http://web.archive.org/web/20021003001657/htt … .com/index.html
Unirefresh Homepage before it disappeared from the net.

Email -> rob.muller@gmx.net

Thanks to the Waybackmachine.org I was able to look it up for you. I've never used the program.
Never heard of it either but maybe you can get in contact with him.

procerus wrote:

In this case the problem is that UniRefresh no longer works.

I am really curious what response you will get from Rob Muller, let us know, please. I could change VESA refresh rates with Hercules GeForce 2MX in DOS/Windows XP with UniRefresh or VBEHz, but I had same problem as you with GeForce FX5200 Inno3D. I made some enquiries in that matter several months ago. Of course, I got no reply at all from InnoVISION support (they already have my money so why would they bother 😠), but Gerhard Strangar (creator of freeware utility VBEHz) replied immediately (you can find his e-mail address in readme of VBEHz). It did not solve my problem though - he mentioned that there is probably some bug in my card bios and recommended trying other bioses, but I could not find any that would work.
I switched to ATI card then and even if it does not have VBE 3.0 support so UniRefresh or VBEHz cannot work with it, it was possible to change DOS refresh rates settings in bios with Radedit1.1d editor. I know that there is also bios editor for NVIDIA cards available, but unfortunately without possibility to change DOS refresh rates.

I'm sure Rob won't mind if I post his reply to my question about this matter since it may be of interest to others on this forum.

******************************************
I think it has been around 2 years since I removed the ms-dos development
environment from my system and moved on. I am not a gamer myself and I have
not used any of my dos programs for a long time.
I noticed that UniRefresh (at least the setup program) does not work under
Windows XP with my machine (P4 3GHz), I am not sure what the problem is.
When booting from a dos-floppy it works but my NVIDIA FX5200 only works at
60Hz, this is probably the same problem you are experiencing. I guess the
video card BIOS ignores the VBE 3 refresh rate changes. It is probably
extremely hard (if not impossible) to do anything about this.

Currently I am using a TFT monitor and refresh rate problems do not exist
anymore. After receiving your email I have started testing my monitor with
the test screen from the UniRefresh setup program. I noticed that all
resolutions looked bad using the DVI cable. When selecting the analog cable
(available with one knob on the monitor, DELL FP2001, 20"), every
resolution looks really sharp (the native resolution of the monitor,
1600x1200, is not supported by the video card as a VESA mode). It is really
surprising that the analog connection is that much better than DVI (this is
caused by the monitor). I did not expect that a TFT could look that good at
resolutions other than the native resolution, it looks fuzzy with the DVI
cable but sharp with the analog cable.

Maybe it is a good idea to buy a TFT monitor instead of a graphics card? I
don't think every TFT does such a good job on scaling as my monitor though.
The response time of my monitor is not as good as with a CRT but it looks
good enough when watching video/TV (I am one of the DScaler developers,
http://www.dscaler.org, I watch a lot of TV on the monitor).
I can recommend the Dell FP2001, especially since it seems that it costs
only 60% in the USA of what I have paid for it last month.

To be precise, the resolutions that I have tested on my monitor are:
- with the analog cable (sharp) and the DVI cable (fuzzy):
320x200
320x240
320x400
640x400
640x480
800x600
the dos text screen (IIRC 720x400)
- both cables equal quality:
1024x768 (sharp)
12080x1024 (fuzzy)
1600x1200 (sharp)

Regards,

Rob

******************************************

So it looks like the problem with nVidia's later cards is that VESA specific code in the card's BIOS no longer recognises calls to change the refresh rate. So us nVidiots are stuck with 60Hz in DOS.

I appreciate Rob's suggestion that this can be fixed by switching to a TFT monitor. And seeing that some of them are now supporting a decent range of low resolutions I would do so. Except in my case I play a lot of games with 3D stereo glasses and these don't works with LCDs. 🙁

My thanks to Rob for producing UniRefresh in the first place and thereby providing me with hundreds of hours of 85Hz refresh rate DOS gaming. Thanks to him also for providing me with excellent support even to the extent of producing custom TSRs to support specific games.

