Ever saw the pouet.net discussions about splitting up the ZX Spectrum category?
While the general idea is good to find incompatibilities, there're only things you can do wrong 😜

shock__ wrote:

Ever saw the pouet.net discussions about splitting up the ZX Spectrum category?
While the general idea is good to find incompatibilities, there're only things you can do wrong 😜

Nah, didn't see that discussion but I can understand the categorisation issues. Yes, that's the general idea where you can find these incompatibities, but there should be a counter-example where there is 100% compatibility for the same demo. Afterall, someone must have tested a particular demo on working hardware before releasing it.

Here's a counter-example to my above demo link, but 100% working:

Demo link: http://www.pouet.net/prod.php?which=1444
Ran 100%: Yes
Graphics: Trident 8900D, ISA
CPU: Cyrix 5x86 100MHz
Chipset: OPTi 696
Sound: CT3930 SB32

@jaws: I am not sure how those infos should be collected in a useful manner. In your first post you had a config where the demo didn't ran, in your second you posted a config where it ran?!?
Also one can not see if the crucial point for the demo is the choice of sound or graphics card or cpu power...
I captured a lot of those old demos and built sometimes a specific system to get a certain demo running, so the list could get long...

As for Kukoo2 I can tell you that the trident card is already mentioned in the scrolltext in the demo. So the crucial part is the graphics card. It does run also fine on the ATI VGA Wonder series chips.

elianda wrote:

@jaws: I am not sure how those infos should be collected in a useful manner. In your first post you had a config where the demo didn't ran, in your second you posted a config where it ran?!?
Also one can not see if the crucial point for the demo is the choice of sound or graphics card or cpu power...
I captured a lot of those old demos and built sometimes a specific system to get a certain demo running, so the list could get long...

I dont think there is a perfect way to collect this info - this is just a start.

The pivotal info is whether the demo is working 100% or not, the demo name itself is the other, and finally the config used in the template is the third piece of info. From this, I had in mind a 3D matrix where: x-axis = 100% working or not, y-axis = Demo name, and Z axis = config details. Or something along those lines. Any suggestions would be welcome.

For example, in the above matrix, you can scan the demo name quicky, and scan whether it worked 100% or not. Then focus on what made it work or not work. My two examples show that the only difference making that demo work was the graphics card - changing from Cirrus Logic 5430 to Trident 8900D.

elianda wrote:

As for Kukoo2 I can tell you that the trident card is already mentioned in the scrolltext in the demo. So the crucial part is the graphics card. It does run also fine on the ATI VGA Wonder series chips.

The Pouet link also mentions S3, but there are various chipsets, and I couldn't get a single S3 card to scroll smoothly. It also mentions ATI in the Pouet link, but which chipsets? I could only get it working on a MACH64 GT core (in a Rage card), and would not work properly on a MACH32 core, for example.

I can understand that the list can become very large, but hey, there are a lot of hardware variations and demos out there. I would be interested in your specific builds used for specific demos - that would be useful info.

EDIT: Typos.

I usually give system config infos in the description: https://www.youtube.com/watch?v=j9wj1UVqlWo

Demo link: http://www.pouet.net/prod.php?which=289
Ran 100%: Yes
Graphics: S3 Trio3D/2X
CPU: P3 800 MHz
Chipset: 440BX
Sound: EWS64XL
problematic was that it required the DPMI API from Win9x, so it ran only in a DOS window in Win9x.

Demo link: http://www.pouet.net/prod.php?which=2048
Ran 100%: Yes
Graphics: TSENG ET4000AX TC6058AF
CPU: 386DX-40
Chipset: ?
Sound: SB16
requires this specific graphics chip revision

I have encountered various problems with Second Reality:

- On S3 cards (Nitro 3D/Virge GX, Savage4) the screen goes blank/loses sync momentarily at the point where the chessboard first appears
- On NVidia (GeForce2 MX) there is palette corruption during the opening spaceship scene
- All OK on Western Digital WD90C24A
- All OK on ATI Mach64 VT

I think the problem is the nature of the demo.
Some demos are written quite specifically for a certain CPU, graphics card and/or sound card, so it's easier to list the configurations it does work on than the ones it doesn't work on, since they are far more numerous (our own 8088 MPH being a prime example, written solely for real IBM 8088+CGA systems).
Many demos however are quite generic code, and will work on most x86-compatible CPUs, VGA-compatible videocards and SB- or GUS-compatible sound cards. So configurations where it doesn't work are the exception to the rule.

