Custom Resolution Utility (CRU)

[ToastyX]

(08-06-2024 05:45 AM)kurtdh Wrote:  Ok here is test with HDMI connected to the FO32U2P. So you're saying any monitor that uses DSC won't work with CRU at all? I didn't know that was a limitation. What good is this file to you if you can't fix it? Just so I can see info but not change it?

The issue is specific to NVIDIA and is mentioned in the first post. Hopefully NVIDIA will fix it eventually or I find some workaround. Meanwhile this works for custom resolutions but not custom refresh rates: https://www.monitortests.com/forum/Threa...Editor-SRE

You can import this file in CRU to see the extension block:
  FO32U2P.dat (Size: 1.15 KB / Downloads: 71)

Here is a file for HDMI with DSC disabled:
  FO32U2P-no-DSC.dat (Size: 1.15 KB / Downloads: 81)

Can you check if importing the second file and restarting has any effect? If it works, 3840x2160 @ 200/240 Hz should be disabled, but 165 Hz should be available. Run the EDID-test.exe afterwards and post the test.txt. I just want to check if it's actually ignoring the EDID override completely or if it's just not allowing custom resolutions with DSC. Last time I checked it was ignoring EDID overrides completely unfortunately.

In theory it should be possible to work around this for HDMI by using a programmable EDID emulator device with a file exported from CRU, but I don't know if any can handle the HDMI 2.1 bandwidth and two extension blocks, and the programmable ones tend to be expensive.

[Jen]

Good day ToastyX

So I have several questions about HDMI 1.4 and pixel clock limits.

Im currently using a dual gpu system (non mux) with an intel uhd 630 and gtx 1650, and I wanted to know if custom resolutions above 300 or 340 mhz would work, since the hdmi out comes from the intel gpu. Im aware there isn't a pixel clock patcher for intel, but I think I heard you mention something about HDMI support display block a while ago. Would setting the max TMDS clock bypass this? I just want to make sure since my new 1440p 360hz monitor still hasn't arrived and I plan to use a custom resolution of 1920x1080@250 for competetive gaming, since that's the max on hdmi 1.4 at 10bit with yuv 4:2:0.

I used this tool to figure that out: https://tomverbeure.github.io/video_timings_calculator

Also Regarding this tool, the 1920x1080@250 is 543 mhz, obviously far higher than the advertised limit for udh630 and hdmi 1.4; however, It claims that it's possible to work either way. Is this because it doesn't count the gpu pixel clock limit or is this a program error?

Thanks in advance

[ToastyX]

(08-06-2024 08:14 PM)Jen Wrote:  Im currently using a dual gpu system (non mux) with an intel uhd 630 and gtx 1650, and I wanted to know if custom resolutions above 300 or 340 mhz would work, since the hdmi out comes from the intel gpu. Im aware there isn't a pixel clock patcher for intel, but I think I heard you mention something about HDMI support display block a while ago. Would setting the max TMDS clock bypass this? I just want to make sure since my new 1440p 360hz monitor still hasn't arrived and I plan to use a custom resolution of 1920x1080@250 for competetive gaming, since that's the max on hdmi 1.4 at 10bit with yuv 4:2:0.

I used this tool to figure that out: https://tomverbeure.github.io/video_timings_calculator

Also Regarding this tool, the 1920x1080@250 is 543 mhz, obviously far higher than the advertised limit for udh630 and hdmi 1.4; however, It claims that it's possible to work either way. Is this because it doesn't count the gpu pixel clock limit or is this a program error?

Well the pixel clock limit is assuming 8 bpc 4:4:4. 4:2:0 uses half the bandwidth, so in theory it can be done, but in reality there isn't a way to add custom 4:2:0 resolutions for HDMI 1.4 other than 3840/4096x2160 @ 50/60 Hz. Technically 4:2:0 wasn't supported until HDMI 2.0, but some drivers implemented 4:2:0 for 3840/4096x2160 to allow 50/60 Hz with HDMI 1.4.

