*** Windows 10 + NVIDIA WHQL Drivers are Killing Alienware and Clevo LCD Panels ***

All my drivers are grab from http://www.laptopvideo2go.com/drivers

 

Thanks

 

Enjoy bro.

 

Fails on installing display driver every time, even used the display uninstaller from guru3d and same problem regardless. Installed in disable signature enforcement it's just a no go for some reason :/

In api.dev.log this is what it fails at:

inf: Failed to import driver package into driver store

 

Did you downloaded the .inf moded?

 

I did mate, ill try again just in case i did something wrong

Sent from my SM-G900F using Tapatalk

 

@t456 suggested a driver test. Bring back win 10, perform his test with the HIDI2C.sys, and pour through the registry disabling everything similar that may drive the edid flash and break the checksum (Prema suggested it seemed similar to the Nvidia vbios checkheader). As this also effects the 9xxm series, I'd suggest it either flipped a bit there (in the vbios) or it is something using that to break the checksum. Admittedly, I do not know as much as you guys do in these areas. For this reason, just wanted to point to tests and findings to date.

 

Everyone seems to be glossing over the fact that 3 of us experienced 8 beeps after restarting from the 4-5 Windows 10 ready updates. I really think that's worth a look.

 

There is an eeprom on the SEC5448, but it looks like this was incorporated into the controller IC. Of these there are no spec. sheets, making it difficult to determine the appropriate wp method.

Thanks for guinea-pigging, though . You can fix the cable by soldering or twining the wire ends back together, if you want. Some heat shrink will prevent them shorting.

Good, thanks; more data is always useful.

The gpu swap will be interesting to witness.

One thing, though; we've compared vbioses and found nothing. Yet ... ever noticed different sizes of vbios .roms? They're tiny and have no 'normal' size value, unlike a 4MB or 8MB bios size. These would fit 1:1 on the bios eeprom, but that is not true for the vbios; your 780M is 162KB, but that does not match any existing eeprom. A 2MB is possible, 1MB common, 512KB exists and so do 256KB, 128KB, 64KB ... You get the drift; vbios code size hardly ever matches the chips they're stored on.

Quick list of some .roms:

Code:

94.5 KB
103 KB
162 KB
166 KB
181 KB
189 KB

Now, a bios uses padding to make it fit the eeprom. Clearly, these vbioses don't ... ergo; what else is there? That peculiarity was there all the time, just hadn't noticed it before ... but after a cursory glance at techPowerUp's Video Bios Collection:

• AMD vbioses all neatly match standard eeprom sizes.
• Nvidia vbioses are anything-goes and will leave spare storage area on the eeprom.

To read that ' beyond-vbios-space' the nvflash tool is useless, so we need a generic eeprom tool (like flashrom) or direct read with programmer. As for the hidden code itself; spooky-spooky ... the suspense is murderous !!!

Yes, although a few were different, too. But the 5555 is the bulk of the corruptions ... weirdly, too, that Mr. Fox's edid seems to corrupt the same byte over and over again. Actually, that address was affected on every corrupted edid seen so far, suggesting it has some meaning to the code writing to it. Perhaps address 14 serves a special function on the vbios?

Indeed.

Remember Microsoft's UEFI to-do list?



Questions;

1. Was one of the 10-ready updates an nvidia driver?
2. Was nvidia driver re-installed or updated after the 10-ready updates?

Hm, and we need to search the 10-ready updates for anything relating to display, i2c and/or gpu. Anyone know a good database of all updates, including those that may have been surreptitiously pulled along the way?

Curiouser and curiouser, but we're getting closer all the same .

 

The three of us were using Windows 7 at the time. I don't think that 7 did Nvidia drivers through update.

 

Just throwing it out there, both my cards stock vBios where 163kb I've now got SVL7 vBios on both. And are you working on the flashrom software =]?

 

It installed this time, however the drivers won't load.

It did install, but I get this error now after completed install. Only difference is your on 10 and i'm on 8.1

 

Can't you mod driver by yourself and test? It's easy. I did it several times with my 880M

 

Then grab the drivers and inf from here http://forums.laptopvideo2go.com/topic/31765-v35598-windows-8-64bit-nvidia-mobile-and-desktop/

 

Go over to techinferno and download J95s modded inf... Will save you a lot of trouble...

Sent from my LG-H811 using Tapatalk

 

Could be coincidence but I went through the pages until I got to the one that said Windows 8 (353.12) and that worked perfectly. And more importantly SLI is working great with GTA V although I could do with going back on TechInferno as the core is static at 450Mhz on both.. yet it is running at 950Mhz base until I run GTA V so it's downclocking for some reason

 

Thanks very much, if I need to go to go higher I'll give that a bash.. the one I've got working now is working well

 

I never had problems with that. were you installing the correct version?

