Try older NV driver with PowerStrip (v. 2.78 for GF2 and older), as Custom resolution edit tool.
Make sure files from old driver aren't screwing you.
Lastly : If you have one, try using VGA to HDMI adapter. Maybe it's some weird limitation on DVI/edid spoofer.

WARNING : If you mess up, and program saves non-working/non-booting setting, you will be forced to uninstall pstrip in safemode and start from scratch.

agent_x007 wrote on 2022-10-28, 05:30:

Try older NV driver with PowerStrip (v. 2.78 for GF2 and older), as Custom resolution edit tool. Make sure files from old drive […]
Show full quote

Try older NV driver with PowerStrip (v. 2.78 for GF2 and older), as Custom resolution edit tool.
Make sure files from old driver aren't screwing you.
Lastly : If you have one, try using VGA to HDMI adapter. Maybe it's some weird limitation on DVI/edid spoofer.

WARNING : If you mess up, and program saves non-working/non-booting setting, you will be forced to uninstall pstrip in safemode and start from scratch.

Quadro FX 3000 is similar to GeForce FX5900. It doesn't have VGA, only two DVI ports. For now I'm using 77.72, which feels a bit more stable than 8x.xx versions. Don't know which driver is the best for NV3x GPUs.

I don't think the EDID spoofer (dongle) is the problem, as I'm using my monitor's original EDID as a base, and the problem exists with or without the spoofer. Everything works fine on XP.

On the other hand, I'm not sure if DirectX version also matters, as I haven't installed DX9 yet. My WinME is still on DX7. Guess I'll give that a try when I have time.

So far the EDID spoofing stuff does work to some extent as I'm already getting some promising results here. I'm considering exploring EDID modding a bit more.

By the way: I'm using AW EDID Editor to do the modifications. When I'm reading the original EDID dump from my monitor, I got a warning about CEA size exceeded and in the CEA section I found a detailed timing block marked red containing a bunch of DTDs (5 exactly), that appear to be totally garbled, so I deleted them. While I'm not seeing any apparent issue at the moment, I'm wondering what those garbled data really are.

Your Quadro has DVI-I (not DVI-D), you can connect it to VGA => HDMI adapter using DVI to VGA passive one.
In general, for FX series it's worth checking out drivers from 45.23 up to 66.42 (there are topics about it here, exact starting driver depends on model).

agent_x007 wrote on 2022-10-28, 06:52:

Your Quadro has DVI-I (not DVI-D), you can connect DVI-I to VGA passive adapter.
In general, for FX series it's worth checking out drivers from 45.23 up to 66.42 (there are topics about it here, exact starting driver depends on model).

Thanks for the advice. I'll take a look at these driver versions.

A bump on this. I just did some further experiments with EDID and I've some good news and bad news.

Good news: I succeeded in forcing the output to 1920x1080 at 70Hz by modifying the preferred output part. In this state, I can correctly set 1920x1080 at 70Hz with the Win9x driver (77.72) as well.

Bad news: This doesn't survive a reboot. With this resolution set I could no longer boot it to Windows anymore... Also, it seems the changes to EDID kinda messed something else up and now even WinXP can't boot anymore with the monitor showing garbage. Even using VGA mode won't help. Also, even with the monitor actually outputting 1920x1080 I still cannot set it in WinNT 3.51 using the VBEMP driver, though other resolutions are still working, fortunately.

After some further investigation, it seems AW EDID Editor doesn't support everything and the garbled stuffs do have meanings, just that AW EDID Editor can't recognize them properly. Guess I need to start over with a different EDID editor that hopefully supports all extensions (so it will recognize everything inside the EDID without mistake).

Also, it seems using 1920x1080 as preferred output has led to some unexpected issues, namely, I noticed some graphical glitches in certain places, like Windows 9x boot logo, that could also end up affecting games as well. Seeing this made me remember similar issues I had in the past with a Radeon HD 3850 AGP and 4650 AGP with another monitor. Maybe I should consider making 1024x768 or 1280x1024 as preferred output as I don't remember seeing issues when using these resolutions as default output. Will experiment a bit more as it's starting to get interesting.

