***EVGA Precision X and Windows 7/8/8.1 and especially 10 bricking systems***

***Update 11/1/2015 - It seems like its mostly EVGA Precision X causing the problem. If you are using Precision X, grab nVidia Inspector and use it instead - http://www.guru3d.com/files-details/nvidia-inspector-download.html - Inspector has been long respected and I have not heard of any bricking thus far... Still proceed with caution with Windows 10 but @Mr. Fox and a number of others have reported that its a combination of Precision X and new drivers causing the bricks***

I have been debating on putting this up here since it didn't seem to be affecting our machines but it appears that it definitely is. Something is causing LCD EDIDs to be corrupted (making them non-functional) and worse, it appears that there is actual BIOS corruption occurring. It is a widespread issue on Alienware machines (see this thread) but I have confirmation of a few Clevo machines that have been bricked as well.

The models that have been reported are ZM and SM-A series. This doesn't mean that other machines are not susceptible and the recommendation at this time is to avoid Windows 10 until the root cause has been found and fixed. Proceed with caution and if your machine has been affected, please post here and let us all know how your reseller or system builder handles your RMA.

 

Bit of a placeholder reply to this, but I want to speak to our R&D before I reply in full. However at this time I can report that we have had no known issues as described in the above post with Windows 10 and ZM models. I don't know about people upgrading from their original OS to Windows 10 with older model laptops, hence wanting to speak to our engineers in more detail.

 

@ajc9988 had to RMA his P770ZM after upgrading to Windows 10.

Considering the fact that machines have been bricked from Clevo, Dell, Alienware, HP, Lenovo, and Sony laptops, as well as Gigabyte desktop boards after installing Windows 10, its better to be safe than sorry.

 

I'm not denying that a problem may exist, I am simply trying to be as helpful as possible here. Not sure how I should have worded my reply differently so as to not cause offence?

1/ we have had zero reports of this problem with ZM issues and we have a very large data pool to draw conclusions from

2/ I don't know about older models which were launched before Win 10, hence I will speak to our engineers about it

I don't have any details of ajc9988's problem or if he is one of our customers - if he is then we will of course provide as much support as possible, if he isn't then he should contact his reseller/SI. I can forward information posted in this thread to Clevo, but it should really come from the reseller/SI because of the large number of variables.

P.S. it may be helpful to differentiate between laptops purchased with Win 10 installed and owners who update from 7 or 8 to 10 - the other thread suggests the problems are occuring with updates to Win 10. There is a Win 10 specific BIOS and EC for the ZM models, not sure if this has been discussed - I can't read through 156 pages :-(

 

It isn't a huge problem for Clevo systems yet. There's only a handful of reports at this point. With that said, many Clevo owners won't upgrade because they are gamers and Windows 10 is still slower than Windows 7 and Directx 12 functions poorly on nVidia cards so there is no real incentive to upgrade. Not to mention most of us know not to use a new OS version until at least Service Pack 1. Pulling up Google and putting in Windows 10 brick brings up everything from displays going out to hard drives being killed to system BIOS corruption across a wide range of systems. I got lucky I didn't have any issues with it but I have an eDP display which may have saved me.

 

is it safe to install on lecacy bios instead of secure boot?

 

The most recent Clevo failure that actually ended up corrupting the system BIOS was in legacy. Mr. Fox has blown 4 displays in his Alienware, also in legacy. It seems indiscriminate.

@XMG My phone won't let me quote your PS without erasing all over the place so I thought I would tell you that it has all been upgrades, not machines that shipped with it. Its interesting Clevo has a BIOS and EC for Windows 10.

 

I think it's a good decision to put up a warning or caution. If the number of Clevos is three displays all in a short time span, that seems a little more than random. Of course, there is a lot unknown right now but being cautious, delaying an upgrade seems wise.

No, Legacy BIOS is not safe. @Mr. Fox has lost four displays on Legacy in his Alienwares. He had issues with W10 on his Clevo P570WM but no bricked display.

Edit Ninjaed

 

This is so sad. It's horrible that serious problems (even damage) is being randomly caused by Windows 10 to both notebooks and desktops.