 

version? I just went on the website listed above and downloaded the one there. It does say Windows 10 but I assume that means compatible up to 10 rather than 10 specific. Someone (possibly you, I haven't checked) sent a link to those drivers for Windows 8. Regardless thanks to everyone who chimed in on the problem, I've got to a version which has enabled proper SLI utilisation and if need be I'm sure with the information I've been given I could definitely get to a higher version

 

Sorry i thought you were on w10 lol

 

Status Update:

Guys! I may have stumbled onto something completely by accident.

Question: How many of you with this EDID corruption problem have been using EVGA Precision X? I have, for ages, but Windows 10 totally borked something (perhaps a flipped write-protection bit as @t456 has suggested) and my scientific wild guess is that the EVGA Pixel OC Utility is not playing nice with whatever Windows 10 changed and that is allowing EDID corruption when drivers are changed. I never use that feature from Precision X, but it's still there and I have always set EVGA Precision X to load at Windows startup.

Anyhow, it dawned on me that the Alienware 18 has not bricked the LCD since I replaced it a couple of weeks ago. I just assumed the new A12 BIOS update for Maxwell fixed whatever Windows 10 broke, and it may have, in fact, fixed it. But, I noticed that I don't have EVGA Precision X installed or running at Windows startup.

On a whim, I uninstalled EVGA Precision X on the M18xR2, ran CCleaner's registry tool and rebooted. I have since installed 344.75, 355.98 and back to 344.75 and haven't missed a beat. No LCD bricking. Yesterday I was able to brick it "on demand" by changing drivers.

More testing needs to occur to see if my EDID gets corrupted again, but initial impression are positive since merely installing GeForce drivers bricked my LCD like clockwork with EVGA Precision X installed and now it does not.

Another observation is, I no longer have the "Default Display" key in the registry with "BAD_EDID" substring. That's gone, left when I uninstalled EVGA Precision X, and hasn't come back again.

 

Umghh, I did have Precision X installed I think when I installed 10... I have it installed on my Clevo... Uninstalling it for now on safe side... Yesterday 355.82 kept crashing when I tried to use Precision X to OC...

 

I have Bad EDID and never heard of EVGA precission so theyre not related i dont think. But man what a find! I hope others who had bricked EDID confirm they had it installed

Sent from my SM-G900F using Tapatalk

 

I also had PrecisionX installed in Windows 7 when it bricked. It was installed when I had Windows 10 as well.

 

I am pretty sure that @Daniel1983 did also. It's going to suck if we cannot use Precision X anymore, but not as much as having a bricked LCD. Using NVIDIA Inspector is fine, just very inconvenient in comparison.

 

I had Precision X installed and still have it now with 355.98 drivers. I never had it in my win startup though, always launched it manually before starting a game. So far the screen is still fine, but I am uninstalling Precision X just to be on the safe side. Thank you for the update, Mr. Fox.

 

Well, I have cycled through 327.23, 344.75, 345.20, 355.98, 344.75, 345.20 and 355.98 since around 6:00 AM, with a mixture of cold boots and reboots and everything is holding steady, no bricked LCD and no 8-beep crap. This is in stark contrast to having to flash the EEPROM with a good EDID about 4 or 5 times yesterday, each subsequent to a driver version change.

I still have the morbidly wicked 780M SLI throttling on any driver newer than 345.20 as I have since March, but that's another battle.

 

im running 355.98 on my 17 r1 with win 8.1 since win 10 bricked my screen. This was before anyone had any idea what the problem was so i bought a 120hz screen to replace my 60hz that died. I didnt have evga precision x installed and it still bricked. The only thing i did was install my 980m i bought from woodzstack and use ddu to remove drivers before gpu swap. I thought i had installed the card wrong but still got 8 beeps when putting my 770m back in. so i think we have to do more testing before we can conclude that evga precision is the cause. i dont think ive ever had precision x ever installed on my 17

 

No, it's absolutely not the cause. I've used it for ages without incident, as I already mentioned. But, based on what I can see it was causing repetitive EDID corruption, almost daily, after Windows 10 did it's nasty cancer thing to make my LCD subject to corruption. Windows 10 created a EDID corruption vulnerability that never existed before. I believe this is why it is happening on Windows 7 and Windows 8 now. Well, it apparently was until now. We shall see over the next day or two.