[Jen]

Alright I see. Thank you so much for your support

I'll see what can I do with yuv 4:2:2 instead

[kurtdh]

(08-06-2024 01:16 PM)ToastyX Wrote:  

(08-06-2024 05:45 AM)kurtdh Wrote:  Ok here is test with HDMI connected to the FO32U2P. So you're saying any monitor that uses DSC won't work with CRU at all? I didn't know that was a limitation. What good is this file to you if you can't fix it? Just so I can see info but not change it?

The issue is specific to NVIDIA and is mentioned in the first post. Hopefully NVIDIA will fix it eventually or I find some workaround. Meanwhile this works for custom resolutions but not custom refresh rates: https://www.monitortests.com/forum/Threa...Editor-SRE

You can import this file in CRU to see the extension block:

Here is a file for HDMI with DSC disabled:

Can you check if importing the second file and restarting has any effect? If it works, 3840x2160 @ 200/240 Hz should be disabled, but 165 Hz should be available. Run the EDID-test.exe afterwards and post the test.txt. I just want to check if it's actually ignoring the EDID override completely or if it's just not allowing custom resolutions with DSC. Last time I checked it was ignoring EDID overrides completely unfortunately.

In theory it should be possible to work around this for HDMI by using a programmable EDID emulator device with a file exported from CRU, but I don't know if any can handle the HDMI 2.1 bandwidth and two extension blocks, and the programmable ones tend to be expensive.

Can I import both of those files, or is it just one or the other? To be clear, you want me to hook up HDMI, import the second file, do nothing else in CRU, reboot, run edid-test.exe and post the test.txt file correct? Also, 165hz won't work with 10-bit color without DSC correct, so I would have to be using 8 bit in order to use 165hz with no DSC? Also, do I need to use restart64 after I do the import? Is the import and restart64 actually what will disable DSC on my monitor?

[ToastyX]

(08-07-2024 12:27 AM)kurtdh Wrote:  Can I import both of those files, or is it just one or the other?

Each file contains a complete EDID, so you only need to import one.

(08-07-2024 12:27 AM)kurtdh Wrote:  To be clear, you want me to hook up HDMI, import the second file, do nothing else in CRU, reboot, run edid-test.exe and post the test.txt file correct?

Yes. Also running this and posting the edid-dump.txt would help as well: https://www.monitortests.com/edid-dump.zip

(08-07-2024 12:27 AM)kurtdh Wrote:  Also, 165hz won't work with 10-bit color without DSC correct, so I would have to be using 8 bit in order to use 165hz with no DSC?

3840x2160 @ 165 Hz at 10 bpc is possible without DSC with YCbCr 4:2:2, but not 4:4:4. That's assuming this even works.

(08-07-2024 12:27 AM)kurtdh Wrote:  Also, do I need to use restart64 after I do the import?

You can run restart64.exe or reboot. Either way.

(08-07-2024 12:27 AM)kurtdh Wrote:  Is the import and restart64 actually what will disable DSC on my monitor?

Only if NVIDIA's driver is not ignoring EDID overrides for the monitor. Otherwise it won't have any effect. That's what I want to check.

[kurtdh]

Ok, I plugged in HDMI 2.1 to the FO32U2P. I set the monitor to 12 bit 240hz 4k. I opened up CRU. I navigated to FO32U2P (active). I clicked import, selected FO32U2P-no-DSC.dat. I clicked OK to exit the program. I clicked restart64. I then ran edid-test followed by edid-dump. Files are attached. Please let me know what your findings are.

Secondly: I've been speaking with one of the admins of blurbusters. He owns a XG27AQDMG 240hz 1440p oled monitor. He claims the override in CRU works just fine for him at 1440p 240Hz 10-bit on DisplayPort 1.4. This means he's using DSC, but CRU works just fine for him (that's what he says.) I thought CRU didn't work with DSC, so why is it working for him?