LSS10999 wrote on 2022-10-28, 13:08:

A bump on this. I just did some further experiments with EDID and I've some good news and bad news. […]
Show full quote

A bump on this. I just did some further experiments with EDID and I've some good news and bad news.

Good news: I succeeded in forcing the output to 1920x1080 at 70Hz by modifying the preferred output part. In this state, I can correctly set 1920x1080 at 70Hz with the Win9x driver (77.72) as well.

Bad news: This doesn't survive a reboot. With this resolution set I could no longer boot it to Windows anymore... Also, it seems the changes to EDID kinda messed something else up and now even WinXP can't boot anymore with the monitor showing garbage. Even using VGA mode won't help. Also, even with the monitor actually outputting 1920x1080 I still cannot set it in WinNT 3.51 using the VBEMP driver, though other resolutions are still working, fortunately.

After some further investigation, it seems AW EDID Editor doesn't support everything and the garbled stuffs do have meanings, just that AW EDID Editor can't recognize them properly. Guess I need to start over with a different EDID editor that hopefully supports all extensions (so it will recognize everything inside the EDID without mistake).

Also, it seems using 1920x1080 as preferred output has led to some unexpected issues, namely, I noticed some graphical glitches in certain places, like Windows 9x boot logo, that could also end up affecting games as well. Seeing this made me remember similar issues I had in the past with a Radeon HD 3850 AGP and 4650 AGP with another monitor. Maybe I should consider making 1024x768 or 1280x1024 as preferred output as I don't remember seeing issues when using these resolutions as default output. Will experiment a bit more as it's starting to get interesting.

Can you share the exact timings you are using and a copy of both the original EDID and your modified one ?

EDIT : Also are you making sure that the preferred mode you are setting has a pixel clock lower than 165MHz ?
EDIT2 : Reformulated for clarity

EDIT3 :
I use this EDID editor : https://ez.analog.com/video/w/documents/750/a … tiv-edid-editor

The downloads are temporarily offline due to a code review, but they just removed the links from the main page and the files are still hosted and accessible .

https://ez.analog.com/cfs-file/__key/communit … 0_1p01p0276.zip
https://ez.analog.com/cfs-file/__key/communit … t_2D00_0p97.zip
https://ez.analog.com/cfs-file/__key/communit … D_5F00_0p41.zip

WARNING : I don't know why they removed the links pending a code review, but if you want to be extra cautious, I can tell you that the version I use and downloaded in 2017 has the same checksum as the one currently named 1731.setup_5F00_EEdit_2D00_0p97.zip

1                MD5                             SHA-1
2-------------------------------------------------------------------------
3faca1e0d20a1883439c96a24bc3e9a20 31bcb3283d1e8b9cfe2ab9973db702ab7e52ef93 1731.setup_eedit-0p97.zip
4
5                MD5                             SHA-1
6-------------------------------------------------------------------------
7faca1e0d20a1883439c96a24bc3e9a20 31bcb3283d1e8b9cfe2ab9973db702ab7e52ef93 setup_eedit-0p97.zip

They are also available from Archive.org .

https://web.archive.org/web/20221028135952/ht … 0_1p01p0276.zip
https://web.archive.org/web/20220126195317/ht … t_2D00_0p97.zip
https://web.archive.org/web/20220126195317/ht … D_5F00_0p41.zip

darry wrote on 2022-10-28, 13:32:

Can you share the exact timings you are using and a copy of both the original EDID and your modified one ? […]
Show full quote

LSS10999 wrote on 2022-10-28, 13:08:

A bump on this. I just did some further experiments with EDID and I've some good news and bad news. […]
Show full quote

A bump on this. I just did some further experiments with EDID and I've some good news and bad news.

