Unable to modify AVR's EDID (I could previously)

[ashlar]

Hi! I am trying to understand what happened to my configuration...

I have an LG 77C1 and a Denon X3800H. The LG used to connect to the AVR and the AVR connected to the LG.

After a small amount of time, the display got oversaturated colors. So I reset everything and reinstalled the backup I had done when everything was set up as I wanted it. Using restart-64 froze the machine. I went to safe mode, deleted all modifications and tried doing everything from scratch.

Alas, any modification to the AVR EDID, followed by a restart-64, brought the machine to a freeze.
I can't add anything, I can't delete anything. But I did that previously and it worked.

I tried completely uninstalling the video drivers (Nvidia RTX 3070 here) and reinstalling them from scratch. Nothing.

I attach the original EDIDs for both the AVR and the TV and the modified EDID for the AVR that used to work perfectly.

I need to slightly alter refresh rates for both 23.976 and 24.000 in order to have smooth playback with no dropped or repeated frames and... well, basically it's proven impossible now (I resorted to connecting the PC directly to the TV, using a DP-HDMI cable to get audio to the AVR but that's less than ideal).

[ashlar]

Hmmmm... a couple of clarifications: everything I described above was done through CRU. The files I have attached in the previous post are CRU exports, not the actual EDIDs. I realized that now and I am attaching to this post the actual EDIDs for both devices.

The thing that absolutely baffles me is that I changed nothing and the AVR went from being editable in CRU to freezing the machine from even the single slightest edit (like deleting one TV resolution, which I need to do in order to substitute it with a detailed one with the correct refresh rate).

EDIT: my ignorance on display... the EDID and CRU exports are the same thing... I thought it was different but the identical size prompted me to open the files and... I learned something. Sorry for this useless second post, basically

[ToastyX]

NVIDIA has two different bugs that could be the issue:

1. The Denon EDID only has 1 byte left for data blocks in the extension block. NVIDIA's driver has a bug that can cause it to crash or ignore any changes if there's less than 2 bytes left.

2. Some people have reported any change causing the driver to crash even when the EDID is fine. The weird thing is it doesn't happen to everyone, so I don't know what the trigger is. Some people have reported using DDU and reinstalling the driver fixed the problem.

[ashlar]

(02-13-2023 03:31 AM)ToastyX Wrote:  NVIDIA has two different bugs that could be the issue:

1. The Denon EDID only has 1 byte left for data blocks in the extension block. NVIDIA's driver has a bug that can cause it to crash or ignore any changes if there's less than 2 bytes left.

2. Some people have reported any change causing the driver to crash even when the EDID is fine. The weird thing is it doesn't happen to everyone, so I don't know what the trigger is. Some people have reported using DDU and reinstalling the driver fixed the problem.

The really strange thing is that this started happening all of a sudden. I had been spending days modifying the EDID with no problem whatsoever. And then... stop.

Today I tried DDU and installed old drivers (457.30).
GPU connected to the AVR and the AVR to the TV.

With these old drivers the 23.976 refresh was different than with the newer ones. I thought something else had changed, without me realizing it, so I updated to the newest ones (528.49) and the refresh rate went back to what it used to be. I mean they both are 23.976xxx but it's the xxx that changes and that's important for smooth playback in a HTPC scenario.

What I don't understand is:

if the EDID that matters is the one in the AVR, why is it that changing drivers modifies the refresh rate (surely different timings are used)?

Edit: I say "the EDID that matters is the one in the AVR" because CRU doesn't even "see" the TV connected to the AVR (although madVR does.

EDIT 2
I suppose what I'm trying to say is this:

If changing the drivers changes the refresh rate for 23.976, I would expect being able to do something on the GPU side to further modify it. Surely the driver did not change the EDID of the AVR or could it?

[ashlar]

ToastyX do you have any info on what was the last driver without the 2 bytes bug from Nvidia?

[ToastyX]

All versions have that bug, but the problem only happens with EDID overrides, which you can edit with CRU to avoid that problem. The refresh rate issue sounds like something they changed in the driver, not the EDID. You would have to report that to NVIDIA.

[ashlar]

(02-21-2023 04:39 AM)ToastyX Wrote:  All versions have that bug, but the problem only happens with EDID overrides, which you can edit with CRU to avoid that problem.

You mean editing the EDID to a state where it has at least 2 byes left?

[ashlar]

(02-23-2023 09:28 AM)ashlar Wrote:  

(02-21-2023 04:39 AM)ToastyX Wrote:  All versions have that bug, but the problem only happens with EDID overrides, which you can edit with CRU to avoid that problem.

You mean editing the EDID to a state where it has at least 2 byes left?

To answer my question: yes, you meant that. And it works perfectly.