Custom Resolution Utility (CRU)

[Case]

bobithtidus Wrote:Also there is certainly difference in HDR, which is the mode I am primarily using, the HDR seems to tone map differently, which is why I am wondering the monitor is doing something cheap like using internal adjustment to downgrade certain things to be able to do 240hz, just like how DSC is a visual downgrade over the full bandwidth.

bobithtidus Wrote:and worst part changing the HDR static metadata did nothing, just clipping past 600 nits

Editing HDR metadata won't change your monitor's tonemapping. If it's tonemapping changes in different modes, it's likely because of enabling/disabling HGiG vs actual tonemapping. If you're clipping above 600 nits, you probably have HGiG enabled (that's how it's supposed to behave, assuming 600 nits is your monitor's peak brightness), which is what you want for games anyway.

[ArtMov]

(02-08-2025 04:37 PM)ToastyX Wrote:  

(02-07-2025 10:29 PM)ArtMov Wrote:  I understand, I just see a real difference in responsiveness between hdmi 2.1 and dp 1.4 on my monitor, this is not a placebo, mb allm and qft give this difference? I understand dp 1.4 does not have such functions?
Also, vrr works noticeably better compared to adaptive sync, input feels more responsive, that's for sure, although vrr and adaptive sync do give a delay.

There shouldn't be a difference.

ALLM is a signal to the display to switch to a lower latency mode, such as telling a TV to switch to game mode. ALLM is pointless if the display doesn't lag. If it's a gaming monitor, it shouldn't have any lag at the native resolution with VRR or adaptive sync. That's more of a TV issue.

QFT does not necessarily reduce lag. It would make the scanout faster, but it can also delay the scanout, so there is no benefit unless the frame is forced to render later in the refresh cycle somehow, but the difference would be miniscule at 240 Hz. VRR and adaptive sync already work like QFT for lower frame rates because the pixel clock is fixed and the scanout is delayed until the next frame is ready, so adding more blanking wouldn't be helpful. Also I don't think fast Vactive is even implemented with PC drivers.

thx vm for answer. I found an unpleasant problem specifically on the hdmi connection, when goes into standby mode, deepsleep, the monitor does not wake up either from the input or from the osd button, need to switch the input \ reconnect the cable. Mb it is possible to influence this behavior using cru? I know that in the edid settings there is active off and other settings related to standby mode etc. I think that this is not related to the cable, it seems to be a real 2.1 8k cable from Vcom.

[ToastyX]

(02-11-2025 01:56 PM)ArtMov Wrote:  thx vm for answer. I found an unpleasant problem specifically on the hdmi connection, when the screen goes out (goes into standby mode, deepsleep) the monitor does not wake up either from the input or from the osd button, need to switch the input \ reconnect the cable. Mb it is possible to influence this behavior using cru? I know that in the edid settings there is active off and other settings related to standby mode etc. I think that this is not related to the cable, it seems to be a real 2.1 8k cable from Vcom.

That has nothing to do with the EDID. Either it's a bug in the monitor's firmware, or it's an issue with a driver on the PC's end.

[bli1348]

Finally bought a 4k160hz monitor to hook up on my vega 64. First night windows by default show 3840x2160 60 hz 8bit-tethering. I was mad that there were no higher refresh rate and 10bit option. I used AMD adrenalin custom resolution to set a 4k120hz profile and it worked(verified by ufotest)! Then I went to bed then nightmare began.
The next morning, despite win10 still showing 120hz, ufotest shows only 60hz, however started seeing 10bit option in adrenalin. I tried to switch profile back and forth, create new profile but failed to get 120hz again, seemed adrenalin capped g.pixel clock at 650Mhz instead of ~1100Mhz req'd, however then I realized while monitor/adrenalin shows 71hz ufotest still shows 120hz... totally confused..

