VOGONS


First post, by darry

User metadata
Rank l33t++
Rank
l33t++

I have Samsung 204B 20 inch monitor with a native resolution of 1600x1200 that works perfectly under Windows (98, XP, 7) at its native resolution with a multitude of different video cards from Nvidia, ATI/AMD and 3Dfx .

However, under DOS, I have yet to find a video card that will allow more than 1280x1024 . I have tried
- Radeon 9700 over DVI
- BFG Geforce 6600GT over DVI
- Voodoo3 over VGA .

SDD's VBETESET lists no resolution above 1280x1024 at any bitrate and Duke Nukem 3D and Blood will not work above that either .

Is it some DDC or EDID issue with the monitor or are the cards to blame ?

EDIT : changed subject to better reflect nature of issue

EDIT 2 :
The bottom line is as follows .
It seems to depend on the TMDS used .
With Nvidia cards using a Silicon Image TMDS (Geforce 4 TI, some members of the Geforce FX and Quadro FX family), 1600x120 under DOS over DVI works . With a 6600GT and Geforce FX 5500 samples using an Nvidia integrated TMDS, I get 1280x1024 on the 204B
With ATI cards, it seems to be BIOS dependent at least for RV350/RV360 (one 9600 gives 1600x1200 with its original BIOS, but not with any other BIOS I have tried) . One 9550 gives 1280x1024 at 75Hz with original BIOS, but gives 1600x1200 with glitches with a 9600 BIOS . My only Radeon 9700 gives 1280x1024 at 75Hz .
The artifacting I complain about in the following posts was due to a Monoprice HDMI repeater-splitter I forgot I had inline between the monitor video card . Strangely it worked fine with ATI integrated (except maybe Radeon 9700, which I have not retried yet) and Geforce 6 integrated TMDS chips only .

Last edited by darry on 2017-10-03, 01:40. Edited 6 times in total.

Reply 1 of 23, by wbc

User metadata
Rank Member
Rank
Member

S3 Trio3D/2X supports 1600x1200 8/16bpp (don't know interlaced or not), same goes for Savage4

--wbcbz7

Reply 2 of 23, by darry

User metadata
Rank l33t++
Rank
l33t++
wbc wrote:

S3 Trio3D/2X supports 1600x1200 8/16bpp (don't know interlaced or not), same goes for Savage4

Thanks for the info. Alas, I have no working examples of either in my current inventory .

I do have another monitor (LG L246WP) whose native resolution is 1920x1200, so I decided to try that and 1600x1200 works on my 6600GT over DVI !!!!
So it is the Samsung 204B's fault after all . Now I need to find out why and how to fix that . EDID dumping and comparing is in my future .

darry wrote:

SDD's VBETESET lists no resolution above 1280x1024

I also just learned that this tool is limited to 1280x1024 anyway .
The mode listing in VESA24_2.zip is a much better test as it shows 1600x1200 as available with the LG monitor, but not the Samsung .

Reply 3 of 23, by darry

User metadata
Rank l33t++
Rank
l33t++

I have dumped and compared the DVI EDID for both monitors and can find no reason for the 204B's lack of 1600x1200 VESA mode .

Both monitors expose identical timings for 1600x1200, the only difference I can find (other than the CEA extension block in the LG L246WP), is that the LG also exposes a 1920x1200 mode (its native mode) .

I have attached the EDID dumps for both monitors. Hopefully somebody here has an idea as to what is wrong with the Samsung .

Reply 4 of 23, by darry

User metadata
Rank l33t++
Rank
l33t++

OK, I did a quick test and flashed the LG's EDID into the Samsung .

Now every DOS mode gives me a "Non Optimal resolution" message and the monitor's OSD indicates 1920x1200 (even though the display is not cut off and is centered), BUT 1600x1200 is present in the list of VESA modes and Duke Nukem 3D now works at 1600x1200 . Windows also still works at up to 1600x1200, as it should .

