furan wrote on 2020-10-28, 12:24:

The proper way to put it is that 3Dlabs GLiNT chip was supported by RenderMorphics! So, from what it sounds like it was for the original GLiNT - the 300SX, with the right driver.

Page 16 of this magazine has a great article about the GLiNT 300SX, and mentions RenderMorphics support and shows screenshots:
https://retrocdn.net/images/6/69/Edge_UK_015.pdf

Very interesting article for sure, first time I've seen that one!

I agree with you that it was for the 300SX and we can just keep an eye out for that driver to turn up at some point in the future.

I also had a look at Warhammer: Shadow of the Horned Rat which Stiletto mentioned as another RL game. It has the same RMORPH directory/file structure as Cyberspeed. If what you say is true that copying over the Gigi.dll files would enable support that means that this game could also potentially be accelerated if we got hold of the driver.

vetz wrote on 2020-10-28, 14:53:

I also had a look at Warhammer: Shadow of the Horned Rat which Stiletto mentioned as another RL game. It has the same RMORPH directory/file structure as Cyberspeed. If what you say is true that copying over the Gigi.dll files would enable support that means that this game could also potentially be accelerated if we got hold of the driver.

There may be more, that list of Reality Lab games (Re: 3D Accelerated Games List (non-3Dfx proprietary APIs)) is not believed by me to be complete. It's mostly what came up with an Internet search for some Reality Lab filenames.

furan wrote on 2020-10-28, 07:53:

Here are some of the datasheets I got, sort-of related to GiGi:
300SX: https://www.dropbox.com/s/kqnxaeeq57aqcvk/GLI … %201.0.pdf?dl=1
500TX:https://www.dropbox.com/sh/7zgtmp1zo2xg43e/AA … DTrdJfm9va?dl=1

Thanks for these!! Will happily take the rest when you've the time. I forwarded these onto vlask of VGAMuseum just now, and MAMEdev now has a copy for future emulation. I also pinged Bitsavers.org but we'll have to see, he tends to focus on things he scans rather than unedited originals. 😀

Stiletto wrote on 2020-10-28, 17:39:

Will happily take the rest when you've the time.

I had not offered them yet 😀 - I need to see if he's okay with me giving them out - some of them are for unreleased chips during their last days.

furan wrote on 2020-10-28, 09:22:

disassembling the CyberSpeed GLINT drivers, they make ExtEscape calls

Yeah, great job furan!

furan wrote on 2020-10-28, 10:43:

Note that I linked a beta 0.98 driver on my Dropbox earlier in the thread 😀 It's from the DirectX 2.0/3.0 CDs. It doesn't have an escape handler for 0x3D3D, and unfortunately this 1.0 driver you linked does not have it either.

Version numbers can misleading, I think too that the 1.0 is the oldest. The release doc of the beta 0.98 driver has a creation timestamp "1996-Mar-04 12:08".
And the beta 0.98 is 3Dlabs version number but the other three are Creative Labs.
Here are by time (latest file date in the driver files):

1    date    |    company    |     DISK ID     |  Version   | DRV  | VXD  | note
21995-Nov-01 | Creative Labs |   3D-95STD-1-US | 01.00.00 | 1.00.000 | 1.10     | CGL
31995-Dec-04 | Creative Labs |   3D-95STD-1-JV | 01.10.00 | 1.00.000 | 1.11     | CGL
41996-Mar-06 | Creative Labs |   3D-95STD-3-US |  1.11    | 1.10     | 1.13     | CGL
51996-May-09 |     3Dlabs    |                 |          | 0.98beta | 0.98beta | CGL; DirectX; D3D but all slow
61997-Apr-09 | Creative Labs | C3DB-95UPD-1-US |  1.20    | 1.20     | 1.20     | CGL; DirectX; D3D

It seems the oldest one also have no ExtEscape calls, GiGi DLL-s not in use.
I have tried VxD calls with all of the 4 drivers by linked-in-binary CGL 1.61, the oldest freezes while card has found, the other three are not freeze, 3Dlabs 0.98 beta has problems, the other two are works.
I try to make some DirectX 2 D3D tests.

