Custom Resolution Utility (CRU)

[ToastyX]

(02-21-2020 11:00 PM)wacha Wrote:  Long story short, most of my GPUs will only do 1080p using the adapter cable, and I think it's because of the 165Mhz pixel clock limitation of SL DVI. Using your patches I've been able to display 1440p with the supported cards I have in my possesion. The thing is that most of the GPUs I need to display 1440p are older that those supported, even though some of them can use drivers supported by the patchers. Most of the supported cards already do 1440p over HDMI or DP, what I need is to be able to do 1440p with DVI-I only cards over an HDMI converter. Is there a reason the limitation can be overcome with a GTX 400 series card o an HD 5000 series card and not a GTX 295 or HD 3870 X2 that use the same drivers?

Older cards have the limits defined in the firmware, so it's not easy to find where in the driver that's being checked. Newer cards use the values defined in the driver.

There's one trick that might work with older cards if you're trying to do DVI to HDMI:

First, use CRU to delete the extension block, or remove HDMI support from it. That will make it send a DVI signal instead of HDMI.

Then rename the patcher to whatever-patcher-single.exe and patch the driver. That will make it always send a single-link DVI signal.

[ToastyX]

(02-21-2020 04:12 AM)pumchkung Wrote:  I custom res already but when I change res in game,i got black screen how i fix it

If it's a lower resolution, try GPU scaling.

[wacha]

Hi. Thank you for your responses. I've tested some of your suggestions but haven't made any progress.

(02-21-2020 11:59 PM)joevt Wrote:  If you're trying this with a DVI-I port that supports DVI dual link then it makes sense that it won't support greater than 165 MHz for single link because that's where it switches to dual link mode. Are there some GPUs that support single link greater than 165 MHz from a DVI dual link port? I think you found some examples.

DVI-I ports that don't support DVI dual link may support single link up to 340 MHz (HDMI 1.4). This is the case wth Intel motherboards using Intel GPUs that have DVI-I ports that are actually just using the DisplayPort dual mode of the iGPU.

All the DVI-I ports I'm testing should be DL as the GPUs support greater than 1440p digital resolutions, and they only have DVI-I as output. There is one special case that really baffles me. I have three GTX 295s. Whenever I try to install the drivers, they all default to 1440p using the adapter cable, but only one of them is able to give a display. The Gigabyte single-PCB GTX 295 works out of the box, 1440p 60Hz crisp image. The other two won't be able to display anything and I get a "no signal" message. So, my only way to even use one of those two cards is to first install the drivers on the Gigabyte, downgrade to 1080p or less, shutdown, and swap the cards. One of them is a Gainward dual-PCB model so I can understand that there might be hardware differences between the two. The other one is a ZOTAC single-PCB model, exactly identical as far as I can tell. The have the same BIOS version, same PCB layout (even the markings and numbering are the same), same driver version, same monitor, same DVI to HDMI cable, same OS, motherboard and other components. Yet one displays 1440p perfectly and the other just gives up when presented with a greater than 165 MHz pixel clock, it doesn't even default to 1080p like most of my other GPUs.

I also tried deleting the extension block. It showed HDMI 2.0. Then renamed the patcher and applied the patch. It gave me the same results. GTX 295s show a "no signal" image, any other card will show 1440p but with a blurry image and barely unreadable text. I might have done something wrong here, I don't know. I opened up CRU, deleted the only entry in the bottommost box and saved. Then I applied the renamed patch and rebooted.

I know I'm trying to use this software in unintended ways, and that my needs are very specific. I just want to understand why it doesn't work. I can live with it if it's simply not possible, or overly expensive, but the lack of consistency in the behaviour is driving me nuts. So far, these are the results of my testing:

-Gigabyte GTX 295: works out of the box.
-Seemingly identical 295 and dual-PCB: "no signal", nothing works past 165 MHz pixel-clock
-HD 7990: defaults to 1080p, 165 MHz limit. Works with the patcher.
-GTX 590: defaults to 1080p, works with CVT-Reduced blank at 1440p. Patcher fixes it and works with any custom 1440p res.
-GTX 780Ti: works out of the box.
-HD 3870X2: defaults to 1080p, 165 MHz limit. Patcher doesn't work.
-X1950 XTX: same as 3870

Thank you for your time.

[apav]

Hi guys.

Thank you ToastyX for keeping these programs consistently updated.

