing the 1080ti hybrid exhaust)

Putting my hand above the 3080, there really isnt much heat coming off during a cinebench, i even tried maneuvering the 120mm at various angles near the ram but it made little difference.

I guess the sure fire way to tell would be to put the 1080ti back in and retest.
 
Last edited:

markm75

Gawd
Joined
Sep 12, 2008
Messages
760
So i put back the 1080, the issue persists, actually seemed a tad worse. I dont know how changing the video card could have done this, unless something became loosened with the x73 or the 10900k delid under, either way its way higher than it should be.

I have some mx4 on hand, i may take the cooler off and apply the mx4 (remove the arctic silver 5).

edit: now i'm puzzled, so i opened the nzxt software, reran cinebench, now its holding at a max of 81 after a few mins (at the 50x and 1.308 load), nothing changed. I closed the software and retested now the same results, before it was near 94C.
 
Last edited:

arestavo

[H]ard|Gawd
Joined
Mar 25, 2013
Messages
1,630
You very well might have bumped it, or there might be air in the pump. Try turning the case on it's side for a few minutes while it's running, then setting it back up. This will help air pockets move away from the pump, and hopefully fix the CPU heat problem.

If not, a repaste and remount might be in order.
 

TheSlySyl

Gawd
Joined
May 30, 2018
Messages
746
Yeah, sounds like you messed up the mount somehow when installing the videocard. Remount it and things should stabilize back to where they were.
 

markm75

Gawd
Joined
Sep 12, 2008
Messages
760
You very well might have bumped it, or there might be air in the pump. Try turning the case on it's side for a few minutes while it's running, then setting it back up. This will help air pockets move away from the pump, and hopefully fix the CPU heat problem.

If not, a repaste and remount might be in order.

Everything is flowing fine now. Coolant temp around 32-34C. I dont know if maybe the rpms werent kicking in with the pump before (didnt check, now the pump is at 2790 rpm at max for sure though) and now they are (after running the kraken software), but either way its normal again. No remount needed so far.
Also restored my 1.35 bios 1.339 load llc 4 and 5.1ghz with 46x ring and the temps after 2 mins are about 85C with the case on, callin it a win here, still not sure what was going on. Might have been related to the bios reset that occurred when i was first trying the clean install of the driver and getting the black screen.
 
Last edited:

motqalden

[H]ard|DCOTM x3
Joined
Jun 22, 2009
Messages
1,687
If you are curious about the caps, have a look at the back plate area above where the GPU die would be. You should see 4 large black caps and 2 sections of much smaller silver caps in the center area. This would be a " 4 + 2 " array. The original FTW3 cards that went to reviewers had 6 of the large black SP caps.

This is from my FTW3

Caps.png
 

vjhawk

Limp Gawd
Joined
Sep 2, 2016
Messages
323
Yup it was a driver issue causing blackouts.

Yes because the 3080 is a 320w card vs your 250W 1080 Ti, that's 28% more watts, it will cause your case to run hotter and thus affect your CPU temps.

If you had the Founder's Edition model, due to its dual flow fan design, it would have exhausted a lot of that heat out the back and you probably wouldn't be seeing your CPU temps affected as much.

You might just want to tweak your case fan configuration to produce more airflow.
 

markm75

Gawd
Joined
Sep 12, 2008
Messages
760
Well, temps have been fine since the other day, but now i'm plagued with random bsod's.

nvlddmkm.sys PAGE_FAULT_IN_NONPAGED_AREA 0x00000050

I've gone through in safe mode and did the DDU too and installed fresh, thinking maybe it was the driver, but i keep getting this, either randomly during use or if the pc sleeps, when it comes back on it reboots instantly and the last error was this one.

Only other thing i can think of is either the overclock is suddenly unstable (though this error points at the video), or maybe that 850 watt psu isnt enough, though i only have mostly nvme ssds and one 7200rpm.