Good news: I succeeded in forcing the output to 1920x1080 at 70Hz by modifying the preferred output part. In this state, I can correctly set 1920x1080 at 70Hz with the Win9x driver (77.72) as well.

Bad news: This doesn't survive a reboot. With this resolution set I could no longer boot it to Windows anymore... Also, it seems the changes to EDID kinda messed something else up and now even WinXP can't boot anymore with the monitor showing garbage. Even using VGA mode won't help. Also, even with the monitor actually outputting 1920x1080 I still cannot set it in WinNT 3.51 using the VBEMP driver, though other resolutions are still working, fortunately.

After some further investigation, it seems AW EDID Editor doesn't support everything and the garbled stuffs do have meanings, just that AW EDID Editor can't recognize them properly. Guess I need to start over with a different EDID editor that hopefully supports all extensions (so it will recognize everything inside the EDID without mistake).

Also, it seems using 1920x1080 as preferred output has led to some unexpected issues, namely, I noticed some graphical glitches in certain places, like Windows 9x boot logo, that could also end up affecting games as well. Seeing this made me remember similar issues I had in the past with a Radeon HD 3850 AGP and 4650 AGP with another monitor. Maybe I should consider making 1024x768 or 1280x1024 as preferred output as I don't remember seeing issues when using these resolutions as default output. Will experiment a bit more as it's starting to get interesting.

Can you share the exact timings you are using and a copy of both the original EDID and your modified one ?

EDIT : Also are you making sure that the preferred mode you are setting has a pixel clock lower than 165MHz ?
EDIT2 : Reformulated for clarity

EDIT3 :
I use this EDID editor : https://ez.analog.com/video/w/documents/750/a … tiv-edid-editor

The downloads are temporarily offline due to a code review, but they just removed the links from the main page and the files are still hosted and accessible .

https://ez.analog.com/cfs-file/__key/communit … 0_1p01p0276.zip
https://ez.analog.com/cfs-file/__key/communit … t_2D00_0p97.zip
https://ez.analog.com/cfs-file/__key/communit … D_5F00_0p41.zip

WARNING : I don't know why they removed the links pending a code review, but if you want to be extra cautious, I can tell you that the version I use and downloaded in 2017 has the same checksum as the one currently named 1731.setup_5F00_EEdit_2D00_0p97.zip

Copy code to clipboard

1                MD5                             SHA-1
2-------------------------------------------------------------------------
3faca1e0d20a1883439c96a24bc3e9a20 31bcb3283d1e8b9cfe2ab9973db702ab7e52ef93 1731.setup_eedit-0p97.zip
4
5                MD5                             SHA-1
6-------------------------------------------------------------------------
7faca1e0d20a1883439c96a24bc3e9a20 31bcb3283d1e8b9cfe2ab9973db702ab7e52ef93 setup_eedit-0p97.zip

They are also available from Archive.org .

https://web.archive.org/web/20221028135952/ht … 0_1p01p0276.zip
https://web.archive.org/web/20220126195317/ht … t_2D00_0p97.zip
https://web.archive.org/web/20220126195317/ht … D_5F00_0p41.zip

Yeah, it seems to be the better tool to use. Thanks for the link. Will try tweaking the EDID file again using it.

The pixel clock for 1920x1080 at 70Hz would be at around 162.5MHz (with CVT reduced blank), which is very close to single-link DVI's 165MHz limit.

For now I can provide the original EDID that I dumped from the monitor in case you're interested. This monitor uses HDMI 2.0 and it defaults at 4K/60Hz, though 120Hz can be used for 2560x1440 or 1920x1080 if supported by the video card.

Off-topic: I'm not sure about DisplayPort, though. This monitor is DP1.4 capable and supports 4K at 120Hz/144Hz, but I cannot use any EDID override feature (to retain the presence of the monitor when I'm switching to another input), as the EDID file exported appears to only contain information up to DP1.2 (nearly equal to HDMI 2.0).

agent_x007 wrote on 2022-10-28, 06:52:

In general, for FX series it's worth checking out drivers from 45.23 up to 66.42 (there are topics about it here, exact starting driver depends on model).

