hcyandluodayou wrote on 2025-04-23, 08:05:
Thank you very much for your professional explanation. This card indeed has only one VGA output interface. Can the three S3 chips work together like Voodoo to enhance the color depth or resolution of the graphics card? After all, this might be a RasterOps card, but it's strange because all the RasterOps cards I know of use NuBus interfaces. One of the S3 chips even has a manually soldered resistor. Is it possible that this is an engineering sample card?
That is a good guess, but I don't think it would be possible for them to work together in that way. The RAMDAC is the thing determines the final output capabilities to the VGA display and this one looks pretty standard for the time period with very limited resolution support.
I did a little bit of digging regarding RasterOps and came up with the same thing as you. Almost entirely NuBus cards, and I don't recall any of them having any resemblance to this with three complete, modern graphics chips on one board.
I noticed the hand-soldered resistors too! It is funny how they are connected together, floating at right angles... this is an usual way to add them to the board. That said, it used to be quite common for production cards to have manually soldered components of some kind. I think the reason is that it was often just not possible (due to time and expense) to design and produce new PCBs once a flaw was found in a design (for example, a glitch in certain situations that did not show up until a production run had completed), so they would just have parts added on like this to fix the problems after production. So, it is most likely not an engineering sample or prototype since it otherwise has normal markings and labels, but it is always a possibility.
PD2JK wrote on 2025-04-23, 09:33:
I'm thinking of some kind of software toggle switch. Three semi-virtual displays you can select from. Awesome card nonetheless.
Ah, that is an interesting idea!
I wonder, is there any precedent for such a card to exist? What exactly would be the use case for it? I guess technically it could be done for performance reasons... for example, if special software told the card to hold different high(ish) resolution images in the 1MB of memory available to each chip, the software could then allow fast switching between those images without having to bring them over the slow ISA bus... but that seems like an extremely niche product, and I can't honestly imagine a practical use for it that wouldn't also be usable (if slightly slower) with a normal 1MB card. I don't think that would do anything to facilitate faster real-time interaction with the display since the ISA bus would still be the bottleneck.
Maybe if you were, like, doing 2D animation on a computer and you had to quickly flick between 2-3 frames to line everything up while drawing... but wow, that is super obscure and probably not what this is for. 🤣
Man... now I *really* want to know what this is! I think it needs it's own thread in the video section.