For myself I may have to either get on with building that "legacy" system or else think about installing dual monitors.

procerus

Just my two cents about this very old thread. I am experiencing this 60hz problem when running DOS games in Windows XP command line mode, and unirefresh does not help too, BUT :

I discovered that running these games through DOSBOX in fullscreen mode they run at a higher refesh rate. I get refresh rates ranging from 70hz to 100hz 😀 😀

thierrybo wrote:

Just my two cents about this very old thread. I am experiencing this 60hz problem when running DOS games in Windows XP command line mode, and unirefresh does not help too, BUT :

I can also confirm that the NVidia 6600GT (Leadtek PX-6600GT) does not work with UniRefresh v1.42 under MS-DOS v6.22, the BIOS seems locked to 60Hz on a CRT monitor. The (S)VGA registers to change the clock rate ARE functioning, since the 60Hz refresh rate is constant at all VGA and SVGA graphics modes. The text modes are at 70Hz. The recent NVidia video adapters simply do not allow VESA Int 10h calls to go above 60Hz in graphics modes and 70Hz in text modes.

This makes sense from NVidia and its partners from a support perspective, many of its customers use flat screen LCDs (vice CRTs) and would not take kindly to an 85Hz input. So NVidia locks in the traditional 60/70Hz limits and avoids any support headaches...

Too bad they did not used one of the reserved VESA Int 10h functions to allow this behavior to be modified.

dvwjr

This darn issue, got it with my GFmx440 too, I would love to see it solved. Not only does 60Hz flicker slightly, optically, in addition my monitor starts giving this very high tone too... arg. I understand that since it is disabled in the hardware, chances are slim for a software override... although I figure the windows driver does just that... I mean if it is possible for windows to set the refresh rates, why can't a MS-DOS utility imitate the windows procedure?

Maybe LCD screens are not that bad after all, since they don't have a refresh rate in that way. but these things will probaly go scaling my pixels? I don't use LCD screens much, I don't know really.

I have the same issue on newer nVidia VGAs starting 6600GT, 7300GT and 7600GS. This problem is probably present on 5xxx and all later cards. It doesn't matter who manufactured your VGA card because they mostly use nVidia reference BIOS with slightly modification to timings.

I was talking about it with author of VESA VBEMP driver for WinNT and he told me that nVidia REMOVED some part of VBE code responsible for refresh rate settins so there's no way to cure poor BIOS.

Interesting is that Scitech (before well known its UNIVBE) is now making SNAP SDK - universal graphics drivers for DOS, Win, OS/2 and others. Their SNAP 3.1 nvidia driver IS ABLE TO SET higer refreesh rate on GeForce 6xxx but not on 7xxx and newer (maybe later). But unfortunately SNAP SDK is only driver and API for nevelopers not TSR like UNIVBE was. Then is useless for us. The driver is provided in binary form only so it would be hadr to grab the refresh code but not impossible.
Another way is to do some reverse engineeriing of windows/linux driver to know how they did it. But it will be probably hardware dependent for different chips. It exceed my skills to hacking the drivers but maybe someone else can do it.

Life would be much easier if HW manufacturers will give full documentation to the chips (e.g. like intel do) to let others to write their own drivers/tools instead making some closed windows-only crap. Grrrh!

I also encountered this problem, but I did not find a universal solution.

Unfortunately, there is still no comparison table of all existing VBE 3.0 compatible video cards, which would reflect UniRefresh support and vBIOS version.

I experimented with different cards and UniRefresh worked great with the GeForce FX 5800 Ultra. But it didn't work with FX 5950 Ultra! I also tried GeForce FX 5600 Ultra and initially UniRefresh didn't work, but when I ran the card with vBIOS from another similar card (Based on the same GPU), the situation changed, UniRefresh worked. I also tried to find a suitable vBIOS for the FX 5950 Ultra, but without success.

If someone experienced could compare these two vBIOS from FX 5600, find the difference that makes Unirefresh work and make a patch for vBIOS from other cards based on this (First of all from 6/7 series), it would be very useful.

The situation with PCI-E cards (GeForce PCX, 6 and 7 series) is also not entirely clear. Is there at least one known card among them that UniRefresh still works with?

Yes, some 7xxx cards were reported to work.

The Serpent Rider wrote on 2022-11-24, 18:26:

Yes, some 7xxx cards were reported to work.

I doubt so. Most nV cards use nvidia reference BIOS with some modifications of timings etc. but the VESA core should be the same for whole family.

RayeR wrote on 2022-11-24, 22:23:

I doubt so. Most nV cards use nvidia reference BIOS with some modifications of timings etc. but the VESA core should be the same for whole family.

But it works - Re: Geforce 7 series backwards compatibility

From my understanding, some manufacturers somehow had screwed up BIOS since GeForce 4 Ti days. Perhaps Nvidia reference BIOS has some bug, which can be triggered with very specific adjustment. For example - adding brand ID .
I also have an example where manufacturer have managed to completely kill VBE 1.2 support on GeForce 2 card.