Gona wrote on 2020-10-29, 07:16:

Yeah, great job furan! <snip> It seems the oldest one also have no ExtEscape calls, GiGi DLL-s not in use. I have tried VxD call […]
Show full quote

Yeah, great job furan!
<snip>
It seems the oldest one also have no ExtEscape calls, GiGi DLL-s not in use.
I have tried VxD calls with all of the 4 drivers by linked-in-binary CGL 1.61, the oldest freezes while card has found, the other three are not freeze, but there is no signal to the monitor...
I try to make some DirectX 2 D3D test with 0.98 beta driver.

Thanks. Note that the driver won't be making ExtEscape calls - it'll either have a CONTROL handler or DrvEscape handler to handle the ExtEscape calls.

Stiletto wrote on 2020-10-28, 17:39:

Thanks for these!! Will happily take the rest when you've the time. I forwarded these onto vlask of VGAMuseum just now, and MAMEdev now has a copy for future emulation.

I wanted to get permission first. I got permission. I think it's good for this stuff to be archived but please let me know if you're going to do stuff like that. 😀

Shared folder: https://www.dropbox.com/sh/gqdg6aqyprx7x8o/AA … 9c61duq-Va?dl=0
Download:https://www.dropbox.com/sh/gqdg6aqyprx7x8o/AA … 9c61duq-Va?dl=1

You have the president of The Khronos Group’s permission to share these. I never expected him to respond to my email but he did. 😀

Gona wrote on 2020-10-29, 07:16:

I have tried VxD calls with all of the 4 drivers by linked-in-binary CGL 1.61 ... no signal

Aldhough I have used a CRT monitor on the 3D Blaster VLB, that monitor not likes the signal and do this without any message this is why was blank. I have retested the 3 late driver on another monitor and all gives picture. The 3Dlabs 0.98 was unplayable, but the Creative 1.11 and 1.20 versions are works with CGL.

Gona wrote:

Aldhough I have used a CRT monitor on the 3D Blaster VLB, that monitor not likes the signal and do this without any message this is why was blank. I have retested the 3 late driver on another monitor and all gives picture. The 3Dlabs 0.98 was unplayable, but the Creative 1.11 and 1.20 versions are works with CGL.

Awesome!

I have tested Windows 3.1 driver also and CGL works with this driver too.

furan wrote on 2020-10-31, 02:17:

You have the president of The Khronos Group’s permission to share these. I never expected him to respond to my email but he did. 😀

Hats off to you guys.

Well i think you already know it, not sure, but will post it, in any case:

This program can extract strings from binary files and place it much more friendly to be readed.
i use
https://docs.microsoft.com/en-us/sysinternals … wnloads/strings
"strings.exe PDLTB170.CGL > PDLTB170.txt" command will create text file

and i found in these text files :
(PDLTB170.txt)
"*** VXD locked ***" - hmm it reminds win98 drivers but it can't be? its dos
"APPLE PIEAPPLE PIEAPPLE PIEAPPLE PIEAPPLE PIEAPPLE PIEAPPLE PIEAPPLE PIEAPPLE " // its padding it reminds me of C or C++ compiled binary

(L3DTB170.txt)
the biggest file, compiled in debug mode?
weird all CGL drivers have ~full set of exports , this one have no "cglBoardConfigure"
"WATCOM C/C++32 Run-Time system. (c) Copyright by WATCOM International Corp. 1988-1995. All rights reserved." //so yes its C or C++
"MODE_1024X768X16_3" // and more strings like that , list of resolution modes supported by driver?
"dumping register file:" "dumping 2D register file:" //guess its contains debug information
"[mem.c] NULL passed to _kfree!" "[mem.c] Invalid magic number: %08X!" // more sure its C , less C++
"RATIONAL DOS/4G" //dos4gw ? runs in protected mode ?
this file contains much much more strings than others files

(VRITB170.txt)
the most smallest file, probably because rendition verite have 'everything' to render, most perfect renderer (hardware) for that period

This is just what i think on looking on these strings =)

And *.cgl files is dynamic, so know format and creating cgl file its probably possible to make emulation , or even wrapper like glide2x.ovl can be used from dosbox with nGlide or dgVoodoo

RaVeN-05 wrote on 2020-12-12, 18:45:

