Custom Resolution Utility (CRU)

[ToastyX]

(01-29-2020 08:11 PM)Drasko Wrote:  Greetings,

I have the AOC 24G2U 144Hz monitor and an AMD gpu (RX 5700), so when I set the refresh rate to 144Hz the memory no longer down clock (it's fine for 120Hz though). Using DisplayPort cable that came with the monitor.

I tried all 3 settings for LCD monitors, but only the 'Automatic - LCD standard' seems to let the gpu memory down clock. The problem is that as soon as I start a 3D game the monitor goes black and says 'Input No Support'. I have to blindly alt+F4 out of the game and select the DisplayPort as input again in order to gain back picture.

Is there anything I can do to make 144Hz work?

If the refresh rate works on the desktop, then the same refresh rate should work in games. The only thing I can think of is maybe the monitor can't handle those timing parameters with FreeSync enabled.

[Drasko]

(01-30-2020 09:40 PM)ToastyX Wrote:  If the refresh rate works on the desktop, then the same refresh rate should work in games. The only thing I can think of is maybe the monitor can't handle those timing parameters with FreeSync enabled.

You're right, I disabled the FreeSync and the game works fine now, kind of a bummer, thanks for the help though

[joevt]

I tried CRU to override an EDID for a display. It works great. The only major issue is that it supports only 3 extension blocks. The Apple Pro Display XDR has 6 extension blocks (7 blocks total). The E-EDID specification says there can be 255 extension blocks.

Some minor suggestions:
- for data blocks that CRU doesn't understand, it could show at least hex values and maybe allow editing them (the hex should not include the tag, extended tag, length, or OUI).
- For TV resolutions, there is a check box to make all resolution only 4:2:0 or only not 4:2:0 (RGB/4:4:4/4:2:2 depending on the bits in byte 3 of the CTA extension block). There exists a third option that allows a resolution to have both. This is done using a third CTA extension block called Y420CMDB which contains flags for each SVD in all the VDBs that also support 4:2:0. For the user, the simplest UI might be to have the VDB, Y420VDB, and Y420CMDB all be editable at the same time using the same dialog. Each SVD has two checkboxes (or a menu with three options) which indicate if it goes into the VDB, the Y420VDB, or both the VDB and the Y420CMDB. Extra blocks can be created automatically if the number of SVDs exceeds the max for each block type. Additional CTA extension blocks may need to be created automatically. For simplicity, the user should not be able to move a VDB above or below a different VDB, in order to maintain the order of the flags in the Y420CMDB block(s).

For reference: edid-decode has been updated recently to decode almost everything in an EDID (it's missing Apple, Nvidia, and AMD vendor specific data block decoding). Most or all the specs used by edid-decode are publicly available.
https://git.linuxtv.org/edid-decode.git/about/

edit: max number of extension blocks for Apple Pro Display XDR is 6

[Bryan Stephanov]

Hello friends, will there be any way to have support for HDMI 2.1 and have dynamic HDR metadata and be able to edit them? also support for SMPTE ST 2086? Maybe in a new version of CRU? Greetings!

[ToastyX]

(02-07-2020 07:13 PM)Bryan Stephanov Wrote:  Hello friends, will there be any way to have support for HDMI 2.1 and have dynamic HDR metadata and be able to edit them? also support for SMPTE ST 2086? Maybe in a new version of CRU? Greetings!

HDMI 2.1 uses the same data block as HDMI 2.0 but with additional features. I don't have access to the HDMI 2.1 specification, so I don't have enough information yet on what the additional features are, but I want to include them eventually.

The HDR dynamic metadata block isn't very interesting. It only lists metadata types and versions. I might add it later.

[ToastyX]

(02-07-2020 09:37 AM)joevt Wrote:  I tried CRU to override an EDID for a display. It works great. The only major issue is that it supports only 3 extension blocks. The Apple Pro Display XDR has 6 extension blocks (7 blocks total). The E-EDID specification says there can be 255 extension blocks.

That's because the last time I checked, both AMD and NVIDIA would not read more than 3 extension blocks. It's rare for a monitor to have more than 2 extension blocks.

(02-07-2020 09:37 AM)joevt Wrote:  - for data blocks that CRU doesn't understand, it could show at least hex values and maybe allow editing them (the hex should not include the tag, extended tag, length, or OUI).

I thought about that but haven't gotten around to it. My implementation would let you edit everything including the tag. The length would be automatically calculated.

