Monitor Tests Forum

(09-30-2022 01:48 PM)ToastyX Wrote: [ -> ]You can just copy the EDID from the LG to the EDID emulator, but HDMI 2.0 can only support 3840x2160 @ 60 Hz at 8 bpc with RGB/YCbCr 4:4:4. 120 Hz would only be possible with YCbCr 4:2:0. Normally these are defined as TV resolutions in a CTA-861 extension block.

Thanks for the reply Toasty, I appreciate it

I see, I recall that I had something like this working before with this emulator, but I did a driver clean and the resolution is now gone.



I have already cloned the EDID (hence it is appearing as an LG TV), but I can only select the HDMI 2.0 compliant resolutions. Also, I had created this through Nvidia Control Panel but I can't seem to recreate this anymore (as it always picks 4:2:0 without HDR, even for 2560x1440). I was then trying the cru tool to create this, but the resolutions don't show up in the Windows 10 resolution menu at all.

Basically, I think it was sending a scaled HDR10 2560x1440 @ 90Hz signal to the emulator. I think this is fine, since the stream was still 4K HDR10 @ 90Hz.

Summarizing my question, it's not possible to push 4K 120 HDR over HDMI 2.0 due to pixel clock limitations, but I am wondering if I can still stream this resolution. For that I need to be able to select this resolution from Windows regardless of the signal sent to the emulator. Not sure if my understanding is correct here though.

Many thanks!

(10-01-2022 09:26 AM)semaphore Wrote: [ -> ]I have already cloned the EDID (hence it is appearing as an LG TV), but I can only select the HDMI 2.0 compliant resolutions. Also, I had created this through Nvidia Control Panel but I can't seem to recreate this anymore (as it always picks 4:2:0 without HDR, even for 2560x1440). I was then trying the cru tool to create this, but the resolutions don't show up in the Windows 10 resolution menu at all.

Basically, I think it was sending a scaled HDR10 2560x1440 @ 90Hz signal to the emulator. I think this is fine, since the stream was still 4K HDR10 @ 90Hz.

Summarizing my question, it's not possible to push 4K 120 HDR over HDMI 2.0 due to pixel clock limitations, but I am wondering if I can still stream this resolution. For that I need to be able to select this resolution from Windows regardless of the signal sent to the emulator. Not sure if my understanding is correct here though.

CRU can't create scaled resolutions, but what you can do is use CRU to get rid of all the 4K resolutions and make 1920x1080 the native resolution. Then you can enable NVIDIA DSR 4x to scale 3840x2160 down to 1920x1080. That should allow you to get 120 Hz and 10 bpc with HDR.

(09-28-2022 08:13 PM)ToastyX Wrote: [ -> ]I don't know if it will work. CRU is meant for overriding monitor EDIDs, but a CRT TV doesn't have an EDID unless the adapter provides one. CRU can still create EDID overrides for monitors without EDIDs, but that doesn't seem to work anymore with Windows 11. If you have Windows 10 or earlier and the GPU supports VGA signals and interlaced (K2200 should), then it might be possible.

If you see "PNP09FF - Generic Non-PnP Monitor (active)" with the TV connected, then it might work if you add 480i. To get 480i with the Automatic timing modes, you have to enter 1440x240 @ 59.94 Hz with interlaced checked. CRU only supports separate sync, but I assume the adapter converts that to composite sync if it supports VGA signals.

It's possible to damage a fixed-frequency CRT with the wrong parameters if it doesn't do any checks, but most displays are designed to reject unsupported signals. Usually it just won't display correctly if the timing is off, or it won't display anything at all.

Thank you for the answer, I have already carried on a little.

Yes I'm on windows 10. The k2200 supports interlaced scanning from the DVD-I output (I'm using a DVI 24+5 to VGA adapter).
I already have feeded the SD CRT with all sort of resolutions (also years ago when I bought the UMSA), it never got any damage.

About the UMSA sync, this is from official page features:

"Composite Video Sync Generatin by logic. Thus a wider range of graphic cards and TV are supported. This results in more stable signal as well. "

Source: https://arcadeforge.net/UMSA/UMSA-Ultima...anguage=en

When I managed to see a perfect picure on the CRT tv (res. was 320x240 however, so windows is not usable), I think I used CVT timing but not remember for sure (I was also testing many manual timings i found on web).