Usually the biggest problem is trying to find a boot configuration that makes the demo work properly (conventional memory, upper memory blocks, EMS, XMS etc).

@ elianda: Thanks for those links - something to work with.

@ bjt: Which Second Reality demo are you referring to? I get several results when I search on Pouet. Please provide link and/or use template above. Cheers.

@ Scali: Some good points there. A main assumption here is that one has the software side all setup correctly. Even so, something as simple as a BIOS setting can throw things off.

It is kinda hard to get a feel for what is the norm and what is the exception. I used Gonas compatibility matrix as a premise: Take the most compatible cards and this should translate to the most compatible cards for demos. However, so far, I'm not seeing this. It's interesting that bjt mentions a Mach64 chipset - while it is so-so in Gonas matrix, I've found this chipset to be one of the most compatible ones in the limited testing I've done so far.

It is actually a lot more complicated. Often demos use some low res high color mode. The availability of those modes does not depend only on the graphics chipset but also if it is available in the vesa bios. So you may have the same card with different bios versions and one card runs the demo, the other doesn't. Most cards do not offer all relevant modes, like 512x384 at 8, 15, 16, 24 and 32 bpp.

Then there appear quirks like this:
Planet Logus T5
I tried with different cards, where most showed just graphics trash (i.e. assuming wrong memory layout for a specific mode)
Then an ET6000 worked for the main scenes, but did not worked for the end scroller.
A S3 Trio3D/2X however worked for the end scroller but not for the main demo.
In the title scene I had flickering which I could not get rid of with various cards that ran the main demo.
This means basically I found a setup where it runs to 75%, but no satisfactory solution.

edit: The demo has a CPU speed problem as well and will not run on anything faster than a P2 233 MHz.

elianda wrote:

It is actually a lot more complicated. Often demos use some low res high color mode.

Indeed, an extra complication there can be that the video card can do it, but your monitor cannot sync to it. LCD screens seem to be more picky than the old multisync CRT screens in that respect.

elianda wrote:

The availability of those modes does not depend only on the graphics chipset but also if it is available in the vesa bios. So you may have the same card with different bios versions and one card runs the demo, the other doesn't. Most cards do not offer all relevant modes, like 512x384 at 8, 15, 16, 24 and 32 bpp.

Yup, some demos just have native code for a few common chipsets (VESA BIOS Extensions didn't exist until later in the 486/Pentium era).
ET4000 is your best pick for such demos, it was one of the fastest cards around, and very popular with demo coders.
For demos that use VBE, you'll either need a card with good onboard VBE implementation (rare), or one that is supported well enough through Univbe (Scitech Display Doctor).

I would say that the ET4000, a GUS and an SB Pro (not 16) are the cards you must have for DOS-era PC demos.
Even then, some stuff is just broken. I did a blog on Crystal Dream a few years ago: https://scalibq.wordpress.com/2012/03/14/olds … ol-demo-fixing/
It used to work on my SB Pro 2.0 when I had a 386SX-16. When I tried to run it on my 486DX2-80 with the same card, it didn't work.
As I found out, it did work when I disabled the turbo... And I could re-enable the turbo once the music started playing.
So it just failed to detect/initialize the card in a faster system.
Funny enough various clone cards did not have this problem because their DSPs responded more quickly than a real SB Pro 2.0.

So, sometimes it's a certain combination of hardware that makes it work/not work.
In the end I tracked down the faulty SB detection/initialization code and replaced it, so now it worked fine on my 486 with turbo.

Scali wrote:

In the end I tracked down the faulty SB detection/initialization code and replaced it, so now it worked fine on my 486 with turbo.