I recently upgraded from a GTX 770 to a GTX 1080 Ti. My monitor has been running at 110hz using the LCD Reduced setting in CRU for years, without any artifacting or other issues. After swapping the cards out I uninstalled my Nvidia drivers in safe mode, reinstalled them normally, used the Pixel Clock Patcher, added my resolutions to CRU, then restarted. However when I boot up Windows I get these weird artifacts which never happened with my GTX 770.

https://streamable.com/5vxcf
https://streamable.com/sk3rb

Sometimes it happens immediately when it goes to the lock screen, sometimes it happens when I press anything to get to the password entry screen. After that it doesn't seem to do it while browsing the desktop, watching videos or playing games. Again this is at 110hz, I seem to be having no issues anywhere at 60hz. Did I miss something or enter something incorrectly on CRU? Nothing has changed with my monitor, so I'm guessing it is either a software issue or the GPU itself (though not sure what exactly would be the issue if its the GPU).

Please help! Thank you so much.

[ToastyX]

(02-24-2020 10:06 PM)apav Wrote:  I recently upgraded from a GTX 770 to a GTX 1080 Ti. My monitor has been running at 110hz using the LCD Reduced setting in CRU for years, without any artifacting or other issues. After swapping the cards out I uninstalled my Nvidia drivers in safe mode, reinstalled them normally, used the Pixel Clock Patcher, added my resolutions to CRU, then restarted. However when I boot up Windows I get these weird artifacts which never happened with my GTX 770.

https://streamable.com/5vxcf
https://streamable.com/sk3rb

Sometimes it happens immediately when it goes to the lock screen, sometimes it happens when I press anything to get to the password entry screen. After that it doesn't seem to do it while browsing the desktop, watching videos or playing games. Again this is at 110hz, I seem to be having no issues anywhere at 60hz. Did I miss something or enter something incorrectly on CRU? Nothing has changed with my monitor, so I'm guessing it is either a software issue or the GPU itself (though not sure what exactly would be the issue if its the GPU).

That looks like the memory clock is changing when it shouldn't. Try raising the vertical blanking/total and see if that makes it stop.

[ToastyX]

(02-22-2020 01:32 PM)wacha Wrote:  All the DVI-I ports I'm testing should be DL as the GPUs support greater than 1440p digital resolutions, and they only have DVI-I as output. There is one special case that really baffles me. I have three GTX 295s. Whenever I try to install the drivers, they all default to 1440p using the adapter cable, but only one of them is able to give a display. The Gigabyte single-PCB GTX 295 works out of the box, 1440p 60Hz crisp image. The other two won't be able to display anything and I get a "no signal" message. So, my only way to even use one of those two cards is to first install the drivers on the Gigabyte, downgrade to 1080p or less, shutdown, and swap the cards. One of them is a Gainward dual-PCB model so I can understand that there might be hardware differences between the two. The other one is a ZOTAC single-PCB model, exactly identical as far as I can tell. The have the same BIOS version, same PCB layout (even the markings and numbering are the same), same driver version, same monitor, same DVI to HDMI cable, same OS, motherboard and other components. Yet one displays 1440p perfectly and the other just gives up when presented with a greater than 165 MHz pixel clock, it doesn't even default to 1080p like most of my other GPUs.

I also tried deleting the extension block. It showed HDMI 2.0. Then renamed the patcher and applied the patch. It gave me the same results. GTX 295s show a "no signal" image, any other card will show 1440p but with a blurry image and barely unreadable text. I might have done something wrong here, I don't know. I opened up CRU, deleted the only entry in the bottommost box and saved. Then I applied the renamed patch and rebooted.

I know I'm trying to use this software in unintended ways, and that my needs are very specific. I just want to understand why it doesn't work. I can live with it if it's simply not possible, or overly expensive, but the lack of consistency in the behaviour is driving me nuts. So far, these are the results of my testing:

-Gigabyte GTX 295: works out of the box.
-Seemingly identical 295 and dual-PCB: "no signal", nothing works past 165 MHz pixel-clock
-HD 7990: defaults to 1080p, 165 MHz limit. Works with the patcher.
-GTX 590: defaults to 1080p, works with CVT-Reduced blank at 1440p. Patcher fixes it and works with any custom 1440p res.
-GTX 780Ti: works out of the box.
-HD 3870X2: defaults to 1080p, 165 MHz limit. Patcher doesn't work.
-X1950 XTX: same as 3870

Thank you for your time.

The behavior will depend on whether it's treated as DVI or HDMI.

Newer cards will output an HDMI signal over DVI ports if HDMI support is defined in the extension block. Older cards might only output DVI no matter what.

If it's treated as DVI, you will get a blurry image beyond 165 MHz pixel clock because that's where it switches to a dual-link signal, but HDMI is single-link, so you lose half the pixels. Renaming the patcher to whatever-patcher-single.exe will make it always send a single-link signal, but that might not work with older cards.

If it's treated as HDMI, AMD/ATI cards will output an HDMI signal over DVI ports, but only up to 165 MHz pixel clock unless the driver is patched. With the HD 7990, you shouldn't need the patch if you plug the monitor directly into an HDMI port.