I guess the fun of trial and error begins now if I want to get rid of the "Non Optimal resolution" . I will probably start by removing the HDTV modes from the CEA extension block (the Samsung EDID does not have an extension block) and adding a 1600x1200 detailed mode (assuming that it will let me) .

If anybody could share and EDID dump from a 1600x1200 monitor that does work under DOS over DVI at native resolution, that would probably help.

Reply 5 of 23, by darry

User metadata
Rank l33t++
Rank
l33t++

At this point, I am pretty much giving up .
Even with only legitimate supported timings, I still get the "Non Optimal resolution" message . I have noticed that the video card seems to be sending 1600x1200 at all time at 73.8 KHz (60Hz refresh) in DOS and 75.4 KHz in Windows at 1600x1200 where I have no issues .
So either
a) There is some bug in the BIOS implementation on the video card
b) The monitor is simply mis-detecting .
c) Something else I just do not understand is going on .

For now, I will try to live with the message. If it annoys me too much, I can always flash back the original EDID and live with the video card running to 1280x1024 (or scaling to 1280x1024) and the monitor scaling that the rest of the way to 1600x1200 .

Reply 6 of 23, by Rawit

User metadata
Rank Oldbie
Rank
Oldbie

It might be a "reduced blanking" issue. In order to save bandwidth, reduced blanking was introduced, which slightly changes some variables in the signal. Perhaps the Samsung has issues with that. It explains the difference between DOS and Windows too, as the Windows driver can detect optimal monitor settings and therefore doesn't apply reduced blanking.

YouTube

Reply 7 of 23, by darry

User metadata
Rank l33t++
Rank
l33t++

AFAIK , reduced blanking was introduced to allow extra short blanking intervals in order to extend the resolutions possible through single-link DVI . Also AFAIK, it is not required for 1600x1200 at 60Hz .

Anyway, the detailed timings I am using are the following :

Reply 8 of 23, by darry

User metadata
Rank l33t++
Rank
l33t++

I just can't give up, it would seem . I just happened to score an FX5900, which works at 1600x1200 in DOS with the original EDID on the 204B !
So the 6600GT is the one behaving oddly, at least with my 204B . I guess I'll have to live with that .

Why not use the FX5900 (which has an Silicon Image TMDS)? Well it artifacts rather badly after a few minutes of use in DOS. The caps on it have not leaked, but the main fan/heatsink was severely dust-logged before I cleaned it . According to the temperature probe in the Windows drivers, it is not currently overheating . I might eventually try cooking it and changing the caps anyway, but I digress.

Reply 9 of 23, by Rawit

User metadata
Rank Oldbie
Rank
Oldbie

Now that you mention the Silicon Image TMDS, I remember reading about the terrible compatiblity of the nVidia on-chip TMDS before during my quest for DVI/DFP under DOS/Windows 3.11. So it might be just that...

YouTube

Reply 10 of 23, by darry

User metadata
Rank l33t++
Rank
l33t++
Rawit wrote:

NI remember reading about the terrible compatiblity of the nVidia on-chip TMDS before during my quest for DVI/DFP under DOS/Windows 3.11. So it might be just that...

I remembered that my Quadro FX1100 also has a Silicon Image TMDS and, low and behold, it works at 1600x1200 in DOS, BUT it also seems to have become defective (green corruption in Quake 3 textures and menus and some weird pixels in the Windows splash screen) .

I seem to have a lot of dead/dying video cards in my inventory :

- Quadro FX1100 with display corruption
- MSI FX5900 with display corruption
- Geforce 3 with black screen

None of these display any visual sign of damage or bloated caps . I am especially surprised by the FX1100, as it likely would not have been overclocked or in anyway beaten during its previous life . Oh well, nothing lasts forever. That is part of the hobby as well, I guess .

Reply 11 of 23, by darry

User metadata
Rank l33t++
Rank
l33t++

Ordered a Quadro FX3000 and a Quadro FX1100 . If both work, I will have a backup .

