Custom Resolution Utility (CRU)

[l337fool]

(08-08-2023 03:06 AM)ToastyX Wrote:  

(08-08-2023 01:12 AM)l337fool Wrote:  Appears to read the EDID and the extension block (CTA) shows up but it seems changes made there have no impact.

Does deleting the "Extension override" data block make a difference? What about if you leave the data block and add two additional extension blocks?

This whole mess is caused by the HDMI 2.1 standard adding an extension override data block to override the extension block count, so the extension block count is both one and three at the same time, depending on whether the GPU's driver supports the data block or not. Then on top of that, NVIDIA implemented the data block by modifying the reported EDID in strange ways. This issue mainly affects RTX 3000/4000-series GPUs, which I don't have, so I can't test this myself.

I'm working on a better solution to read all the extension blocks, but every solution has some problem. If I implement the extension override data block as the HDMI 2.1 standard intends, older GPUs will ignore additional extension blocks added by the user. If I delete the data block, newer GPUs will ignore the additional extension blocks that exist by default. If I change the main extension block count, older GPUs won't be able to read the additional extension blocks anyway, so that would cause the "Default extension block" problem with older NVIDIA GPUs.

That leaves me with two solutions:

1. Create a separate extension block list in the CTA-861 extension block for additional extension blocks that newer GPUs will read but older GPUs will ignore. This would best represent what's actually happening, but this would definitely confuse people and would be annoying to implement.

2. Delete the extension override data block and change the main extension block count if the additional extension blocks are readable. That would be the simplest to implement and should work with all GPUs, but this would change how the EDID is represented and hides what's actually happening. It also technically violates the EDID 1.3 standard without an extension block map, but GPU drivers don't seem to care about this anyway.

I feel like the HDMI Forum didn't take into account what kind of impact this data block would have. I'm not the only one scratching my head trying to figure out the best way to implement this.

Just a suggestion but maybe provide a fork for new GPUs (perhaps in the form of a 2nd exe in the zip folder labeled as such). It sucks when a standardization organization and/or manufacturer drop the ball like this. In my particular situation all I am looking to do is change the report PQ values of my display which are incorrect (clipping at 950 nits when it should be 1150 nits). Whatever you ultimately decide to do that provide me with the ability to address that concern will be greatly appreciated.