NVIDIA cards will output an HDMI signal over DVI without the 165 MHz limitation, but keep in mind older cards might not have the hardware to handle higher pixel clocks because old versions of HDMI only supported up to 165 MHz. The one GTX 295 probably has a better TMDS transmitter that is capable of higher pixel clocks.

[apav]

(02-25-2020 05:40 PM)ToastyX Wrote:  

(02-24-2020 10:06 PM)apav Wrote:  I recently upgraded from a GTX 770 to a GTX 1080 Ti. My monitor has been running at 110hz using the LCD Reduced setting in CRU for years, without any artifacting or other issues. After swapping the cards out I uninstalled my Nvidia drivers in safe mode, reinstalled them normally, used the Pixel Clock Patcher, added my resolutions to CRU, then restarted. However when I boot up Windows I get these weird artifacts which never happened with my GTX 770.

https://streamable.com/5vxcf
https://streamable.com/sk3rb

Sometimes it happens immediately when it goes to the lock screen, sometimes it happens when I press anything to get to the password entry screen. After that it doesn't seem to do it while browsing the desktop, watching videos or playing games. Again this is at 110hz, I seem to be having no issues anywhere at 60hz. Did I miss something or enter something incorrectly on CRU? Nothing has changed with my monitor, so I'm guessing it is either a software issue or the GPU itself (though not sure what exactly would be the issue if its the GPU).

That looks like the memory clock is changing when it shouldn't. Try raising the vertical blanking/total and see if that makes it stop.

Just to confirm so I don't accidentally mess something up, you mean keep increasing this 6 number by one until it stops? Do I touch that 144 number at all?

I just don't understand why the same settings in CRU on the same monitor are giving me a problem now when they didn't before. Can the same settings behave differently based on what GPU you're using? Thanks.

[ToastyX]

(02-25-2020 06:15 PM)apav Wrote:  Just to confirm so I don't accidentally mess something up, you mean keep increasing this 6 number by one until it stops? Do I touch that 144 number at all?

I just don't understand why the same settings in CRU on the same monitor are giving me a problem now when they didn't before. Can the same settings behave differently based on what GPU you're using? Thanks.

Yes, increase the vertical only. The memory clock is not supposed to change if the vertical blanking is too low, but it looks like it's changing anyway for some reason, most likely a driver bug. This is not an issue with the monitor.

[apav]

(02-25-2020 07:20 PM)ToastyX Wrote:  

(02-25-2020 06:15 PM)apav Wrote:  Just to confirm so I don't accidentally mess something up, you mean keep increasing this 6 number by one until it stops? Do I touch that 144 number at all?

I just don't understand why the same settings in CRU on the same monitor are giving me a problem now when they didn't before. Can the same settings behave differently based on what GPU you're using? Thanks.

Yes, increase the vertical only. The memory clock is not supposed to change if the vertical blanking is too low, but it looks like it's changing anyway for some reason, most likely a driver bug. This is not an issue with the monitor.

Before I saw this reply, I tried deleting my resolutions in CRU and recreating them Nvidia Control Panel using the CVT reduced blank setting. I don't seem to be having the artficating issue with this custom resolution, but the pixel clock is about 23Mhz higher than the one I had in CRU.

Since you mentioned memory clock, I forgot to mention that when I use CRU's 110hz custom resolution my GPU stays at much higher clockspeeds when idling. This resolution I made in Nvidia Control Panel seems to keep the GPU at idle clocks, though with this resolution every time after rebooting my computer it flickers between the two clocks for a bit before finally settling on the idle one. Both CRU and Nvidia Control Panel's 96hz custom resolution that I made doesn't seem to do this.

Do you recommend I try to get this working correctly in CRU for the lower pixel clock, or should I leave it with Nvidia Control Panel as is?

[ToastyX]

(02-25-2020 08:05 PM)apav Wrote:  Before I saw this reply, I tried deleting my resolutions in CRU and recreating them Nvidia Control Panel using the CVT reduced blank setting. I don't seem to be having the artficating issue with this custom resolution, but the pixel clock is about 23Mhz higher than the one I had in CRU.

Since you mentioned memory clock, I forgot to mention that when I use CRU's 110hz custom resolution my GPU stays at much higher clockspeeds when idling. This resolution I made in Nvidia Control Panel seems to keep the GPU at idle clocks, though with this resolution every time after rebooting my computer it flickers between the two clocks for a bit before finally settling on the idle one. Both CRU and Nvidia Control Panel's 96hz custom resolution that I made doesn't seem to do this.

Do you recommend I try to get this working correctly in CRU for the lower pixel clock, or should I leave it with Nvidia Control Panel as is?

"LCD standard" will give you the same timing parameters as CVT reduced blank. That should allow the memory clock to change without artifacts. You can try for a lower pixel clock, but it's not necessary if you're not having any problems.