Always use 45.23 with FX series cards, not anything newer. There are compatibility problems with newer drivers - Thief 2, Need for Speed Porsche and EA Sports games run fine with 45.23. With later driver things start to break.

GeForce FX cards do have visual issues like illegible fonts in some games even with 45.23 but it's rare. For the absolute best compatibility GeForce 4 is needed (MX440, Ti).

An update. Guess I must have done something irreversible to the video card and I don't think I could fix it on my own.

It seems the graphical glitch that just appeared has nothing to do with the EDID dongle or anything. I still see them even after connecting the card directly to the monitor, without that dongle (so no forced resolutions or refresh rates according to the monitor OSD). The monitor itself is fine as my other systems are still working without issues, and the card was working fine until now (none of these issues I'm having now were present before).

- WinME can boot into safe mode or using default 16 color output, but there are now some color distortions that were not present previously. It can no longer boot to higher color modes anymore as Windows will freeze before displaying the UI.
- WinXP still cannot boot, with the card still outputting garbage like that time (VGA mode does not work). Safe Mode works fine, though.

Looks like I must have really really damaged the card's functionality with this experiment...

LSS10999 wrote on 2022-10-29, 05:04:

An update. Guess I must have done something irreversible to the video card and I don't think I could fix it on my own. […]
Show full quote

An update. Guess I must have done something irreversible to the video card and I don't think I could fix it on my own.

It seems the graphical glitch that just appeared has nothing to do with the EDID dongle or anything. I still see them even after connecting the card directly to the monitor, without that dongle (so no forced resolutions or refresh rates according to the monitor OSD). The monitor itself is fine as my other systems are still working without issues, and the card was working fine until now (none of these issues I'm having now were present before).

- WinME can boot into safe mode or using default 16 color output, but there are now some color distortions that were not present previously. It can no longer boot to higher color modes anymore as Windows will freeze before displaying the UI.
- WinXP still cannot boot, with the card still outputting garbage like that time (VGA mode does not work). Safe Mode works fine, though.

Looks like I must have really really damaged the card's functionality with this experiment...

If all you did was experiment with EDID, I doubt you did anything bad to the video card, IMHO. I did lots of weird stuff with EDID on an FX5900 and never managed to damage it.

Strangely enough, the symptoms you describe resemble those I saw with the only FX3000 I bought, which was like this from the start. It was sold as used but working, so I got a refund.

I don't know if FX3000 cards (the GPU, RAM or another component) are somehow susceptible to a common failure mode or if this is just a coincidence .

darry wrote on 2022-10-29, 08:11:

If all you did was experiment with EDID, I doubt you did anything the video card, IMHO. I did lits of weird stuff with EDID on a […]
Show full quote

LSS10999 wrote on 2022-10-29, 05:04:

An update. Guess I must have done something irreversible to the video card and I don't think I could fix it on my own. […]
Show full quote

An update. Guess I must have done something irreversible to the video card and I don't think I could fix it on my own.

It seems the graphical glitch that just appeared has nothing to do with the EDID dongle or anything. I still see them even after connecting the card directly to the monitor, without that dongle (so no forced resolutions or refresh rates according to the monitor OSD). The monitor itself is fine as my other systems are still working without issues, and the card was working fine until now (none of these issues I'm having now were present before).

- WinME can boot into safe mode or using default 16 color output, but there are now some color distortions that were not present previously. It can no longer boot to higher color modes anymore as Windows will freeze before displaying the UI.
- WinXP still cannot boot, with the card still outputting garbage like that time (VGA mode does not work). Safe Mode works fine, though.

Looks like I must have really really damaged the card's functionality with this experiment...

If all you did was experiment with EDID, I doubt you did anything the video card, IMHO. I did lits of weird stuff with EDID on an FX5900 and never managed to damage it.