So, let's get to the point. Connecting the UMSA to the PC is like connecting nothing. The adapter is connected to the electric current and when connected to the CRT Tv it automatically change the input to AV1 (the source where it is connected), I cannot even change it. Unfortunately it doesn't seems to send anything to the PC for making it aware.

That's the main reason I never managed to make it work. But now on the quadro I get the workstation tab on nvidia control panel, see:

[attachment=1008]

It is pratically a software EDID emulator. I can export a EDID, then import it on any input and windows recognizes it as a connected TV/monitor.

So I exported a couple EDID (from a HD-CRT Tv and a LCD Tv that should have PAL/NTSC standard res.) and the plan is to Edit it with CRU (making PAL or NTSC as the "first choice"/ native res), then load it on the DVI-I port while I'm using the display port monitor. Easy and safe (safe because when changing to monitor 2/UMSA if I don't get a correct picture on the CRT I'll just leave Windows to revert back).

About this, do you suggest to edit the EDID file, then load it on Nvidia CP, or should I load it first then EDIT with CRU while Windows is seeing it as a already connected monitor?


As I said I'm new to these things, so i have a few more questions if you don't mind

It is the EDID of a display a fixed file? I'm asking because on CRU I have i.e. 3 times the edid of the LCD Tv (the Benq monitor I'm on now have even 4). Exporting them, they have 3 different MD5. Shouldn't they be the same? Perhaps if I connect the PC with different cables (or extracting using another PC) it could get a different EDID file?

I also noticed that exporting a EDID through Nvidia CP and CRU (both as TXT, since Nvidia CP only exports as TXT), the MD5 doesn't match.
Opening them on Notepad++ though, I see the values are the same but on CRU letters are uppercase and ordered in more lines instead of a long one. Perhaps this is enough for triggering a different MD5?

[attachment=1009]

I hope there won't be some compatibility issues. Do you think it'd be better to export an EDID with Nvidia CP, then edit it on CRU?

Last question: what's the reason i have to put that strange res. instead of just 720x480i / 720x576i?
I'll set the automatic CRT timing. The Tv is a SD Mivar 28" CRT Tv (from 2001). It is only PAL/NTSC (NTSC colors only from RGB input), not progressive scan support. It has very minimal electronics, no flicker filter, abl or any digital alteration, a RGB signal directly drive the cathodic ray (you cannot even adjust the color saturation when using a RGB signal). Great image quality

(09-18-2022 02:28 PM)ToastyX Wrote: [ -> ]

(09-18-2022 02:05 AM)egagah Wrote: [ -> ]I have dual monitor setup and RTX2070super as below:
PRISM+ X270 PRO (running at 2560x1440 @ 119.988Hz)
AOC2402 24G2W1G4 (running at 1920x1080 @ 120.000 Hz)
NVIDIA driver vers. 516.94

When I set my PRISM+ to 59.951Hz, my gpu can downclock, but at 119.988, it won't downclock. Both monitors can go to 144Hz but I dare not set this since the GPU will never down clock at those Hz.

Problem arise is for the PRISM+ monitor, the target resolution of 1440p 120Hz above is only found under Extension blocks, and in here, even adding just 10 to the blanking vertical lines will cause the following events AFTER using restart64:
1) PRISM+ goes totally dark
2) AOC flashes quickly, mouse move super slow, flash again, repeat 3-5 times, then PC freezes.
A restart after above event will cause the pc to load up to the login screen where it just freezes the moment the windows time appears on the AOC screen (PRISM+ stays dark the entire time).

Normally for the GPU to clock down with multiple monitors, both monitors have to be perfectly synchronized so the vertical blanking periods happen at the same time, but this is hard when the resolutions don't match. I don't know how it clocks down at 59.951 Hz because that doesn't match the other monitor exactly. Freezing would be an NVIDIA driver bug, but that might be happening if the monitor isn't staying connected properly. If the PRISM+ can't handle custom parameters, that makes it much more difficult to synchronize both monitors, and I can't even guarantee synchronizing them would allow the GPU to clock down because the driver might not handle all cases where the resolutions are different. If the PRISM+ can handle a very slight tweak to the timing parameters, making all the parameters and the pixel clock a multiple of 4 would allow making a 1920x1080 equivalent by multiplying the parameters by 0.75 and the pixel clock by 0.5625. Technically these two are perfectly synchronized:




You should also check the TV resolutions and delete 1920x1080 @ 120 Hz if it's in there. I don't know if that's enough blanking for the GPU to clock down, but if it isn't, I'm not sure what you can do if the PRISM+ won't accept higher vertical totals.