Then I resorted to CRU. Firstly I saw range limit of 48-160 hz, 255-255khz, 1490Mhz, yay! 1490M not 650M in adrenalin, there is already an display ID1.3 extension block, inside a detailed resolutions with 2 entries 4k143.997hz and 4k160hz, neither shown in win10 I suppose cuz of the horizontal range limit of 255khz. I added a third entry of 4k117hz to stay at 255khz horizontal freq using exact reduced timing. restart, and then change win10 refresh rate to 117hz, the whole screen image SHAKES left and right with adrenalin showing CRU param of 1020M pixel clock and 117hz refresh. tried the same 117hz on 2k same Shaking left and right...
please where did I do wrong??

[Kas_37]

Hello, ToastyX.
Are you already tested rtx50 series and dsc displays? As I can see, CRU changes now works correctly, not ignored by driver. DLDSR is available also. So nvidia finally changed something at driver side.
Though it seems to bring a lot of problems: black screens, crashes, etc

[ToastyX]

(02-16-2025 10:45 AM)Kas_37 Wrote:  Hello, ToastyX.
Are you already tested rtx50 series and dsc displays? As I can see, CRU changes now works correctly, not ignored by driver. DLDSR is available also. So nvidia finally changed something at driver side.
Though it seems to bring a lot of problems: black screens, crashes, etc

I don't have the hardware to test this. NVIDIA didn't change anything on the driver side. The RTX 5000-series just has a higher single-head limit. Older GPUs still have issues with DSC.

[ToastyX]

(02-14-2025 03:56 PM)bli1348 Wrote:  Finally bought a 4k160hz monitor to hook up on my vega 64. First night windows by default show 3840x2160 60 hz 8bit-tethering. I was mad that there were no higher refresh rate and 10bit option. I used AMD adrenalin custom resolution to set a 4k120hz profile and it worked(verified by ufotest)! Then I went to bed then nightmare began.
The next morning, despite win10 still showing 120hz, ufotest shows only 60hz, however started seeing 10bit option in adrenalin. I tried to switch profile back and forth, create new profile but failed to get 120hz again, seemed adrenalin capped g.pixel clock at 650Mhz instead of ~1100Mhz req'd, however then I realized while monitor/adrenalin shows 71hz ufotest still shows 120hz... totally confused..

Then I resorted to CRU. Firstly I saw range limit of 48-160 hz, 255-255khz, 1490Mhz, yay! 1490M not 650M in adrenalin, there is already an display ID1.3 extension block, inside a detailed resolutions with 2 entries 4k143.997hz and 4k160hz, neither shown in win10 I suppose cuz of the horizontal range limit of 255khz. I added a third entry of 4k117hz to stay at 255khz horizontal freq using exact reduced timing. restart, and then change win10 refresh rate to 117hz, the whole screen image SHAKES left and right with adrenalin showing CRU param of 1020M pixel clock and 117hz refresh. tried the same 117hz on 2k same Shaking left and right...
please where did I do wrong??

What monitor is it? Are you using DisplayPort or HDMI? What timing parameters are you using? Vega 64 only supports HDMI 2.0, which is limited to 600 MHz pixel clock, so I'm assuming you're using DisplayPort. Vega 64 supports DisplayPort 1.3/1.4 but not DSC. DisplayPort 1.3/1.4 without DSC is limited to 1080 MHz pixel clock at 8 bpc, and 864 MHz pixel clock at 10 bpc. The range limits has nothing to do with the problem. You should be able to get 3840x2160 @ 120 Hz at 8 bpc with "Native PC" timing, but there's no guarantee the monitor will accept that.

[ArtMov]

(02-11-2025 06:50 PM)ToastyX Wrote:  

(02-11-2025 01:56 PM)ArtMov Wrote:  thx vm for answer. I found an unpleasant problem specifically on the hdmi connection, when the screen goes out (goes into standby mode, deepsleep) the monitor does not wake up either from the input or from the osd button, need to switch the input \ reconnect the cable. Mb it is possible to influence this behavior using cru? I know that in the edid settings there is active off and other settings related to standby mode etc. I think that this is not related to the cable, it seems to be a real 2.1 8k cable from Vcom.

That has nothing to do with the EDID. Either it's a bug in the monitor's firmware, or it's an issue with a driver on the PC's end.