[Jen]

(08-06-2024 08:53 PM)ToastyX Wrote:  

(08-06-2024 08:14 PM)Jen Wrote:  Im currently using a dual gpu system (non mux) with an intel uhd 630 and gtx 1650, and I wanted to know if custom resolutions above 300 or 340 mhz would work, since the hdmi out comes from the intel gpu. Im aware there isn't a pixel clock patcher for intel, but I think I heard you mention something about HDMI support display block a while ago. Would setting the max TMDS clock bypass this? I just want to make sure since my new 1440p 360hz monitor still hasn't arrived and I plan to use a custom resolution of 1920x1080@250 for competetive gaming, since that's the max on hdmi 1.4 at 10bit with yuv 4:2:0.

I used this tool to figure that out: https://tomverbeure.github.io/video_timings_calculator

Also Regarding this tool, the 1920x1080@250 is 543 mhz, obviously far higher than the advertised limit for udh630 and hdmi 1.4; however, It claims that it's possible to work either way. Is this because it doesn't count the gpu pixel clock limit or is this a program error?

Well the pixel clock limit is assuming 8 bpc 4:4:4. 4:2:0 uses half the bandwidth, so in theory it can be done, but in reality there isn't a way to add custom 4:2:0 resolutions for HDMI 1.4 other than 3840/4096x2160 @ 50/60 Hz. Technically 4:2:0 wasn't supported until HDMI 2.0, but some drivers implemented 4:2:0 for 3840/4096x2160 to allow 50/60 Hz with HDMI 1.4.

Hey ToastyX

So is there no known way of getting around the 300.99 mhz limit on intel? Is it at least possible to get it up to 340mhz? I've researched on it quite a bit but couldn't find anything useful.

It also seems that changing the bit depth or color format doesn't affect the mhz, at least according to that tool I used.

[ToastyX]

(08-07-2024 04:36 AM)kurtdh Wrote:  Ok, I plugged in HDMI 2.1 to the FO32U2P. I set the monitor to 12 bit 240hz 4k. I opened up CRU. I navigated to FO32U2P (active). I clicked import, selected FO32U2P-no-DSC.dat. I clicked OK to exit the program. I clicked restart64. I then ran edid-test followed by edid-dump. Files are attached. Please let me know what your findings are.

Did it have any effect? I'm finding that Windows recognizes the EDID override is there, but NVIDIA's driver seems to ignore it.

(08-07-2024 04:36 AM)kurtdh Wrote:  Secondly: I've been speaking with one of the admins of blurbusters. He owns a XG27AQDMG 240hz 1440p oled monitor. He claims the override in CRU works just fine for him at 1440p 240Hz 10-bit on DisplayPort 1.4. This means he's using DSC, but CRU works just fine for him (that's what he says.) I thought CRU didn't work with DSC, so why is it working for him?

I looked more into it and NVIDIA says this: "NVIDIA DSR, NVIDIA DLDSR and NVIDIA Image Scaling are supported when DSC mode is enabled, if the pixel rate needed to drive the display mode does not exceed the GPU's single head limit. If GPU uses two or more internal heads to drive the display, NVIDIA DSR, NVIDIA DLDSR and NVIDIA Image Scaling are not supported."

That doesn't explain why they need to disable EDID overrides, but it seems tied to this. They don't say what the single head limit is, so I don't know the exact logic that NVIDIA is using to disable EDID overrides and DSR/DLDSR. All I know is everyone with this problem has a monitor with DSC, but apparently it has more to do with the bandwidth required for the highest resolution and refresh rate. Regardless of the cause, the limitation is definitely on NVIDIA's end and not the monitor or CRU.

[ToastyX]

(08-07-2024 12:25 PM)Jen Wrote:  So is there no known way of getting around the 300.99 mhz limit on intel? Is it at least possible to get it up to 340mhz? I've researched on it quite a bit but couldn't find anything useful.

That would require patching the driver, but that would also cause issues with anti-cheats blocking the driver.

(08-07-2024 12:25 PM)Jen Wrote:  It also seems that changing the bit depth or color format doesn't affect the mhz, at least according to that tool I used.

The pixel clock is the number of pixels per second, but the actual bandwidth required is halved with 4:2:0, and 4:2:2 is 2/3rds of the bandwidth of 4:4:4.