Well i think you already know it, not sure, but will post it, in any case:

If you take the time to read this thread you'll find we used strings a year or two ago. We now have an SDK and understand how the API works. 😀

vetz wrote on 2020-08-14, 18:01:

Seems like we need to create a new compatibility matrix for the Permedia NT 😀

I have bought a Diamond Fire GL 1000 card for reasonable price. I just finished with CGL testings. We have only the "Integer Register" type driver, so "Integer Stack" and "Float Register" applications cannot find appropriate driver. Diamond Fire GL 1000 also can not work as "2nd VGA" (add-on 3D accelerator mode), or this CGL driver not support this, so BattleRace; Fatal Racing/Whiplash are switches over to nothing and no signal. (3D Blaster PCI with CGL API can works as add-on 3D accelerator but with other APIs like Speedy3D API, works only as primary card).
Diamond Fire GL 1000 has a weak DOS VBE (2D) compatibility, there are 2D problems with Battle Arena Toshinden, Actua Soccer and Euro 96. With Battle Arena Toshinden it is temporary and easily playable, but with Actua Soccer and Euro 96 serious, it was a bit hard to me to get to 3D part.

I think these problems caused by the BIOS of the Diamond Fire GL 1000 card, because with my Leadtek WinFast 3D L2200 (also a Permedia NT card) there is no problem with 2D parts of these games.
UNIVBE can not helps on these problems. Diamond has released a VBETUNE program for the Diamond Fire GL 1000. With Battle Arena Toshinden and Actua Soccer the VBETUNE temporary has solved the 2D problem, but when you exit from the CGL part and 2D loaded newly the 2D is bad again. VBETUNE do nothing with Euro 96 2D problem at all.
3D parts.
Battle Arena Toshinden CGL part looks the same as 3D Blaster PCI.
Actua Soccer CGL is very similar to the 3D Blaster PCI, but captions has some more small bugs as on the 3D Blaster PCI has:

Euro 96 CGL part looks also similar to the 3D Blaster PCI, but captions has a more significant bugs

Flight Unlimited (bundled with 3D Blaster PCI) works withoud problems, although texture the filtering is really different as the texture filtering of the 3D Blaster PCI

But sill much better than without texture filtering:

Although because the color of sky and border of the clouds the 3D Blaster PCI's drawing looks much better

Rebel Moon (bundled with 3D Blaster PCI) also looks a bit different because the distinct texture filtering technique. And on Diamond Fire GL 1000 it runs faster, almost too fast for me.

The DTDTEST is perfect on Diamond Fire GL 1000. The fps is slower than on 3D Blaster PCI but this one might looks a littlebit better on Diamond Fire GL 1000.

I have updated my page about these results. Here it is the compatibility matrix on image (scaled down):

Amazing work Gona!

I'm sure if some ex-Creative employee found all this they would get impressed of all the information. It's obvious this was internal development of unreleased products/features.

Remarkable job indeed, Gona. Although the blending you talk about is actually texture filtering.

Thanks guys!

Putas wrote on 2021-02-13, 19:43:

the blending you talk about is actually texture filtering.

I have used "blending" because this is an on/off option in the Flight Unlimited CGL versions, but "texture filtering" is the exact term so I have replaced.
Is there an exact filtering method name what 3DBlaster VLB uses?

Gona wrote on 2021-02-15, 06:44:

Is there an exact filtering method name what 3DBlaster VLB uses?

No, Gigi can only texture by point sampling.

Putas wrote on 2021-02-15, 13:34:

Gona wrote on 2021-02-15, 06:44:

Is there an exact filtering method name what 3DBlaster VLB uses?

No, Gigi can only texture by point sampling.

OK. So GiGi has no supports textute filtering instead uses point sampling (or nothing).
Later with Vérité, CGL driver can use texture filtering methonds.
I have updated my CGL page according this.
Thank you.

in Direct 3D , a point sample means using neighbor nearest filtering =)
At least here we see some kind of billinear filtering imitation.
is hard to name it properly 🤣

something like Unreal do in software render mode.

Diffusing pixels.

Dithering

Edit: looks like this "dithering" is done with original game palette , all colors available from game diffused