I would encourage anyone with NVIDIA GPUs that has run into display panel problems with their notebooks after a Windows 10 upgrade to also post the details about the problem in the thread linked below.

Windows 10 + GeForce Drivers are Killing Samsung and LG Notebook LCD Display Panels| GeForce Forum

 

Can you shed a little more light on this Windows 10 specific BIOS for the ZMs, please? What does it do differently if anything? What are the recommendations for W10 install with this BIOS, upgrade OK, clean install, BIOS flash first, et cetera?

 

very good question, ive already considered to update my bios/ec to that win10 version, but have refrained from it thus far, sticking with premamod...

 

Does Prema have a specific W10? I don't have 10 on my machine. Also, I'm with a Prema partner ~ Mythlogic. So, I should be getting BIOSes from them..

 

doesnt look like it yet...

prema´s latest v2 is based on 1.03.15 BIOS and 1.03.09 EC, whereas XMG´s W10 versions are 1.05.01 Bios / 1.05.01a EC / 1.05.01 EC2

got the former flashed and the latter downloaded and in my BIOS/EC archive, but im not changing ANYTHING in my current config until that terrible W10 bios / display edid corruption thing is solved... same goes for display drivers and boot mode

 

I'm on 1.05.01 MYT (Mythlogic). since I got it a month ago.

 

sounds good! still being careful for now also, i prefer premamod over latest stock

 

At this moment in time all I can really say is that 1.05.01 provides full support for Win 10, therefore we ship all Win 10 systems with this BIOS/EC and recommend that customers who upgrade from 8.1 to 10 also update their BIOS/EC. Based on the information that we currently hold, there is nothing to suggest that running Win 10 on an older BIOS & EC will cause any problems and updating is not essential.

We are already discussing the problem reported by Ethrem with Clevo, I think it's best to wait until these discussions are concluded before discussing further. A lot of our customers have updated to Win 10 themselves and there are no reports from them either about issues, so we really need to get specific details and numbers of who and what has been affected.

Apart from that there isn't really much more I can recommend. The people on here who have had an issue with a Clevo laptop should get in touch with their SI / reseller and so they should have more information on the issue than XMG will.

 

The 2x ZM systems that I have seen, which developed black screen problem in W10, where both on the 1.05.01 BIOS/EC (which btw Clevo still hasn't pushed upstream onto their mainstream server, yet).
That being said, the only real difference with the W10 BIOS base is a fix for a "potential bug" in the secureboot code...which can not even be used under Legacy Boot.
So BIOS revision shouldn't really matter in this case.

I was traveling for the past 2 months and just came back to base.
Will also be looking into this W10 thing as soon as I get a chance...looks like W10 is granted low-level write access to the LCD firmware and for some reason overwrites it (maybe in conjunction with NVIDIAs new g-sync LCD communication and as long as the specific LCD has its firmware writing PIN connected)...anyone with this problem please PM me.

Also if anyone can provide detailed info about the exact LCD panel model that was in use or has the ability to dump the LCD firmware from bricked panels throw them right into my inbox.

 

Indeed, I noticed this too but it has been supplied directly to SIs.

What we need is chassis number, full specification etc etc then we can pass this info on to Clevo. @Prema - you're the only person on here who has seen this issue first hand (apart from @Ethrem but are they the same chassis that you have both seen?) so please do share info via pm. At the moment all we have is reports of the issue, we need some solid information to investigate further.

 

OK, I got the 'before' and 'after' dump info from a P7xxZM user with Chi Mei N173HGE-L11 (CMO1720) screen.

"W10/NVIDIA" has overwritten his LCD internal EDID, which was using the EDID revision 3 standard, with one based on revision 4. Main problem here apart from writing 'adjusted' frequency values was that "W10/NVIDIA" broke the checksum while doing so.

So whatever is generating the updated EDID table is not correcting the checksum and thus rendering the LCD 'unbootable'.

@XMG I'll upload the dumps and data for your tech-team. EDIT: DONE

P.S.: The screen was working just fine after re-flashing the EDID part of its firmware manually via Linux!

 

@Prema Excellent, thanks. This is the kind of information that we have been hoping to get hold of!

