sager (Sager NP8358F2/Clevo PB51DF2) Strange bug regarding display output. Doesn't output to external monitor depending on GPU setting in BIOS...

[JQM]

If it's a VBIOS issue, I don't think I can do anything about that. Sager refuses to help and nVidia defers to the manufacturer (Sager).

[JQM]

I tried some other VBIOS (Tongfang, ASUS, and Gigabyte) for the hell of it. All of them resulted in a brick each time. Seems to be something about the PCI-E location being different on each laptop? Thankfully I was able to, with TeamViewer, just put my regular VBIOS back each time and it'd come back normally. I had to get a special version of nvflash that basically ignores all actual checks on what it's doing since the checksum and PCI-E location were different. 😓

[Maks0098]

what kind of nvflash is this that ignores all this? I would need it

[JQM]

5 hours ago, Maks0098 said:

what kind of nvflash is this that ignores all this? I would need it

I found it here: https://www.techpowerup.com/forums/threads/cant-override-pci-subsystem-id-with-6.262665/page-2
https://drive.google.com/file/d/1fYE-t51O7fYNpRAcUZydpMaVXiFa3DrA/view for the download link specifically.

 

Note that I don't recommend doing this specific thing. 😅 It was very risky to flash seemingly random VBIOS. Hopefully it helps fix your issue though/

[JQM]

Well for science I tried a much less risky thing today. Ran DDU to completely remove my drivers, then restarted, so I was using the Microsoft Basic Display Adapter drivers. Obviously not good. Tried installing the "desktop" driver for my GPU from Nvidia's site.

 

Same issues under Discrete, no OS. I think it's actually the same driver, judging from the downloaded file being the same size as the notebook one. Why are they even different options then? Weird...

 

I truthfully do not know how these Chinese companies make their mini PCs from laptops at this point---am I cursed with one from a rare vendor that just doesn't allow the laptop to have full functionality without an LCD? Did the Chinese companies hack the BIOS or do something to the internal wiring? I don't know.

[JQM]

I found three VBIOS with the same DeviceId, since you're supposed to look for this, apparently, I didn't know this previously, that at least loaded graphics under MSHybrid, but had various issues otherwise (they didn't allow proper HDR, and had other issues in the NVIDIA Control Panel).

 

All had the same issues under Discrete. VBIOS won't solve this, or at least any VBIOS I can conceivably find. Wesley from Sager said he was going to contact the Chinese team (odd since Clevo is a Taiwanese company, Tongfang is the Chinese one) and inquire if there's anything that can be done last week, no response yet...

[incendery]

I suspect it may not be a vbios thing at this point, it may be something with the EC and enabling the GPU.

[JQM]

58 minutes ago, incendery said:

I suspect it may not be a vbios thing at this point, it may be something with the EC and enabling the GPU.

What is EC?

EDIT: Oh, Embedded Controller? Like the Super I/O? 😞

[JQM]

Well unfortunately I don't think this issue will actually be "fixable". Sager's coy replies insinuate an LCD is needed. I doubt the actual screen itself, the whole 15.6" thing, is needed, but probably something on the PCB of it. The EDID data, maybe? I don't know. They refuse to say the specific thing because they want me to send it in and pay a thousand dollars (at least) in repairs.

 

But there's another large issue and no trickery like the above can fix it.

When I took my screen off last year I did a thorough deep clean of my laptop, repasted it, you know, the works. Performed good. It's the one year anniversary now. How are things?

 

My CPU on average performs 8% worse than last year, using Cinebench and video encoding tests. For some reason, the percentage changes for the worse, and it's 13% if undervolted. I assume my CPU needs more power than it did last year, I don't know.

 

But it gets worse...
My GPU on average performs 60% worse than last year, using video encoding and upscaling tests.

 

I noticed this happening a few months ago... Back in March, I was regularly playing high demand games in the emulator "RPCS3", for PS3 games. By May already I had noticed that I couldn't play it anymore. After like ten minutes, it'd go to a crawl.

 

Temperatures weren't too unusual. 86C on the GPU as always, because it's set to never go higher, and 95C on CPU, similar deal, set to never go higher. It would literally always be at this, it wasn't throttling back in March. It's a gaming laptop, it performs at its highest temperature if it's being used to its limits.

 

So this explains when it started to degrade, fast. I'm not surprised it is degraded so badly either. Remember that when the short happened to kill my Wi-Fi card and screen, I was on the "Discrete" setting. Meaning my NVIDIA GPU was connected directly to the shorted screen, and dead Wi-Fi card, which it also shared a PCIe connection with too.

 

I'm reminded of a quote that fits here:
“It is possible to commit no mistakes and still lose. That is not a weakness; that is life.” - Jean Luc Picard

 

I'm at a loss of words. I don't have any backup plan.

[JQM]

PROBLEM. SOLVED.

After much arguing with Sager and them insisting an LCD was needed, I consulted ChatGPT (not always the best idea for a final decision) and a hobbyist friend of mine who makes handheld gaming consoles on what the "LCD" meant that it needed.

 

It was likely some form of "EDID" data, data about what the screen is supposed to be. Hmm, from the PCB that the eDP cable plugs into? There's a chip there that holds data, generally. It could be looking for refresh rate or other things, but that'd be saved... on the PCB?

YES.

https://imgur.com/Mf7Lkai

 

That's the bottom part of my broken LCD. The PCB. Cut the two flat wires that were going to it, since they weren't needed (using some tracing I was able to see they had to deal with the backlight, which wasn't working anyway). And... look at this:

https://i.imgur.com/NyGCYqu.png
IT'S ON DISCRETE, NOT MSHYBRID!

 

I need to make a better way to keep this, since obviously the bare PCB exposed is not a good idea. I have some ideas how to do this and it's not an issue.

 

And yes, there was a massive performance increase from doing this. Activities involving the GPU are performing at a breakneck pace compared to on MSHybrid.