**I also retried the driver installer and did the clean option (while the same newer one was on there), again it shuts off the pc then boots to bios recovery each time, i did this on my test bare windows 10 image as well (had a previous driver on that one).
Event logs show The computer has rebooted from a bugcheck. The bugcheck was: 0x0000007e
 
Last edited:

arestavo

[H]ard|Gawd
Joined
Mar 25, 2013
Messages
1,630
Setting everything to default in the motherboard BIOS and running the card totally stock is a good next step.

If the crashing stops, it's either an unstable OC somewhere, or power. You can narrow that down by testing OCs individually (CPU, then RAM, and then GPU - each without any other OC).
 

markm75

Gawd
Joined
Sep 12, 2008
Messages
760
Setting everything to default in the motherboard BIOS and running the card totally stock is a good next step.

If the crashing stops, it's either an unstable OC somewhere, or power. You can narrow that down by testing OCs individually (CPU, then RAM, and then GPU - each without any other OC).
I had tried that though I left ram at xmp. Same issue

Tried again with pure mb defaults. Pc still shuts off during install. One note of corse is thst with ddu and booting normally with Lan unplugged I can run the driver install without this power down issue but I still run into those random bsod
 
Last edited:

arestavo

[H]ard|Gawd
Joined
Mar 25, 2013
Messages
1,630
I had tried that though I left ram at xmp. Same issue

Tried again with pure mb defaults. Pc still shuts off during install. One note of corse is thst with ddu and booting normally with Lan unplugged I can run the driver install without this power down issue but I still run into those random bsod
Sounds like a GPU or PSU problem.

Easiest way to test is to try the video card in another system, be it a friend's or a repair shop's.

Edit: just an FYI, unplugging the internet makes it so Microshaft doesn't automatically install old Nvidia drivers after an uninstall.
 

markm75

Gawd
Joined
Sep 12, 2008
Messages
760
Someone in a different thread mentioned they were having the same bsod and shutdowns on sleep

They rolled the driver back to the last driver and havent had issues.

So far i used DDU in safe mode, unplugged internet, booted normal.. installed the older driver 456.71 (no crash on install), put the pc to sleep, waited 1 hour, woke pc, it was resumed normally. So either a fluke (i had a few times it worked before), or the driver has indeed cured the issue at least with sleep.
 
Last edited:
Joined
Apr 22, 2011
Messages
600
Oh nvidia and your anti-amd driver propaganda you propagate in tech forums, how the turntables... :woot::ROFLMAO:
 
Last edited:

arestavo

[H]ard|Gawd
Joined
Mar 25, 2013
Messages
1,630
Oh nvidia and your anti-amd driver propaganda you propagate in tech forums, how the turntables... :woot::ROFLMAO:
I believe that I've seen a reference to AMD taking 4 months to even acknowledge the 5700xt black screen issue and work on a fix. Or was it 2 months to acknowledge, 4 to (kinda, but not totally) fix? I dunno - last AMD cards were 7990s that didn't work with the high end Dell DVI monitors that I had, so I had to return them.

Nvidia, in my experience, doesn't usually take quite that long to put out fixes for important bugs like that.

So there's that.

But here's hoping that with those firings and restructuring of their driver team, it will be much better.
 

markm75

Gawd
Joined
Sep 12, 2008
Messages
760
Little side note.. my pc with the 3080ftw ultra now pulls about 705 watts during a game (with the 10900 at 5.1 at 1.339 load) total (also includes a sony 43" lcd and a few peripherials). It only pulls about 365 when not in a game.
I guess expected, i think the wattage was rated around 395 or so for the ftw3.

What i might have miscalculated was using the 1080ti hybrid ftw, i swear i was only pulling 380 watts (total) before with the 10900k, but maybe i misread my ups reading at the time. I thought the 1080ti hybrid was around 280 watts (should have been about 600 total i think).
If i'm at 365 without a game, that should have been 365 + 280 = 645
 
Last edited:
Top