makes me wonder if it has enough logic to emulate this card https://www.youtube.com/watch?v=0W1t2_EJG9w
though I guess the drivers would need to be rewritten to use the pci bus.

That's really cool, but thinking about it I'm not surprised.

Despite popular believe, parallel and serial ports are neither primitive nor obsolute.
Rate the contrary. In industrial applications, they're still well alive.

The LPT port gained many features over the years. There's DMA operation, SPP mode, PS/2 mode, Nibble mode (4-Bit), ECP mode, EPP mode.
Supporting all this requires complicated silicon.
It's just logical to use FPGA here, especially if the old chips are no longer available or if newer standards must be supported.

Serial ports have outperformed the original 8250/16450 compatible legacy modes and provide non-standard modes up to a few dozen MBit/s.
They even can be used by their separate native APIs.

And that's not just since yesterday.
Even by the late 90s, highly programmable USB Serial dongles existed.
They contained a fully programmable microcontroller and
could be used with Macintosh serial port hardware (RS-422), as well. Not just RS-232.

Edit: PCI Express and PCI are the same, software wise.
So I think it might be possible to operate PCIe chips with a PCI compatible command set.
Since this is an FPGA, it might be possible to use the I/O pins to support serial or parallel I/O, as needed.
That's why little to mo extra hardware is necessary to interface with outdated PCI slots.
So maybe the FPGA payed off twice for the makers of the PCI Parallel Port card.

weedeewee wrote on 2023-02-01, 19:48:

makes me wonder if it has enough logic to emulate this card https://www.youtube.com/watch?v=0W1t2_EJG9w
though I guess the drivers would need to be rewritten to use the pci bus.

Possibly, though it comes accross as documentation being so bad you need a whole original setup to probe for bus timing and command sets etc. I think the main attraction with that though is having earliest optical drive working for earliest PCs, since there's ways of doing it on 5 more recent bus standards for another 5+ drive interfaces if you just want to read CDs... Laserdisk drives tho...?? There was people working on that, laserdisks might have been the future but all the tech that reads them stayed stuck in the early 80s.

BitWrangler wrote on 2023-02-01, 21:50:

weedeewee wrote on 2023-02-01, 19:48:

makes me wonder if it has enough logic to emulate this card https://www.youtube.com/watch?v=0W1t2_EJG9w
though I guess the drivers would need to be rewritten to use the pci bus.

Possibly, though it comes accross as documentation being so bad you need a whole original setup to probe for bus timing and command sets etc. I think the main attraction with that though is having earliest optical drive working for earliest PCs, since there's ways of doing it on 5 more recent bus standards for another 5+ drive interfaces if you just want to read CDs... Laserdisk drives tho...?? There was people working on that, laserdisks might have been the future but all the tech that reads them stayed stuck in the early 80s.

I was just wondering since I have 3 cd drives which have such a 15 pin interface, 2 philips (461 & 462) and 1 ... Laser magnetics..., can't remember the name., yet I haven't come across any interface cards. 😒

Yah, it's tough with those kind of things, there's an Aztech soundcard with that kind of interface on Re: Aztech Sound Galaxy cards very easy to overlook and realise that it's a CDROM interface at all since everything else had a lot more pins.

Thought of something that logically "Makes no kind of sense whatsoever", the kind of thing I might have done in my 20s as an educational experience..

8 bit per pixel, possibly quite slow, 3D accelerator, DX3 or something, 8 bits per pixel for two reasons, that's all the bits this Spartan appears to be wired up to suck and blow, and 8 bit per pixel with a pallette is what you can feed into a VESA feature connector I think (Basically why they never get used, it seemed like a good idea in 1991? when 16 colors seemed amazing I guess) ... Sounds unlikely, but if there's 5000 gates, that puts it in the same realm as million transistor chip, which is around what early accelerators were. Might not get any local memory beyond a scratch buffer, so PCI bus will make it slow, BUUUT, you can bang output into 2D card mem through vesa connector. Stretch goal, some kind of cheap "jellybean" ARM or something as a geometry processor, might actually make it "fast". IDK but got this feeling that done heroicly, it could eventually get "software rendering" style depth and resolution at increased framerates over what the CPU could do... if it were a P133 down to DX2/66 .. but that might be hoping for too much and it would be a longass slog of a project.

Hm. I would have loved hardware mp3 playback for my 5x86 back in the day. Always had to play them at 22khz and mono. These days I don’t have an mp3 collection 🙁

its $50 and XCS05 is _tiny_. Would be cheaper to design own board from scratch for something like $8 T20F256C3 if you really really want PCI card with FPGA, or buy $5 Altera EPM7064 based PCI Diagnostic card.

EPM7064 is even smaller, and has no distributed ram capabilities as xilinx part.
I'm working on some cartridge stuff for my old ZX Spectrum clone, the VGA pixels serializer taking two bytes (bitmap and attributes) and pushing out 6-bit color pixels takes 41 macrocells out of 128 available on EPM7128.
XCS05 has just 238 LEs (4-input lookup tables), it's somewhat larger than my EPM7128, but still it's nowhere near to be good for anything more complex than "pin extender" or simplistic protocols translation applications. You can fit some I2C core there, or implement 16550 uart (few), but that's about it.

Something caught my eye over at opencores, AC97 controller, I wonder if you can hijack onboard AC97 codec with it, then pretend to DOS it's a soundblaster of a common type. