Esoteric Graphics Questions/Findings (another T221 thread)

[jeremywork]

SwitchResX 3.8.7 on Tiger (10.4.11) I presume?

Yeah, and 4.3.6 on the 2008 (10.6.8)

That's curious. Does it not work with the TH2G-SE at all, or not at >360 MHz?

That's a good point. I should remove the override before I plug it in next time in case it's failing to send 452.24 MHz.

In that case, I'd list only prototypes, close siblings and offshoots (and Toshiba's TN version), to avoid cluttering up the page.


This would make sense if the 3280×2048 and 4096×2560 (16:10) medical displays were also included.

This tablet might also be worth mentioning because of its display (20" 3840×2560; 3:2; 230 ppi) and HDMI 2.0 input (2015 model only). On the other hand, it’s quite recent (introduced in 2013) so may not fit the bill. For me, it’s about ultra-high-res displays that were way ahead of their time.

Maybe a combination of both options. Monet, Roentgen, the T210, and CDU2510A appear to exist in the same 'storyboard' so to speak, but the uncommonness of any 'Quad' derived XGA resolutions kind of merits mentioning all 2048 x 1536 and greater resolution support, at least for displays predating DisplayPort. The modern displays which are clearly of the DP age but use multiple DVI links anyways are kind of in a weird zone.

A 22" IBM display with a resolution of 3548×2661 (4:3) is mentioned in a late-2000 issue of German computer magazine c’t (c’t 24/2000, page 62). It's behind a paywall, but I can make it available. I couldn't find anything about this display, so I guess they may have been talking about Bertha but got the resolution wrong. Or maybe......

I'd appreciate taking a look. The odd numbered vertical resolution makes me think this is a total image signal including the intervals.

Spoiler: Ramblings

Shooting completely from the hip, supposing IBM was using the same approach as is evident in Bertha's EDIDs:
Using GTF to determine the vertical intervals and custom definitions for horizontals, considering a target somewhere in the 40-60Hz range for four stripes would give 2661 at

around 38.0 Hz for 2604 (perfect 4:3)
around 41.0 Hz for 2600
around 42.5 Hz for 2598 (perfect 4:3)
around 54.5 Hz for 2580 (perfect 4:3)
around 68.0 Hz for 2560

But there's the issue of 3548. If it's a four stripe configuration it would require an odd numbered horizontal total. If it's a four tile it could use an even horizontal, but then would require inequal vertical totals for 2661. Assuming the former:

((3548/4)-(8+32+23))*4 = 3296 (824 x 2472 @ 4:3; too small for vertical GTF unless 128.5 Hz; 297 MHz)
((3548/4)-(8+32+15))*4 = 3328 (832 x 2496 @ 4:3; too small for vertical GTF unless 112 Hz; 262 MHz)
((3548/4)-(8+32+7))*4 = 3360 (840 x 2520 @ 4:3; too small for vertical GTF unless 95.5 Hz; 225 MHz)
The horizontal totals are bordering on implausible, so I'm not sure this is at all close.

Doesn't seem like this approach works unless I'm off by a factor of two for some reason. Those total timings are plausible for single link stripes if cut in half...

On Bertha's 'custom' timing modes, the vertical intervals aren't straight GTF, only similar to GTF totals. So I could be going wrong there...

Could also just be an LCD die cut at 3548 x 2661

 

[jeremywork]

I've added an 'Overview and history' section above the existing content and in the process discovered a slide deck which also lists another couple interesting ones: NEC 1600 x 1200 (11.3", 177ppi, 1998 and 9.4", 211ppi, 1999) as well as a Toshiba 3200 x 2400 (20.0", 200ppi, 1998)

https://www.talisman.org/~erlkonig/misc/flatpanels.pdf

 

[Amethyst1]

[...] the uncommonness of any 'Quad' derived XGA resolutions kind of merits mentioning all 2048 x 1536 and greater resolution support, at least for displays predating DisplayPort.

I'll see what I can hunt down in terms of ≥2048×1536 LCDs then. Good point regarding the appearance of DisplayPort, i.e. basically 2008 or older.

I think I'll just add the list to this post right here, editing it as I find more (announcement dates):

---

1920×1920:
EIZO FlexScan EV2730Q (November 2014) [DP+DVI]

---

2048×1536:
Barco Coronis 3MP (?) [DP+DVI]
Barco Nio Color 3MP (?) [DP+DVI]
EIZO RadiForce G31 (July 2003)
EIZO RadiForce R31 (June 2005)
EIZO RadiForce G33 (June 2005)
EIZO RadiForce GS310 (March 2006)
EIZO RadiForce GS320 (November 2006)
EIZO RadiForce GX320 (November 2006)
EIZO RadiForce RX320 (November 2008)
EIZO RadiForce GX340 (September 2011) [DP+DVI]
EIZO RadiForce RX340 (September 2011) [DP+DVI]
EIZO RadiForce RX350 (November 2015) [DP+DVI]
EIZO RadiForce RX360 (October 2018) [DP+DVI]
EIZO RadiForce RX370 (June 2021) [DP+DVI]
NDS Dome E3 (?)
NDS Dome E3cHb (?) [DP+DVI]
NDS Dome S3 (?) [DP+DVI]
NDS Dome S3c (?) [DP+DVI]

---

2048×2048:
EIZO Raptor SQ2801 (March 2008)
EIZO Raptor SQ2802 (February 2012)
EIZO Raptor SQ2804 (November 2012)
EIZO Raptor SQ2814 (January 2013) [DP+DVI]
EIZO Raptor SQ2825 (September 2016) [DP+DVI]
EIZO Raptor SQ2826 (February 2023) [DP+DVI]

---

2560×2048:
Barco Nio 5MP (?) [DP+DVI]
Barco Coronis 5MP (?) [DP+DVI]
EIZO RadiForce G51 (July 2003)
EIZO RadiForce GS510 (June 2006)
EIZO RadiForce GS520 (November 2007)
EIZO RadiForce GS521 (September 2009) [DP+DVI]
EIZO RadiForce GX530 (November 2010) [DP+DVI]
EIZO RadiForce GX540 (November 2012) [DP+DVI]
EIZO RadiForce GX550 (November 2016) [DP+DVI]
EIZO RadiForce RX560 (August 2017) [DP+DVI]
EIZO RadiForce GX560 (June 2018) [DP+DVI]
NDS Dome E5 (?)

---

2800×2100:
Barco Nio Color 5MP (?) [DP+DVI]

---

3280×2048:
Barco Coronis Fusion 6MP (?) [DP]
Barco Coronis Fusion 6MP DL (?) [DP+DVI]
EIZO RadiForce RX650 (November 2013) [DP+DVI]
EIZO RadiForce RX650-AR (February 2015) [DP+DVI]
EIZO RadiForce RX660 (November 2016) [DP+DVI]
NDS Dome S6c (July 2014)

---

4096×2560:
EIZO RadiForce GX1030 (July 2010)
NDS Dome S10 (November 2010)

---

I've added an 'Overview and history' section above the existing content and in the process discovered a slide deck which also lists another couple interesting ones: NEC 1600 x 1200 (11.3", 177ppi, 1998 and 9.4", 211ppi, 1999) as well as a Toshiba 3200 x 2400 (20.0", 200ppi, 1998)

Awesome, I wasn't aware of these NEC and Toshiba displays. The 16.1" 1280×1024 (1994) IBM display is called "Halo" in c't 9/1996, page 163.

 

[jeremywork]

I'll see what I can hunt down in terms of ≥2048×1536 LCDs then. Good point regarding the appearance of DisplayPort, i.e. basically 2008 or older.

If you can find anything about that Toshiba QUXGA, I've only found a press release and a forum post about it. I guess it ended up 20.8" / 192ppi

Telecompaper

Telecompaper - Telecom news and research

www.telecompaper.com

Toshiba's new 20.8" 3200x2400 LCD display

OK, not sure if anyone else noticed, but both IBM and Toshiba announced some new kick-ass displays at CeBIT this year.Toshiba's display: "It supports up to 3,200 by 2,400 pixel QUXGA (Quad Ultra Extended Graphics Array) resolution, has a 0.132-millimeter dot pitch or 192 pixels per inch, can...

arstechnica.com

I think I'll just add the list to this post right here, editing it as I find more:

---

2048×1536:
EIZO RadiForce G31 (July 2003)

---

2560×2048:
EIZO RadiForce G51 (July 2003)

Awesome! I think I'll do a section with a straight list of these, since they're tangentially related but probably don't need commentary as much as a spec chart. Shouldn't detract much since the page is already kind of cluttered with data

Awesome, I wasn't aware of these NEC and Toshiba displays. The 16.1" 1280×1024 (1994) IBM display is called "Halo" in c't 9/1996, page 163.

Thanks, I added that to the prototypes section.

 

[Amethyst1]

Awesome! I think I'll do a section with a straight list of these, since they're tangentially related but probably don't need commentary as much as a spec chart. Shouldn't detract much since the page is already kind of cluttered with data

First slew of displays added. Inputs are DVI-SL or DVI-DL only unless stated otherwise.

If you can find anything about that Toshiba QUXGA, I've only found a press release and a forum post about it. I guess it ended up 20.8" / 192ppi

https://www.global.toshiba/ww/news/corporate/2001/03/pr1502.html
https://www.hartware.de/2001/03/17/nochmals-lcds/
https://www.zdnet.de/2055913/toshiba-kuendigt-20-zoll-lcd-an/

---

Spoiler: Some more noteworthy LCDs released before Bertha

• SGI 1600SW (September 1998): first 1600×1024 & first widescreen & only OpenLDI "desktop" (17.3") LCD released
• Apple Cinema Display (August 1999): first 22" & first 1600×1024 IPS LCD released; (possibly) first DVI LCD announced
• Dell (1999): 15" 1400×1050 LCD available in Inspiron 7500 (c't 21/1999, page 43; PC Magazine Vol. 18 No. 21, December 1st, 1999, page 46)
• Toshiba (1999): 15" 1600×1200 LTPS LCD shown off (c't 23/1999, page 28)
• Dell (August 2000): 15" 1600×1200 LCD available in Inspiron 5000
• Samsung 240T (September 2000): first 24" & first 1920×1200 LCD released
• Eizo L771 (November 2000): first 1600×1200 "desktop" (19.6") LCD released

 

[jeremywork]

First slew of displays added. Inputs are DVI-SL or DVI-DL only unless stated otherwise.

Brilliant work!

I've started an 'Other relevant displays' section and I'll work more on this shortly.

http://stormdecay.com:1984/displays_ibm_bertha.html#related

 

[Amethyst1]

Brilliant work!

EIZO have also released a couple of 2048×2048 1:1 displays so I've now added these.

---

Addendum:

The Sharp Zaurus MI-EI1 (1999) has a 4” 640×480 (200 ppi) LCD, but it’s a PDA, not a full-blown laptop.

The service manual for the Dell Inspiron 4100 (December 2001) lists a 14.1” 1600×1200 (142 ppi) LCD option.

The Sharp Zaurus SL-C700 (November 2002) has a 3.7” 640×480 (216 ppi) LCD, but it’s a PDA, not a full-blown laptop.

The manual for the Dell Inspiron 8500 (July 2003) lists a 15.4” 1920×1200 (147 ppi) LCD option. [This is beaten by 15” QXGA though.]

The Sony Vaio PCG-U1/U3 (2002) have a 6.4” 1024×768 (200 ppi) LCD.

The Sony Vaio VGN-UX series (2006 — too new?) have a 4.5” 1024×600 (264 ppi) LCD. They’re stretching the definition of “laptop” though.

---

And what about 30” 2560×1600 LCDs that predate DisplayPort?

 

[jeremywork]

Made a dent today.

And what about 30” 2560×1600 LCDs that predate DisplayPort?

I added the original 30" Cinema Display - I think this is the first.
There was an HP and at least two Dells (with DVI-DL and with DP); the Samsung 305T, the later revamped Apple 30. Any others?

Addendum:

The Sharp Zaurus MI-EI1 (1999) has a 4” 640×480 (200 ppi) LCD, but it’s a PDA, not a full-blown laptop.

The service manual for the Dell Inspiron 4100 (December 2001) lists a 14.1” 1600×1200 (142 ppi) LCD option.

The Sharp Zaurus SL-C700 (November 2002) has a 3.7” 640×480 (216 ppi) LCD, but it’s a PDA, not a full-blown laptop.

The manual for the Dell Inspiron 8500 (July 2003) lists a 15.4” 1920×1200 (147 ppi) LCD option. [This is beaten by 15” QXGA though.]

The Sony Vaio PCG-U1/U3 (2002) have a 6.4” 1024×768 (200 ppi) LCD.

The Sony Vaio VGN-UX series (2006 — too new?) have a 4.5” 1024×600 (264 ppi) LCD. They’re stretching the definition of “laptop” though.

These are in there

• SGI 1600SW (September 1998): first 1600×1024 & first widescreen & only OpenLDI "desktop" (17.3") LCD released
• Apple Cinema Display (August 1999): first 22" & first 1600×1024 IPS LCD released; (possibly) first DVI LCD announced
• Dell (1999): 15" 1400×1050 LCD available in Inspiron 7500 (c't 21/1999, page 43; PC Magazine Vol. 18 No. 21, December 1st, 1999, page 46)
• Toshiba (1999): 15" 1600×1200 LTPS LCD shown off (c't 23/1999, page 28)
• Dell (August 2000): 15" 1600×1200 LCD available in Inspiron 5000
• Samsung 240T (September 2000): first 24" & first 1920×1200 LCD released
• Eizo L771 (November 2000): first 1600×1200 "desktop" (19.6") LCD released

These too

1920×1920:
EIZO FlexScan EV2730Q (November 2014) [DP+DVI]

---

2048×1536:
Barco Coronis 3MP (?) [DP+DVI]
Barco Nio Color 3MP (?) [DP+DVI]
EIZO RadiForce G31 (July 2003)
EIZO RadiForce R31 (June 2005)
EIZO RadiForce G33 (June 2005)
EIZO RadiForce GS310 (March 2006)
EIZO RadiForce GS320 (November 2006)
EIZO RadiForce GX320 (November 2006)
EIZO RadiForce RX320 (November 2008)
EIZO RadiForce GX340 (September 2011) [DP+DVI]
EIZO RadiForce RX340 (September 2011) [DP+DVI]
EIZO RadiForce RX350 (November 2015) [DP+DVI]
EIZO RadiForce RX360 (October 2018) [DP+DVI]
EIZO RadiForce RX370 (June 2021) [DP+DVI]
NDS Dome E3 (?)
NDS Dome E3cHb (?) [DP+DVI]
NDS Dome S3 (?) [DP+DVI]
NDS Dome S3c (?) [DP+DVI]

---

2048×2048:
EIZO Raptor SQ2801 (March 2008)
EIZO Raptor SQ2802 (February 2012)
EIZO Raptor SQ2804 (November 2012)
EIZO Raptor SQ2814 (January 2013) [DP+DVI]
EIZO Raptor SQ2825 (September 2016) [DP+DVI]
EIZO Raptor SQ2826 (February 2023) [DP+DVI]

---

2560×2048:
Barco Nio 5MP (?) [DP+DVI]
Barco Coronis 5MP (?) [DP+DVI]
EIZO RadiForce G51 (July 2003)
EIZO RadiForce GS510 (June 2006)
EIZO RadiForce GS520 (November 2007)
EIZO RadiForce GS521 (September 2009) [DP+DVI]
EIZO RadiForce GX530 (November 2010) [DP+DVI]
EIZO RadiForce GX540 (November 2012) [DP+DVI]
EIZO RadiForce GX550 (November 2016) [DP+DVI]
EIZO RadiForce RX560 (August 2017) [DP+DVI]
EIZO RadiForce GX560 (June 2018) [DP+DVI]
NDS Dome E5 (?)

---

2800×2100:
Barco Nio Color 5MP (?) [DP+DVI]

---

3280×2048:
Barco Coronis Fusion 6MP (?) [DP]
Barco Coronis Fusion 6MP DL (?) [DP+DVI]
EIZO RadiForce RX650 (November 2013) [DP+DVI]
EIZO RadiForce RX650-AR (February 2015) [DP+DVI]
EIZO RadiForce RX660 (November 2016) [DP+DVI]
NDS Dome S6c (July 2014)

---

4096×2560:
EIZO RadiForce GX1030 (July 2010)
NDS Dome S10 (November 2010)

I've been doing these in release order so far, but maybe I should only do a dated list of 'firsts' and then switch to a resolution-based order like you've done here. It might be more straightforward. I've marked off the Eizo G31, G51, R31, and G33 to keep track of what I've added.

I should probably also split another column for LCD process (DD-IPS, TN, etc) and maybe another for input channels/CLUT data, since that's cluttering up the notes section a bit.

The hyperlinks are going to need refinement but most of them are at least there.

 

[Amethyst1]

Any others?

Displayspecifications lists a couple, using a 2560×1600 filter.

 

[Amethyst1]

Displayspecifications lists a couple, using a 2560×1600 filter.

Just checked that list and it doesn't include the older ones.

I added the original 30" Cinema Display - I think this is the first.

So do I.

There was an HP and at least two Dells (with DVI-DL and with DP); the Samsung 305T, the later revamped Apple 30. Any others?

Apple 30" (revised), Dell 3007WFP, Dell 3007WFP-HC, HP L3065, LG W3000H, Samsung 305T.

 

[jeremywork]

Miscellaneous updates:

Re-testing with the Titanium PowerBook reveals 48, 32, 64; 2, 2, 2 are truly its tightest working intervals with the T221. Although it does produce a stable image at up to 165.00 MHz, I have noticed some shimmering green pixels begin to appear especially in dark places as the PowerBook warms up. It doesn't appear to matter what resolution or whether the blanking intervals are tight or not, as long as the pixel clock is near 165 MHz. Reducing to 162 MHz with the same tight intervals alleviates this noise entirely. Since I'm fond of that PowerBook I'll stop pushing it now.

Tangent: screensavers work fine on the TiBook at 2048 x 1536, but not at 2560 x 1600 or higher for some reason (except the machine name screensaver; that one works fine even at 3840 x 2400)

In other news I was able to find a couple DVI EDID override passthrough adapters which use 24C02C chips/256 bytes.


I have confirmed a basic test with a single-link display's EDID successfully masking the hardware display I was plugged into on a Rage 128. I wrote both the Left and Right tiles adapted for the T221 on Jack Doerner's page and tested the following:
2015 MBP (Iris Pro 5200, 10.14.6):

-> 2x Apple A1306 -> 2x EDID override (L and R) -> 2x LFH-60 adapters -> Bertha

-> Startech MST hub -> 2x Accell B087B-002B -> 2x EDID override (L and R) -> 2x LFH-60 adapters -> Bertha

In both cases the computer doesn't appear to detect any new display plugged in, not even a flash of the display or a hesitation in the cursor. Upon replugging one of the DVI cables between the DP-DVI adapter and the EDID override, there was some hesitation in the cursor, but no display was enabled.

These look to be EDIDs for both tiles of the Dell UP2414Q.

I then wrote the two EDID blocks from here and tried the same configurations but this produced the same results.

The only thing that did sort-of work was starting with the Startech -> 2x Accell -> 2x LFH-60 -> Bertha (which gives a mirrored 2x 1920 x 2400 @ 48 Hz) and then hotplugging one side or the other with the EDID Override dongles. At least both scalers still received a signal in those configurations, but the EDID is still being read from the T221 itself.


I'm not sure why the EDID override passthrough seems to defeat the passive split in the LFH-60 adapters. Looking closely it seems the two 960 x 2400 stripes contain opposite pixels, so it's not losing the second link...


If I hotplugged both sides, the display would simply stay black, and even unplugging the Startech wouldn't remove the detected Bertha display from the MBP's 'Arrange' tab; even clicking 'Detect Displays' wouldn't remove it. I'd have to hotplug back to direct connection(s) to the T221 then remove the Startech for the computer to update the active displays (or reboot.) This was also the case with A1306 adapters or Delock 62603s when hotplugged.

Catching up here a bit:

https://forums.macrumors.com/threads/early-2011-13-macbook-pro-intel-hd-graphics-3000-driving-two-or-three-external-monitors-thunderbolt-1-awesomeness.2309427/

The Dell UP2414Q doesn't use a .mtdd file, and an override should be present in Mac OS since 10.9 if I understand correctly.

Is there any reason these EDID override dongles behind active DisplayPort adapters would return EDIDs differently than a 'real' UP2414Q? It seems like the EDID of the T221 is read fine this way. The issue of the DL stripes being split wrong would be the next problem, but I'd expect if it were the only problem Mac OS would present a 3840 x 2160 @ 60 Hz display at least.

 

[Amethyst1]

The Dell UP2414Q doesn't use a .mtdd file, and an override should be present in Mac OS since 10.9 if I understand correctly.

This is a shot in the dark, but what about flashing the two tile EDIDs for a dual-link SST (e.g. 5120×2880) display that uses two discrete DisplayPort connections rather than MST?

Attached are the EDIDs for my HP Z27q which uses an .mtdd file. Each tile is 2560×2880@60Hz (483.25MHz) though so something will be overrun. (Ignore the fake EDID in the archive that only has a 5120×2880 mode. It comes from the .mtdd, not from the display itself.)

The Dell UP2715K can do 2560×2880@30Hz (238.25MHz) on each tile so its EDIDs may be better, but macOS Mojave and older doesn’t expose 30Hz if a 60Hz mode is present (multiple timings in an .mtdd file require Catalina AFAIK). I have attached its EDIDs too. You could try removing the 60Hz timing from them I suppose.

However, dual-link SST may not work on the Iris Pro 5200 in general (it doesn’t with the Dell UP2715K), and the HP Z27q in particular may not work on NVIDIA GPUs (I’ll have to test this — it works fine on AMD GPUs). Without hacks, NVIDIA GPUs can do dual-link SST on the Dell UP2715K only IIRC. They also have a very annoying cursor bug that made me switch to an AMD GPU.

 

[Amethyst1]

Assuming all else is identical this mode should work for you...

Finally got round to testing the DH2G-DP with the 9503-DG3. 345.2 MHz works fine.

Some other early >100 ppi LCDs that might be worth adding to your page:

• Iiyama ProLite 39a (1999): 15.4" 1280×1024 (106.44 ppi) TN
• NEC MultiSync LCD1525S (September (?) 1999): 15.4" 1280×1024 (106.44 ppi) TN
• Iiyama AU4831D (September 2001): 19" 1600×1200 (105.26 ppi) IPS

 

[jeremywork]

However, dual-link SST may not work on the Iris Pro 5200 in general (it doesn’t with the Dell UP2715K), and the HP Z27q in particular may not work on NVIDIA GPUs (I’ll have to test this — it works fine on AMD GPUs). Without hacks, NVIDIA GPUs can do dual-link SST on the Dell UP2715K only IIRC.

This helps to know. Perhaps that MacBook Pro isn't revealing anything useful about the viability of this strategy. I should probably scrounge for an AMD solution- an RX 480 or 580 makes the most sense in my current exploratory context since its OS support leaves me legacy options.

Finally got round to testing the DH2G-DP with the 9503-DG3. 345.2 MHz works fine.

That's good to know! I'll note it on the page.

I haven't had the time to really delve into this yet, but I did think of something else I could try (in the meantime):

I pulled out my older Mac Pro 3,1 with an MVC-flashed GTX 780 3GB. It holds a weird place in the venn diagram of twin-DVI-DL equipped Metal-capable cards which can run under systems modern enough to have newer overrides/mtdd files. I already had an OCLP install of 10.12.6, and with this I was able to cold-boot with the two override passthroughs installed (Jack Doerner's EDIDs) and this time one of the ports detected a 1920 x 2400 @ 48 Hz display and SRX was able to pull the overridden EDID, which is the complete 256 bytes and matches the source (attached anyways FWIW.)

I'll work next on trying the other tiled EDIDs and installing a newer version of MacOS to an additional drive/partition. The working 1920 x 2400 signal is still split back into two 960 x 2400 even-pixel and odd-pixel stripes, so something in these passthroughs needs modification to work correctly. To bodge/collect more data I found a pair of Gefen EXT-DVI-EDIDP units that should be able to clone the 'display' (which I'll substitute for my programmed chips) and hopefully these will output correctly since they advertise 3840 x 2400 / DVI-DL support.

Other housekeeping: I've added my other Matrox boxes to the site; both DualHead2Go (Digital SE and ME) and the TripleHead2Go Digital Edition.

http://stormdecay.com:1984/rd/rd_matrox.html

 

[joevt]

I pulled out my older Mac Pro 3,1 with an MVC-flashed GTX 780 3GB. It holds a weird place in the venn diagram of twin-DVI-DL equipped Metal-capable cards which can run under systems modern enough to have newer overrides/mtdd files. I already had an OCLP install of 10.12.6, and with this I was able to cold-boot with the two override passthroughs installed (Jack Doerner's EDIDs) and this time one of the ports detected a 1920 x 2400 @ 48 Hz display and SRX was able to pull the overridden EDID, which is the complete 256 bytes and matches the source (attached anyways FWIW.)

I wonder why two displays weren't detected? AllRez should be able to get both EDIDs (if both DVI ports work). Then EdidUtil.sh can use edid-decode to get more than than SwitchResX's decoder can.
I'm not sure tiled displays work with Nvidia, except the Dell UP2715K. I guess this needs more testing. First need to find a tiled display that works with modern AMD (like RX 580) then try on Nvidia.

I'll work next on trying the other tiled EDIDs and installing a newer version of MacOS to an additional drive/partition. The working 1920 x 2400 signal is still split back into two 960 x 2400 even-pixel and odd-pixel stripes, so something in these passthroughs needs modification to work correctly.

I don't understand what "two 960 x 2400 even-pixel and odd-pixel stripes" means. Need a picture.

To bodge/collect more data I found a pair of Gefen EXT-DVI-EDIDP units that should be able to clone the 'display' (which I'll substitute for my programmed chips) and hopefully these will output correctly since they advertise 3840 x 2400 / DVI-DL support.

Not sure what this means. What EDID are you going to clone?

Are you saying the override passthroughs only pass single link while 1920 x 2400 48 Hz requires dual link? That would cause every second vertical line to be missing or something. Need pictures.

So using the Gefen instead of the passthroughs would be better as the Gefen supports DL.

 

[Amethyst1]

I'm not sure tiled displays work with Nvidia, except the Dell UP2715K. I guess this needs more testing. First need to find a tiled display that works with modern AMD (like RX 580) then try on Nvidia.

I'll test the HP Z27q (dual-link SST) with the NVIDIA NVS 510. It works great on the AMD Radeon RX 460.

I don't understand what "two 960 x 2400 even-pixel and odd-pixel stripes" means. Need a picture.

#111

 

[jeremywork]

I wonder why two displays weren't detected? AllRez should be able to get both EDIDs (if both DVI ports work). Then EdidUtil.sh can use edid-decode to get more than than SwitchResX's decoder can.
I'm not sure tiled displays work with Nvidia, except the Dell UP2715K. I guess this needs more testing. First need to find a tiled display that works with modern AMD (like RX 580) then try on Nvidia.

Both DVI ports can send 1920 x 2400 @ 48 Hz when directly connected to the display- maybe I should re-seat or re-flash the chip in the secondary EDID override dongle to eliminate the possibility that it just isn't readable. I'll check AllRez and EdidUtil tomorrow and report back. I also intend to try as @Amethyst1 suggested and just run the UP2715K EDIDs regardless of the pixel difference to see what the OS does.

I don't understand what "two 960 x 2400 even-pixel and odd-pixel stripes" means. Need a picture.

It's the same as in my prior post (#111.) It's a near duplicate of two images, though technically one stripe is all the odd pixels and the other all even, so both links are being passed successfully. From today's test I snapped this picture during startup:

Not sure what this means. What EDID are you going to clone?

Just a lazy way of getting the EDIDs from my TL866II+ onto the Gefen units, since I don't see any explicit mention of the Gefens accepting an EDID over USB or anything. They can clone the EDID of the display connected to the output side, so I'll just connect the EDID override passthrough units I've already pictured to the output side so the Gefen units can assume their EDIDs.

Are you saying the override passthroughs only pass single link while 1920 x 2400 48 Hz requires dual link? That would cause every second vertical line to be missing or something. Need pictures.

Sort of, but I do get both links- just mis-assembled. This may be due to the passives in the LFH-60 to DVI-DL adapters from IT Plaza/Akihabara which assume the function of IBM's official DVI-DL converter box which used active signaling to produce the LFH-60 signal.

So using the Gefen instead of the passthroughs would be better as the Gefen supports DL.

In terms of the on-screen visual. But this issue seems entirely separate from whether or not the OS recognizes and assembles the two tiles. If I can get Mac OS to recognize a single 3840 x 2400 @ 48 Hz logical display, the apparent splits in the displayed image are a solvable problem.