Strangely enough, the symptoms you describe resemble those I saw with the only FX3000 I bought, which was like this from the start. It was sold as used but working, so I got a refund.

I don't know if FX3000 cards (the GPU, RAM or another component) are somehow susceptible to a common failure mode or if this is just a coincidence .

No idea. So far it seems the video card's VGA resolutions are kinda damaged. It can no longer correctly output 640x480 resolutions anymore (regardless of color depth), and ironically, the color distortion with 16 colors (especially around the mouse pointer) kind of resembles the one I noticed on Maxwell and later nVidia video cards.

Still, I'm not sure if editing EDID could kind of reprogram the video card somehow as the "damage" appears irreversible. I continued experimenting with EDID and was able to get some interesting results afterwards.

- The preferred timing matters most. It will always be considered first if applicable. If not, the video card will choose an alternative from the EDID. There are several places where resolutions can be defined.
- You can set the preferred timing to anything you want, and the video card will output using that timing as long as it looks good to the video card, though you may not get output or OUT OF RANGE if configured incorrectly.
- It seems some video cards behave oddly with particular resolutions such as 1920x1080. In my case, forcing 1920x1080 caused one of the system to get black borders around the output. Using 1680x1050, however, works without black borders.
- I did manage to get my FX 3000 to output at 120Hz with a 1280x720 resolution (should also do with 1280x800, though I haven't tested it yet). This should work with any resolution as long as the pixel clock is less than 165MHz. However, it seems using 1280x720 caused resolutions above 800x600 to be unavailable. 1024x768 should in theory be available if using 1280x800, though.
- VESA drivers such as VBEMP will usually not touch your current timing. As such you can use this EDID hack to make some OSes that rely on VESA for video operate at 120Hz or even more, namely ArcaOS and Haiku.
- Generally, my monitor appears to use the preferred timing for any resolution below that, so all lower resolutions (640x480, 800x600, etc.) will all appear as the preferred resolution and refresh rate according to the OSD.

Should say that EEditGold is indeed the right tool for the job. It correctly recognized every single aspect of the EDID file after importing. However, it doesn't have a built-in calculator for preferred timings so I have to use online tools to get the right values.

LSS10999 wrote on 2022-10-29, 10:21:

No idea. So far it seems the video card's VGA resolutions are kinda damaged. It can no longer correctly output 640x480 resolutio […]
Show full quote

darry wrote on 2022-10-29, 08:11:

If all you did was experiment with EDID, I doubt you did anything the video card, IMHO. I did lits of weird stuff with EDID on a […]
Show full quote

LSS10999 wrote on 2022-10-29, 05:04:

An update. Guess I must have done something irreversible to the video card and I don't think I could fix it on my own. […]
Show full quote

An update. Guess I must have done something irreversible to the video card and I don't think I could fix it on my own.

It seems the graphical glitch that just appeared has nothing to do with the EDID dongle or anything. I still see them even after connecting the card directly to the monitor, without that dongle (so no forced resolutions or refresh rates according to the monitor OSD). The monitor itself is fine as my other systems are still working without issues, and the card was working fine until now (none of these issues I'm having now were present before).

- WinME can boot into safe mode or using default 16 color output, but there are now some color distortions that were not present previously. It can no longer boot to higher color modes anymore as Windows will freeze before displaying the UI.
- WinXP still cannot boot, with the card still outputting garbage like that time (VGA mode does not work). Safe Mode works fine, though.

Looks like I must have really really damaged the card's functionality with this experiment...

If all you did was experiment with EDID, I doubt you did anything the video card, IMHO. I did lits of weird stuff with EDID on an FX5900 and never managed to damage it.

Strangely enough, the symptoms you describe resemble those I saw with the only FX3000 I bought, which was like this from the start. It was sold as used but working, so I got a refund.

I don't know if FX3000 cards (the GPU, RAM or another component) are somehow susceptible to a common failure mode or if this is just a coincidence .