 

BTW, the 'wrong' checksum is generated in a way that would be an appropriate routine for a NVIDIA vBIOS with advanced header data...

 

Thanks Prema for stepping in and helping out as always!

 

Yeah thx for getting a step closer to this issue. Don't want to risk Installing W10 again, using W8.1 right now with 355.84 Dev drivers.

Gesendet von meinem Nexus 9 mit Tapatalk

 

Do you know what can be done with this information? Is there any way your BIOS or vBIOS could prevent EDID overwriting, or can anything else be done now that you know this?

 

W

I guess it could be done via BIOS or by simply cutting a wire to the screen, but best would be to disable the writing feature via registry in W10 to target everyone at once...but there are people better suited than me to dig up the right register...

 

Perhaps the best solution is for a whole bunch of people to boycott Windows 10 and let Micro$oft and their cancer OS fail real hard. That sorry outfit has become pert near worthless. Think about it... we have to use Linux to fix their idiotic mistakes that mess up LCD firmware. What's wrong with this picture, LOL. How much complete and utter crap are we willing to put up with for a DX12 API?

 

Get game designers to adopt Vulcan!!!

 

The root cause is bad hardware engineering really... Write command to EDID EEPROM should not be allowed (since there is no legit use for it on consumer PC). Or EEPROM with write-protect enabled should be used.
Software tends to have bugs and hardware should be resilient to software-induced breakage and can't just depend on software never doing incorrect thing.

Unfortunately its just easier for manufacturer to connect some cheap EEPROM without write protect directly to DDC/I2C bus (esp. when EEPROM already uses I2C itself) - allowing all commands to be sent. I had same trouble when I've tried to upgrade GPU on Thinkpad - due to VBIOS bug it sent write instead of read, overwritten ~10 bytes at start of EDID EEPROM and panel went into test mode. Luckily this was just a known EDID header and I was able to reflash it using external monitor.

I recommend backing up the EDID for your panel, using etc PowerStrip utility.

 

I disagree, at least partially. Should it have been write protected? Yes, of course, but that is not a legitimate excuse. Should I lock my front door? Of course, but if I don't, that does not give a burglar authority to waltz into my living room uninvited.

Should Windows 10 and/or NVIDIA be attempting to write ANYTHING ? Heck no! That is a very serious trespass. An operating system or device drivers should not ever be a piece of trojan malware that carries a firmware altering payload. The extent of their "authority" should be to add folders and place files in drive space and add or modify registry entries, and even then only if and when we ask them to by downloading and installing whatever they are offering... if we actually feel like doing so. The moment they do any alterations to any component (including their own branded hardware after it leaves the factory) at a firmware level without full disclosure, advance knowledge and express permission of the computer owner they become an extremely unethical, untrustworthy and deviant vendor that deserves whatever manner of adversity such acts of indiscretion may bring upon their brand.

 

Do you really think they NVIDIA or Microsoft issuing write command by intent? Its most likely a bug.
They should not attempt to write anything, but changing read command to write in I2C protocol is just a single bit flip - and multitude of bugs can cause this quite easily. My point was, display panels not supposed to accept write commands into their EDID EEPROM by DDC protocol - it was never intended for that. If they actually do, it just bound to eventually cause trouble (and it did).

Its a bit akin placing your computer BIOS into file on usual file system, and saying "deleting this file is not allowed, if you do, your PC will be bricked and you will need specialized hardware to recover". Is it unrealistic? Certainly not. Is it bad non-robust design? Absolutely yes.

 

Who said anything about legal? This is a civil case, not criminal. It is called a tort.

 

Sorry, I didn't meant "illegal" in any conjunctions to the courts/law, just in general sense.

Not sure if you can even have a legal case here, arguably both sides broken rules here (panel manufacturer by allowing DDC writes to affect EEPROM, OS/driver developer by issuing the write command)

 

This is why you name everyone as a party and allow them to fight about indemnification...