As it turned out, it was a problem with the nvidia driver, they fixed it in 572.47 waking up from in extended sleep mode.

sorry if this hasn't been discussed before, I have fast vactive and fapa disabled in hdmi 2.1 features, does enabling them give any profit at the moment?

[edidcru42]

Hello,

Newcomer to CRU, discovered it this past weekend during troubleshooting (lack thereof) custom resolution setting now missing from Intel graphics driver setting (unlike the Nvidia and AMD still providing such tweaks).

Here's a quick background on my situation and problem:

I'm using the same "dummy" display adapters at DisplayPorts on PC1 and PC2 computers. PC1 has Nvidia graphics adapter running Windows 10. PC2 has the Intel graphics running in Windows 11. I have another PC3 with Intel graphics in Windows 11, and that also has the same behavior as PC2.

Screenshots below has PC1 on the left. And PC2 at the right.

PC1, without any custom resolution tweak in Nvidia driver setting, shows the resolution of 1920x1200, which this "dummy" display emulator do support. But in PC2, the same dummy DisplayPort emulator is not showing this 1920x1200 resolution.

Intel has (in recent days) removed ability to tweak custom resolutions from their Graphics Command Center. And workaround has been to tell user to use older, beta version of the Intel Graphics Command Center, or point them to CRU. And that's how i found CRU, learning about it, watching view tutorials in YouTube.

I've shown side by side, of CRU for this DP1080P60 dummy display emulator between PC1 and PC2. And trying to understand where in EDID INF difference(s) may lie for PC1 to show 1920x1200 resolution but not for PC2. But more so, how can i leverage the PC1's EDID to cause 1920x1200 to show up in PC2 for the same dummy display.

Anyone has similar experience in trying to create custom resolution this way, if even possible at all?

Trying to troubleshoot why PC2 is not plug-and-play for the same dummy display adapter that lacks 1920x1200, where PC1, granted Nvidia graphics, shows 1920x1200 as supported for the same dummy display.

I can collect and upload more details from CRU to help with this troubleshooting. There are lots of data to go through and learn from CRU.

Thank you!

[ToastyX]

(02-24-2025 09:42 PM)edidcru42 Wrote:  I've shown side by side, of CRU for this DP1080P60 dummy display emulator between PC1 and PC2. And trying to understand where in EDID INF difference(s) may lie for PC1 to show 1920x1200 resolution but not for PC2. But more so, how can i leverage the PC1's EDID to cause 1920x1200 to show up in PC2 for the same dummy display.

Anyone has similar experience in trying to create custom resolution this way, if even possible at all?

The EDID is part of the dummy plug, so the EDID should not be different on different computers. The issue is DisplayPort dummy plugs are usually just single-link DVI dummy plugs with a different connector, like a passive DisplayPort to single-link DVI adapter, so they should be limited to 165 MHz pixel clock.

That leaves four possibilities:

1. NVIDIA is incorrectly allowing higher pixel clocks. If 3840x2160 @ 30 Hz and 2560x1440/1600 @ 60 Hz are available with NVIDIA but not with Intel, then this is the case.

2. NVIDIA is automatically applying reduced blanking to 1920x1200 @ 60 Hz and Intel is not. Normally standard resolutions are not reduced blanking, so 1920x1200 @ 60 Hz would exceed 165 MHz pixel clock. If 1920x1200 @ 60 Hz is available but 3840x2160 is limited to 17 Hz and 2560x1440/1600 are limited to 30 Hz, then this is the case.

3. 1920x1200 is one of the default scaled resolutions with NVIDIA's driver, so it might just be scaling 1920x1200 to 2560x1440 @ 30 Hz. If 1920x1200 is limited to 30 Hz and the active resolution is 2560x1440, then this is the case.

4. The dummy plug is actually DisplayPort, and Intel is incorrectly limiting the pixel clock for some reason. With real DisplayPort, 6 bpc color is usually available as an option, while DVI can only be 8 bpc.

DisplayPort and DVI don't need anything special in the EDID, so you can just delete all the resolutions and extension blocks and just create the resolutions you want. 1920x1200 @ 60 Hz with "Automatic PC" timing should work in all cases since that does not exceed 165 MHz pixel clock.