No idea. So far it seems the video card's VGA resolutions are kinda damaged. It can no longer correctly output 640x480 resolutions anymore (regardless of color depth), and ironically, the color distortion with 16 colors (especially around the mouse pointer) kind of resembles the one I noticed on Maxwell and later nVidia video cards.

Still, I'm not sure if editing EDID could kind of reprogram the video card somehow as the "damage" appears irreversible. I continued experimenting with EDID and was able to get some interesting results afterwards.

- The preferred timing matters most. It will always be considered first if applicable. If not, the video card will choose an alternative from the EDID. There are several places where resolutions can be defined.
- You can set the preferred timing to anything you want, and the video card will output using that timing as long as it looks good to the video card, though you may not get output or OUT OF RANGE if configured incorrectly.
- It seems some video cards behave oddly with particular resolutions such as 1920x1080. In my case, forcing 1920x1080 caused one of the system to get black borders around the output. Using 1680x1050, however, works without black borders.
- I did manage to get my FX 3000 to output at 120Hz with a 1280x720 resolution (should also do with 1280x800, though I haven't tested it yet). This should work with any resolution as long as the pixel clock is less than 165MHz. However, it seems using 1280x720 caused resolutions above 800x600 to be unavailable. 1024x768 should in theory be available if using 1280x800, though.
- VESA drivers such as VBEMP will usually not touch your current timing. As such you can use this EDID hack to make some OSes that rely on VESA for video operate at 120Hz or even more, namely ArcaOS and Haiku.
- Generally, my monitor appears to use the preferred timing for any resolution below that, so all lower resolutions (640x480, 800x600, etc.) will all appear as the preferred resolution and refresh rate according to the OSD.

Should say that EEditGold is indeed the right tool for the job. It correctly recognized every single aspect of the EDID file after importing. However, it doesn't have a built-in calculator for preferred timings so I have to use online tools to get the right values.

Did you ever get to test the 45.23 driver? Because my hope was that those older drivers somehow enable dual link dvi, which is broken in all the 77.72 and later versions that I can test with my Quadro FX 3450 and 7900gs/gto under 98.

Jonsmith0815 wrote on 2024-04-30, 22:02:

Did you ever get to test the 45.23 driver? Because my hope was that those older drivers somehow enable dual link dvi, which is broken in all the 77.72 and later versions that I can test with my Quadro FX 3450 and 7900gs/gto under 98.

Nope... I'm afraid the card was likely dead shortly after that topic. I didn't get to try older drivers as I'm not sure which was the most optimal according to the release notes.

Originally I thought I might be able to address the issues through a BIOS flash, as the card apparently had a BIOS update from HP. But no. Instead of improvements, it made the card even worse to the point no VGA resolutions ever work (my Win10-based PE even fell back to a greyscale resolution).

Then I tried flashing back the old image and it ended up totally borked. The card could still boot, but no video outputs ever work anymore. Perhaps this card be another victim to the well-known "bumpgate" that were relevant to those generations of nVidia cards?

On the other hand, my problem had little to do with dual-link DVI at present. Adapters that could actively convert dual-link DVI signals simply do not exist. If I ever want to make use of such old video cards for high resolutions and refresh rates, I have to use a lesser monitor that does have dual-link DVI connetion, like Asus VG248QE(Z).

Finally, my monitor (XG27UQR) simply isn't fit for such old nVidia video cards. It has a default timing for 4K and 1440p, but not 1080p. As such, nVidia cards incapable of outputting 4K or 1440p will be forced down to a very low default like 1024x768 and can't really drive any other timings unless the EDID is hacked to give something better and valid.

Ok, sorry to hear, that the card is still broken. If it wasn't bumpgate, I wonder if a new bios would help. Maybe use an external programmer and solder the bios chip if it has no socket. I dont know what bios chip the card uses.
So maybe the older driver would be a solution, although I am sceptical.
I have a dual link dvi benq xl2411, I believe it also supports 144hz only at 1080p, not at lower resolutions.