Reply 12 of 23, by Rawit

User metadata
Rank Oldbie
Rank
Oldbie

Hope it works out for you. Googling 6600GT DVI problems gives a huge list of results. nVidia's on-chip TMDS is kinda flakey.

YouTube

Reply 13 of 23, by darry

User metadata
Rank l33t++
Rank
l33t++
Rawit wrote:

Hope it works out for you. Googling 6600GT DVI problems gives a huge list of results. nVidia's on-chip TMDS is kinda flakey.

Thank you for the good wishes, but lady luck does not seem to be on my side at this point .
I just received and tried the second Quadro FX1100 and it has similar, but not identical, artifacts in DOS and almost identical ones in some of Quake III's textures. As a "bonus" over the first FX1100, it makes the PC quite unstable (one of the crash scenarios is 100% reproducible by running Quake III at 1024x768) .

Hopefully, the Quadro FX3000 will be better, but I am really starting to have doubts about the Geforce 5/Quadro FX family .

Reply 14 of 23, by Rawit

User metadata
Rank Oldbie
Rank
Oldbie

Does the fan work? Quadro cards should have an ok build quality. It does sound a bit like video memory issues/overheating. I assume your PSU is up for the task?

YouTube

Reply 15 of 23, by darry

User metadata
Rank l33t++
Rank
l33t++
Rawit wrote:

Does the fan work?

Yes

Quadro cards should have an ok build quality. It does sound a bit like video memory issues/overheating.

The temp sensor in Windows drivers indicates everything is OK (50ish Celsius).

I assume your PSU is up for the task?

465W Enermax (regularly cleaned, no cap bloat/leak) for a P3 1400Mhz with a PCI Voodoo 3, 3 sound cards, a SATA controller, an HD and a DVD drive .
6600GT works fine with the same setup, so I assume the PSU is OK .

I just got the Quadro FX 3000 and guess what ? No picture, just a black screen on one DVI out and severely corrupted display on the other .

I am either very unlucky lately or have been suckered twice .
Maybe this is some weird incompatibility with the FX series and my Ipox 3ETI23 board as I seem to recall the first FX1100 working fine in my M6TBD board . Looks like I will have to dig out some other AGP boards from storage and test out all those cards before I recycle them .

In the meantime, back to the 6600GT it is .

Reply 16 of 23, by darry

User metadata
Rank l33t++
Rank
l33t++

Bought one of these FX 5500 as a final "test" with Geforce FX family members for now . It uses the integrated TMDS, but it should give me an idea as to whether hunting for other higher end Geforce FX in a working state would be worth it . If it works well enough, I might actually use it .
http://www.ebay.ca/itm/inno3D-GeForce-FX5500- … 54AAOSwlY1ZEPBO

Reply 17 of 23, by Rawit

User metadata
Rank Oldbie
Rank
Oldbie

Well it doesn't sound like there is an issue with your system, but when you're so unlucky with cards... It's not a 1.5v / 0.8v AGP issue? The cards/mainboard are probably keyed correctly.

YouTube

Reply 18 of 23, by darry

User metadata
Rank l33t++
Rank
l33t++

I could not resist, so I ended up ordering an x800 GTO . Looking forward to testing it with Catalyst 6.2's beta support .

Reply 19 of 23, by darry

User metadata
Rank l33t++
Rank
l33t++

Bought one of these FX 5500

This ones gives artifacts too, even in DOS! It looks to be new old stock, so that is strange .

I could not resist, so I ended up ordering an x800 GTO . Looking forward to testing it with Catalyst 6.2's beta support .

No artifacts on this one in DOS, but I never got it to work in Windows. The driver fails to load with error code 23 or 24 . A Radeon 9700 I also have artifacts, even in DOS, but its driver loads in Windows .

I would stick with the 6600GT, but I discovered it crashes Windows while scrolling in Firefox and Opera, unless I lower resolution to 800x600 at 16 bits or use at 60 series driver where it just becomes extra slow (visible redraw) after a while . 3D seems to work fine, though .

PCI Voodoo 3 works fine in both 2D and 3D, incidentally .