Where is described the problem of GF2? I'm loosing in it a bit, is the spreasheet available in other searchable form than just a PNG image?

The thread about GF7 looks very interesting but so incredible to me. I'll continue there (a pitty this theme [VN-VESA bugs] is shattered to many threds here, would be good to cross-link them in 1st post...
I still have my Asus 7900GT that should work with 7900GS too. I would like to try it but need correct vbios dump, I'm going to contact Riikcakirds, hoping he's still visiting here.

RayeR wrote:

Where is described the problem of GF2?

Results are marked in red and mentioned in notes for ASUS V7100.

No success with loading 7900GS vbios, continue in thread Re: Geforce 7 series backwards compatibility

I also tested different video cards and out of the box did not work any of the cards I have from the 7 series. I tried to flash one of them using the two BIOS versions suggested in the next topic, but this also did not give any result. Testing the 7 series is complicated by the fact that each BIOS must be flashed directly to the card, since UniRefresh freezes if the BIOS is loaded via RamBios. I haven't considered the sixth series in depth yet.

I also looked at how it all looks on an old PC and found that changing the frequency under DOS is only functional with an analog connection. While digital connections always use the same frequency (60 or 70 Hertz, depending on the resolution). So CRT monitor seems to be the best choice for DOS after all. At the same time, I made a dump with NV30.

Checked these working dumps on a new PC. With the 7 series, they don't work (at least through RamBios. Flashing, at least, is a dubious idea). With the GeForce PCX 5900, they worked and UniRefresh works with them, but there are also significant artifacts. That is, the frequency changes with the working BIOS from NV30 and NV31, but the BIOS itself is not compatible enough (It gives artifacts). I also tested about 150 BIOSes from the Internet for different FX59XXS, but UniRefresh doesn't work with any of them. Finding genuine PCX5900 BIOSes for testing was no longer so easy. The sites where they were hosted earlier are now broken (They don't allow you to download them), for example, like x-drivers.com. I haven't read the full BIOS collection from old video cards yet. The few versions that I accidentally found (From previous experiences with 5950) also do not support UniRefresh. Although overall compatibility under DOS is excellent, there are no artifacts. It seems that starting with FX59XX it has already been cut thoroughly. Here it remains to try some more versions from FX5700, FX5500 and Quadro, as well as video cards from the 6 series. Although, if there were really working maps in the sixth series, probably someone would have reported it here long ago.

Hurray! I managed to cope with this problem. While testing video cards, I noticed one feature-BIOS'es from different cards, it doesn't matter if AGP or PCI-E, if the GPU is one (Or very close) - are almost completely interchangeable for loading via RamBios, at least under DOS. The success rate reaches 98%. Using this principle, I started testing earlier maps (Since there were no shifts in the new ones) and immediately noticed one of them:

https://www.techpowerup.com/gpu-specs/geforce-pcx-5300.c58

The mysterious "NV37" is nothing more than NV34 + HSI (Transition bridge from AGP to PCI-E). And NV34, as you know, was used in one of the most common and popular cards - GeForce FX 5200 and FX 5500, in which, as one might assume, with some degree of probability, support for UniRefresh could still be preserved (After all, in some instances NV30 and NV31 it was still preserved). Out of the box, this map did not support UniRefresh. Another positive feature is the PCI-E bus, which means the ability to install the card in a modern PC, as an additional one, for the needs of MS-DOS. So I tried it. I pumped up what BIOS I managed to find from the Internet for FX5200\FX5500 and there were only 50 of them, which should have been enough for the first time. But what a surprise it was when almost every second such third-party BIOS correctly loaded via RamBios and fully supported UniRefresh! All VESA modes are available. Under DOS, absolutely full functionality. In total, out of 50 BIOSes tested, 48 worked, of which 25 supported UniRefresh. In addition to this, I also tested these 25 pieces a little more in-depth and chose from them, in my opinion, the most suitable for a particular PCX5300 instance (Both video outputs work with it and the picture scales correctly if you need to connect via DVI). Under DOS, in general, it works perfectly, but for Windows it is not necessary (There are significantly more powerful cards), it will stand exclusively as an additional one, especially for UniRefresh under DOS.

Well, here is another unfinished gestalt associated with moving to a new PC, finally closed!

I just use this: NVPatch - making UniVBE work on NVidia cards

Let's you patch UniVBE to work with your card, and then UniRefresh will work with it too.