(02-07-2020 09:37 AM)joevt Wrote:  - For TV resolutions, there is a check box to make all resolution only 4:2:0 or only not 4:2:0 (RGB/4:4:4/4:2:2 depending on the bits in byte 3 of the CTA extension block). There exists a third option that allows a resolution to have both. This is done using a third CTA extension block called Y420CMDB which contains flags for each SVD in all the VDBs that also support 4:2:0. For the user, the simplest UI might be to have the VDB, Y420VDB, and Y420CMDB all be editable at the same time using the same dialog. Each SVD has two checkboxes (or a menu with three options) which indicate if it goes into the VDB, the Y420VDB, or both the VDB and the Y420CMDB. Extra blocks can be created automatically if the number of SVDs exceeds the max for each block type. Additional CTA extension blocks may need to be created automatically. For simplicity, the user should not be able to move a VDB above or below a different VDB, in order to maintain the order of the flags in the Y420CMDB block(s).

I'm aware of the capability map, but implementing a dialog that manages multiple data blocks at once is complicated, so this is not a high priority for me. If I did implement it, I would probably put a "4:2:0" check box next to the "Native" check box when adding/editing a resolution, which would set the appropriate bit in the capability map. This would be grayed out and always selected if it's set to be a 4:2:0 data block. Right now the only 4:2:0 resolution that seems to be in common use is 3840x2160 @ 60 Hz, so it doesn't seem worth the effort to implement all that for one resolution.

(02-07-2020 09:37 AM)joevt Wrote:  For reference: edid-decode has been updated recently to decode almost everything in an EDID (it's missing Apple, Nvidia, and AMD vendor specific data block decoding). Most or all the specs used by edid-decode are publicly available.
https://git.linuxtv.org/edid-decode.git/about/

That's good to know.

[Bryan Stephanov]

(02-07-2020 11:04 PM)ToastyX Wrote:  

(02-07-2020 07:13 PM)Bryan Stephanov Wrote:  Hello friends, will there be any way to have support for HDMI 2.1 and have dynamic HDR metadata and be able to edit them? also support for SMPTE ST 2086? Maybe in a new version of CRU? Greetings!

HDMI 2.1 uses the same data block as HDMI 2.0 but with additional features. I don't have access to the HDMI 2.1 specification, so I don't have enough information yet on what the additional features are, but I want to include them eventually.

The HDR dynamic metadata block isn't very interesting. It only lists metadata types and versions. I might add it later.

That would be great! really thanks for your work,it is much helpful for most people in here.

[Seneral]

Hello,
I need help getting a cheap 2k LCD screen of 1440x2560 portrait mode to work. It only has one mode it works at, and CRU seems to identify it correctly. But Windows and my AMD drivers won't recognize it correctly, after some fiddling around they finally acknowledge it as a 1440x2560 display but the timing seems to be off (atleast the timing in the radeon software settings differ greatly from the one CRU reads). I verified on a linux machine which does perfectly recognize the LCD that the timing CRU reads is indeed correct.
Now it sometimes just stays black although it is adressed as a 1440x2560 display and LCD led is green indicating no error, sometimes it does actually show a solid image with visual glitches (image persisting until I change a setting or reconnect). Just now I also had a case for the first time where the image with the glitches shows, but flashes at varying frequencies (frequency changes at irregular rates, but image generally shows longer than the black screen does, usually for at east 200-500ms, but as low as 5-10ms). And another new visual glitch, the stripes as usual and rare flickering of the upper half of the display in correct orientation (so upper half was displayed on lower half and vice versa). All in all seems to be the timing.

Here are some images: https://imgur.com/a/ywDMWC1
1. CRU with the correct timing
2. Linux with working LCD showing same timing
3. What radeon software automatically detects
4. Trying to enter CRUs timing information, won't let me accept it, says display doesn't support it. Did I do something wrong / interpret values wrong?
5. The visual glitches I'm talking about (except for the flashes, but when it flashes these visual glitches are also there)

Can anyone help me? Why can't I set the correct timing values in radeon software? I did ask over at the AMD forums before in case you're wondering. Thank you very much!

[jcpq]

Hello
Many thanks for the application.
I have a doubt:
When formatting the disk / windows, the created resolutions will be erased, right?

[ToastyX]

(02-13-2020 01:55 PM)jcpq Wrote:  Hello
Many thanks for the application.
I have a doubt:
When formatting the disk / windows, the created resolutions will be erased, right?

Right.