 

Hello Everyone

So far, we only have information from one customer and this was kindly supplied through Prema, but I need a lot more information from you guys to help Clevo investigate this further. Please can you pm me the information as requested below, there's no need to post it in the forum as it may get lost in the discussion:

Clevo chassis model

Chassis serial number (not essential but it would help a lot)

Panel type (is it the same as came with the alptop or have you upgraded yourself)

BIOS and EC (including any non stock Clevo versions or vBIOS etc)

GPU model

GPU driver (version, GeForce or Clevo driver)

OS version

OS upgrade or clean install, or did you buy the OS installed with the laptop?

A quick description of the issue, for example "installed Windows 10, after updating the GPU driver the LCD panel died". Don't need an essay here (!), just a brief outline of the events leading up to the issue. The main points that we are trying to establish are:

- only in Win 10 (so far YES)

- only with Nvidia GeForce drivers and NOT with stock Clevo GPU drivers (so far YES)

- only when a customer upgradess to Win 10 or carried out a clea install (so far YES, but it seems Windows is not the cause)

Please Note: I ask for this info in order to provide Clevo with as much as possible, to enable them to continue an investigation. None of the information that you provide can be linked back to you personally or your order - therefore whatever you tell me can not in any way affect your warranty. If you could pm the infomration in the same format as above, it will make my life 1% easier

 

Filed it

 

My old acer laptop too died after upgrading to Windows 10. It wont boot pass the boot splash screen anymore. I already did everything I know but just wont boot anymore nor able to enter setup.

 

Another "upgrade" then.... Dunno but it seems to manifest mainly in upgrades.

Only speculating, no hard facts for sure!

 

Hmm very strange indeed. I upgraded mine to win10 with no issues

 

The individual act of upgrading to Win 10 isn't causing the issue as far as we can see.

Just one pm received so far on this (from @Scerate , thanks!) - we could do with some more customer examples and info ( see http://forum.notebookreview.com/thr...-bricking-systems.781467/page-4#post-10095091 above ) please let me know if anyone can provide their information in the above format so we have a larger pool of data to work from. Thanks

 

Question:
Would this be how nvidia is controlling the over clocking of the lcd in the nv control panel?
That it somehow needs write access to the lcd for this. Just curious.
@Mr. Fox
We still use dos also.

 

It might be whatever mechanism they are using to overdrive the LCD that is causing EDID corruption. It could also be something with G-Sync features in the drivers that is causing it. Whatever they're doing, it's pretty messed up. They haven't released a driver that works well since Q1 of this year. Before the LCD bricking problem is was 780M SLI throttling to 405/400 for me. It's going to suck to have to downgrade to AMD GPUs that deliver 2012-level performance and don't overclock worth a damn just to have basic functionality, but it's starting to look that way.

@XMG - info posted here by @t456 might useful.

 

Does this affect LG eDP IPS displays? Since I have a non G-Sync 980M, I have the display overdrive at 75hz with Windows 10 clean install. Do I have to worry about my machine bricking at a later time?

 

thus far the only definitive trigger that has been identified in this is the usage of a current version of precision x on a win10(ready) machine with newer nvidia drivers...

Sent from my Nexus 5 using Tapatalk

 

No plans to overclock 980M right now with Windows 10.

 

Inspector hasn't had any reported problems.

 

Has this issue been resolved yet? You guys might have seen my thread - this happened to me using Windows 7. I believe it was due to EVGA Precision X and the latest Nvidia drivers. Managed to reflash the screen in Linux.

Eventually I'd want to upgrade to Win 10 for DX12, but not as long as a this issue is still unresolved.

 

according to mr. fox it looks like his previously bricked system is stable now without precision x....

Sent from my Nexus 5 using Tapatalk

 

Sorry, I had to move on the drop of a dime so I didn't have a chance to update the post... Title changed, disclaimer posted.

 

PrecisionX, I never liked that tool, it has the buggiest GUI I have ever seen. Like I'd click on profile 1 to save it, then change a few settings to save to profile 2, then click back on profile 1 and the overclocking settings are all back to default. Seems like a 12 year old is writing the code for PrecisionX

 

You must not have much experience with Afterburner then... PrecisionX offered everything that Inspector did (except driver mods) without the annoying P states resetting your overclock every time you would apply it and have to do it one or two times more to get it to stick.