I just wish I would have discovered this before installing nasty poop Windows 8.1 in pure UEFI mode... waste of time for a worthless pile of crap OS, LOL. If this holds, will be dumping that and going back to Windows 7 again in very short order. I can't handle Windows 8's sucky physics performance... drives me nuts just knowing it sucks even if I am doing nothing but web surfing.

 

Hmmm, I had Precision X on Windows 8.1, but ran it manually. After an upgrade to Windows 10, I did a reset and did not installed Precision X again. So, if it's true and Precision X is causing EDID corruption, then that's why my LCD is still fine.

 

@Mr. Fox, can you try using MSI Afterburner? The new PrecisionX isn't based on Afterburner anymore.

 

It may be a contributing factor in the frequency of the soft-bricking, but it certainly isn't the root cause.

 

Yes, I am going to test it. It's actually pretty much worthless to me because it lacks voltage control. The major lack of MXM overclocking functionality of MSI Afterburner is why I stopped using it. I'll be using NVIDIA Inspector since it does everything EVGA Precision X does in terms of overclocking, except for it being very clunky and inconvenient. Of course, it works great if you go the extra step and create batch files, just not "on the fly" like Precision X. The batch files work nice with the TactX keys as well. One-button GPU clock changes FTW.

If EVGA had not tied it's monitor overclocking in with the new design and kept it as a separate utility like it used to be, we could just opt to not install that component. In fact, I will check to see if I saved an old version of it and see if I can use that without having issues.

That was Windows 8.1 though... Windows 10 is what started the mess... after the cancer OS was installed everything started to unravel. Precision X never caused any issues for me whatsoever until Windows 10 created a new vulnerability for EDID corruption. It looks as if maybe Brother @t456 nailed it with his speculation of a possibility of a write-protect bit getting flipped by Windows 10.

 

Ya I had precision X also installed both times. I was hopeful when HWiNFO Was mentioned because it would be a very easy fix. It would definitely be nice if that was the cause. Also very funny if Nvidia didn't use its own software on their test rig so they couldn't get it to brick haha.

 

I'm so glad it's not HWiNFO64... I would be totally lost without that utility and ThrottleStop... both are absolute indispensable and unmatched by anything else, and their authors are my heroes, LOL.

Like this...

 

I've definitely got my fingers crossed. You still plan on fixing screens after a fix is found? Very eager to get that bad boy working again.

 

what version of precision x are you talking about @Mr. Fox? the RTSS era or the post-RTSS era? so we're now thinking it might be caused by this app?

anyone tried win 10 with nvidia inspector?

 

As I suspected, nothing bricked but it's no more useful than before in the absence of voltage control. No point in having MSI Afterburner installed.

Yes, definitely...

 

What's wrong with Nvidia inspector? mind you I say that and my memory won't increase it just resets once I click apply clock and voltage

 

The more recent versions (I had the latest installed) that do not use RTSS and which have the integrated EVGA Pixel Overclocking are what appears to be causing my LCD EDID corruption over and over again (once or more every day, and every time I install a GeForce driver). I am running RTSS 24/7 as a stand-alone application for HWiNFO64 OSD and that's not causing any issues. This is not EVGA's fault because it was fine and did not cause EDID corruption until Windows 10 created a weakness or vulnerability.

Nothing at all. NVIDIA Inspector works flawlessly, but it is inconvenient and cumbersome to use compared to Precision X for quick clock speed, voltage and power/temp target changes. Precision makes identical changes to both GPUs in an SLI system (synchronizes them) and with NVIDIA Inspector it takes longer and then you have to repeat the process on the second GPU. That's why I make NVIDIA Inspector batch files once I have my overclock settings dialed in to where I want them. The batch files turn a pain in the butt to a one-click overclock. NVIDIA Inspector would be more convenient if it allowed you to save profiles. The batch files are a replacement for user-saved profiles.

 

Hi all.
I've had two 120Hz LCD's corrupt and have not had precision X or nvidia inspector installed on any of them.

They did however have MSI afterburner as a temp monitoring program.

 

so that means the usual OC apps are out and my guess is that the only viable OC combo is nvidia inspector + RTSS + HWinfo.

i am ok with nvidia inspector as a basic OC tool as it doesn't run while the system is on all the time and therefore not hogging CPU resources.

 

Yes I seen your Macro key setup for overclocks.. would never have thought to use them in that way it's pretty cool =] post below threw a spanner in the works in regards to precision or inspecter not being installed :/ also any ideas why memory clocks won't stick? core will stay put once I click apply but Memory just shoots back to default when I hit apply?