I've read your post before on that demo. I love how you solved it 😀

vetz wrote:

I've read your post before on that demo. I love how you solved it 😀

Thanks.
I come from the world of C64 and Amiga, where it was somewhat common to fix games and demos.
In the PC world this seems rare, especially for stuff where source is not released. But it's just as possible as on C64/Amiga.
Perhaps more people should do this, more games and demos could use fixing 😀

elianda wrote:

... Then there appear quirks like this: Planet Logus T5 I tried with different cards, where most showed just graphics trash (i.e […]
Show full quote

...
Then there appear quirks like this:
Planet Logus T5
I tried with different cards, where most showed just graphics trash (i.e. assuming wrong memory layout for a specific mode)
Then an ET6000 worked for the main scenes, but did not worked for the end scroller.
A S3 Trio3D/2X however worked for the end scroller but not for the main demo.
In the title scene I had flickering which I could not get rid of with various cards that ran the main demo.
This means basically I found a setup where it runs to 75%, but no satisfactory solution.

edit: The demo has a CPU speed problem as well and will not run on anything faster than a P2 233 MHz.

Did some tests, and both these configs worked better:

Demo link: http://www.pouet.net/prod.php?which=5553
Ran 100%: Not quite, ran high-end settings fine, including end credits, but failed motion-blur option with artifacting.
Graphics: ATI Mach64 GT (Rage Pro), PCI
CPU: K6-III+ 480MHz
Chipset: 430FX
Sound: GUS MAX 2.1

Demo link: http://www.pouet.net/prod.php?which=5553
Ran 100%: Yes, ran high-end settings fine, including end credits and motion-blur option.
Graphics: SiS 6215, PCI
CPU: K6-III+ 480MHz
Chipset: 430FX
Sound: GUS MAX 2.1

If you or anyone else has more demos to share, that would be much appreciated, cheers.

I have a problem, running next demos in dos/win98se
I tried them on iP1-233/64 ram, voodoo1
slot1 п3-600, 128 ram, voodoo3-1000
slot1 p2-350, 64 ram, voodoo 3-1000
sound is aureal vortex2 or awe64 gold

1. yume 2 by inf
http://www.pouet.net/prod.php?which=310
Not enough memory on exec,
mem shows 590кб std memory

2. The Asylum by Moppi Productions
http://www.pouet.net/prod.php?which=70939
crash with error in glide2x.dll, tried a hell-lot of driver versions

3. First Demo II by Smash Designs
http://www.pouet.net/prod.php?which=1213
readme says need a dos4gw версии 1.97 or higher, but I got
dos4gw 1008 can't load executable format

1. Eden by Psychic Monks
https://www.pouet.net/prod.php?which=2613
Refuses to run in any color mode on ATI Mach64, even if using either ATI's VBE utility or UniVBE

2.) Gateways by Trauma
https://www.pouet.net/prod.php?which=295
As with Eden, this absolutely fails to run on a Mach64

That's a really big issue on pouet. Mostly no information about required cpu/ram/win/dx-version/etc.

I've been watching for demos that run on Windows 2000 with extended kernel and BWC drivers.
Hardware is (at the moment) Athlon II X2 280, GTX 650 Ti, 3GB RAM.
Many demos will ask for the user's screen aspect ratio, but then anyone who admits to having 4:3
(or even 16:10) will just get extra black bars slapped on. I've noted below where it might be better
to lie instead 😉

Demos that ran:
Stargazer (lie about aspect ratio to get rid of black bars)
ASD Rapture (uncheck 'force 16:9' to get rid of black bars)
ASD Lifeforce (uncheck 'force 16:9' to get rid of black bars)
Interstase
Timeless (64K)
ASD Happiness is around the bend (worked on Radeon but has trashed geometry on GeForce)
flt Panic Room (64K)
Ziphorse
flt One Wish
Wir sind einstein (lie about aspect ratio to get rid of black bars)
FR-043 (lie about aspect ratio to get rid of black bars)

Though, if you lie about the aspect ratio, are circles still circles on your screen?

elianda wrote on 2021-07-21, 19:26:

Though, if you lie about the aspect ratio, are circles still circles on your screen?

Generally, no. Maybe with a little debugging one could find the projection matrix and patch it to fix that as well.