Hi Toast! Thanks so much for the detailed reply! So I tried the settings as per the screenshot, no problem when I changed the settings on the AOC display, even did a driver64 reset without issue. But the moment i change the settings for the prism+ display and do a driver64 reset, the entire PC freezes again. So I'm going to guess this is just signs that the PRISM+ monitor cannot be modified at all with CRU? Based on your experience, this is really a sign of the monitor issue or it's NVIDIA driver? Like idk how to find a good driver to test with haha.

By the way, not sure if this is the right way, but I just modified the values in the detailed resolution itself (so for the AOC, changing the 144Hz profile to a 119,584Hz profile as per your screenshot, and the PRISM+ one 59.950Hz to the 119.584Hz profile). Every paramter like front back proch etc, i follow exactly as per the screenshots too.

I also checked the standard resolutions + extension blocks, 1080p 120Hz doesn't exist in X270 pro. It does appear in the extension blocks for AOC monitor though

[attachment=1010]
[attachment=1011]

EDIT: I got it to work using NVIDIA'S panel's built-in custom resolution! This is the settings used, hope it helps us understand more what went wrong haha.
I want to say big thank you Toasty for your big support!

[attachment=1012]

(10-02-2022 04:20 AM)egagah Wrote: [ -> ]Hi Toast! Thanks so much for the detailed reply! So I tried the settings as per the screenshot, no problem when I changed the settings on the AOC display, even did a driver64 reset without issue. But the moment i change the settings for the prism+ display and do a driver64 reset, the entire PC freezes again. So I'm going to guess this is just signs that the PRISM+ monitor cannot be modified at all with CRU? Based on your experience, this is really a sign of the monitor issue or it's NVIDIA driver? Like idk how to find a good driver to test with haha.

If any change causes the PC to freeze, then it's an NVIDIA driver bug. I've received sporadic reports of driver crashes since 471.41, but in those cases I didn't find anything unusual with the EDID, so I don't know what the trigger is. I only know of one driver bug that could cause a crash if the CTA-861 extension block doesn't have at least two bytes left for data blocks, which I noted in the first post.

(10-01-2022 08:24 PM)Peppe90 Wrote: [ -> ]About this, do you suggest to edit the EDID file, then load it on Nvidia CP, or should I load it first then EDIT with CRU while Windows is seeing it as a already connected monitor?

It would be better to load the modified EDID using the NVIDIA control panel because that's more direct. Too bad they only offer that for Quadro cards.

(10-01-2022 08:24 PM)Peppe90 Wrote: [ -> ]It is the EDID of a display a fixed file? I'm asking because on CRU I have i.e. 3 times the edid of the LCD Tv (the Benq monitor I'm on now have even 4). Exporting them, they have 3 different MD5. Shouldn't they be the same? Perhaps if I connect the PC with different cables (or extracting using another PC) it could get a different EDID file?

CRU should only have one active entry for each connected monitor. Inactive entries are leftovers in the registry from previous connections or different drivers. Keep in mind the registry only stores the first block, so CRU can't read extension blocks for inactive entries. If you see a "Default extension block" in the extension block list, the exported file won't have the extension block data. Each type of connection can have a different EDID. VGA will say it's analog. DVI will say it's digital. HDMI uses EDID 1.3 and has a CTA-861 extension block with an HDMI data block. DisplayPort normally uses EDID 1.4. The EDID for each port on the monitor normally does not change by itself, although some monitors might have certain settings that can change the EDID. Keep in mind any changes you make in CRU will appear in the exported file. If you didn't make any changes and the monitor is connected and recognized as active, then exporting a file will save the original unmodified EDID.

(10-01-2022 08:24 PM)Peppe90 Wrote: [ -> ]I also noticed that exporting a EDID through Nvidia CP and CRU (both as TXT, since Nvidia CP only exports as TXT), the MD5 doesn't match.
Opening them on Notepad++ though, I see the values are the same but on CRU letters are uppercase and ordered in more lines instead of a long one. Perhaps this is enough for triggering a different MD5?