 

Corrupted by Windows 10? If so, that's the root issue and the source of all the mayhem that occurs under Windows 7 or 8. Once Windows 10 has ruined your computer, you're screwed until we find a fix. I never had any of these problems before that filthy piece of dog poop was installed and opened a door of vulnerability to EDID corruption under any OS. Now that the vulnerability exists, it appears EVGA Precision X is doing something to corrupt the EDID that it wasn't capable of doing before.

 

I've got Nvidia Inspecter and RTSS for OSD and mine is all good.

 

i have windows set the OC when i log in. if something causes the GPU to crash and lose the clock i can just run the shortcut again.

we need confirmed cases and tests done.

 

I have no idea why it does not stick. I've not experienced that issue before. I would try re-flashing the vBIOS on both GPUs if you haven't already. I'm able to change the memory clock even with the stock vBIOS. It's core clock +135 limit and voltage lock that are the main problems with stock vBIOS.

That works fine, but is not a solution for me because I don't overclock my GPUs except when I am benching. There is no point in overclocking for gaming with the additional GPU power of SLI on tap, but I can see where that might be useful on a single GPU system. Plus, I use a variety of overclock settings for benching. Look at how many batch files I use, LOL. Max stable overclock varies by benchmark.

Go back and read posts #2172 and 2178. I've already confirmed EVGA Precision X is causing my EDID corruption every day (once or more). We need a day or two, or three, to pass before we can put our nickle down, but uninstalling it stopped the problem immediately. It's not installed on my Alienware 18 and the new LCD hasn't bricked once since it was installed. Windows 10 did not damage my Clevo and was never installed on my M18xR1 and both of those machines are using the same version of EVGA Precision X without issue because they were not made vulnerable by Windows 10 cancer.

What we don't know is where the vulnerability resides. I think @t456 speculation that Windows 10 flipped a write protection bit is a good guess. It could also be some nasty Windows 10 malware that got tucked away in the surplus vBIOS EEPROM storage space. We know it was not the system BIOS because replacing my motherboard changed nothing, so it almost has to be a change made to the LCD panel or something hidden/persistent in the vBIOS chip. I'm guessing my GPUs are hosting the problem in the surplus vBIOS chip space. I haven't tried moving them to the M18xR1, but that machine is running like a Swiss watch and my appetite for upsetting the apple cart to find out if putting those GPUs into the M18xR1 moves the problem to that machine is very low right now. I've got enough headaches with just the one machine being screwed up, LOL.

 

@Mr. Fox - the last EVGA PrecisionX that was just like Afterburner (before PrecisionX was rewritten) is 4.2.1 which is quite old now. Interestingly enough, the newer versions had serious issues for me on my desktop (for the 780 Ti, haven't tested on my 980 Ti), so I haven't tested the last couple of releases since they have completely ignored our bug reports for months.

@mrsweet1991 - Memory OC not sticking is a common thing if you had an unstable OC and a TDR crash. This happens on my desktop, and the ONLY way to clear it is to full power off then power back on. Restarting (reset button, or Windows restart) doesn't clear the issue for whatever reason.

 

On the contrary - If it isn't on the motherboard, it is either on the LCD or on the GPU. And since you have changed out the LCD for a new one (as well as the mobo) then it is pretty cut and dry that it has to be on the NVIDIA card's VBIOS.

Thus we have an NVIDIA video card based issue that came about because of Microsoft Windows 10.

The way to know for sure is to get an Nvidia card that has never touched Windows 10 and try it to see if the LCD EDID corrupts again. Can anyone let @Mr. Fox borrow one for testing purposes?

 

Old, but it still worked well and included the voltage control that MSI Afterburner is missing. And, it so happens that I have that bad boy stored in my Google Drive... [ Download EVGA Precision X 4.2.1] No reason to avoid using it if it works right. I haven't tested it yet. I have uninstalled Afterburner and everything is still functioning as it should be. I am going to wait a few days before testing anything else so we can have a better idea of whether or not these EDID corruption issues are resolved on my M18xR2.

If memory serves me correctly, (which is questionable now because this has been such a mess,) all of my LCD panels were bricked by Windows 10, including the replacements. So, we haven't ruled out the LCD if that is the case. All of them have been restored by flashing except for the Alienware 18. It was replaced with a new one and has never been "touched" by Windows 10.

Edit: the one I got from @andrewsi2012 has never been bricked by Windows 10. It was on the M18xR1 to begin with. After I learned how to flash the EDID with Linux, I put a repaired LCD on the M18xR1 and have been using the "virgin" LCD from andrewsi2012 for hot-swap booting on the M18xR2 when I have needed to re-flash the LCD. The other LCD (three SEC5448 in total) is in NVIDIA's possession right now.