MD5 requires an exact match. The best way would be to compare the raw binary data, but the NVIDIA control panel only exports text files. The text formatting shouldn't affect the data as long as the hex values are the same.

(10-01-2022 08:24 PM)Peppe90 Wrote: [ -> ]Last question: what's the reason i have to put that strange res. instead of just 720x480i / 720x576i?

720x480i/576i is normally double-clocked, but there's no way to specify a double-clocked resolution in the EDID. 1440x480i/576i produces a similar signal that the TV can handle. The EDID stores the number of lines per field, so 240 for 480i and 288 for 576i.

(10-03-2022 11:31 PM)ToastyX Wrote: [ -> ]It would be better to load the modified EDID using the NVIDIA control panel because that's more direct. Too bad they only offer that for Quadro cards.

I agree, that should be a obvious feature on any GPU.
So you're saying that's better to edit the EDID file when it is not mounted, then when the edit is done I'll mount it with Nvidia CP. If something's not right, I'll unmount it, make other modifications and so on.
This way I'll also edit only the actual EDID txt, no modifications will be made on registry, right?

(10-03-2022 11:31 PM)ToastyX Wrote: [ -> ]CRU should only have one active entry for each connected monitor. Inactive entries are leftovers in the registry from previous connections or different drivers. Keep in mind the registry only stores the first block, so CRU can't read extension blocks for inactive entries. If you see a "Default extension block" in the extension block list, the exported file won't have the extension block data. Each type of connection can have a different EDID. VGA will say it's analog. DVI will say it's digital. HDMI uses EDID 1.3 and has a CTA-861 extension block with an HDMI data block. DisplayPort normally uses EDID 1.4. The EDID for each port on the monitor normally does not change by itself, although some monitors might have certain settings that can change the EDID. Keep in mind any changes you make in CRU will appear in the exported file. If you didn't make any changes and the monitor is connected and recognized as active, then exporting a file will save the original unmodified EDID.

Thank you for the infos, things are much more clear now. I didn't done any modification with CRU yet. I only exported original EDIDS.

(10-03-2022 11:31 PM)ToastyX Wrote: [ -> ]720x480i/576i is normally double-clocked, but there's no way to specify a double-clocked resolution in the EDID. 1440x480i/576i produces a similar signal that the TV can handle. The EDID stores the number of lines per field, so 240 for 480i and 288 for 576i.

I see. For now what matters is to feed the UMSA/CRT Tv with a resolution on which i can use Windows, then I'll do all other needed testings.

I'd like to play PC games sometimes on the CRT Tv, for that reason I wanted to use a SD resolution (the K2200 is equivalent to the 750Ti, I think it can max out even Cyberpunk 2077 with good performance at 720x480i ), 1440x480i is more demanding I think.

I could also try the default 720x576i resolution from the LCD Tv, perhaps it'll work.
I'll try using the LCD Tv EDID as a base. I'll edit the EDID txt with CRU adding your suggested resolution as native, then mount it with nvidia CP.

(09-30-2022 01:48 PM)ToastyX Wrote: [ -> ]

(09-30-2022 05:52 AM)simon52min Wrote: [ -> ]Could you guide me how to create a 2560* 1440@165Hz & 3440x1440@144Hz bin Files via CRU tool ? My video card is RTX3070ti.

What are you trying to do? You can add those resolutions in a DisplayID extension block and export a .bin file.

What i 'd like to do is connect a virtual display device through the hdmi interface of the 3070 graphics card, to create a virtual desktop of 2560 * 1440 @ 165Hz, or 3440 * 1441 @ 144Hz

Can CRU edit the chromaticity coordinates?

I need it so I can clamp to sRGB within my AMD GPU driver settings (which reads this information). I want to replace the stored coordinates with my measured ones (because the default values are somewhat off)

(10-05-2022 11:45 PM)takla Wrote: [ -> ]Can CRU edit the chromaticity coordinates?

No, but you can export a file and use an EDID editor to edit the values, and then import it back into CRU with "Import complete EDID" enabled. You can use any editor except Phoenix: https://www.monitortests.com/blog/list-of-edid-editors/

I do want to add editing the color coordinates to CRU 2.0.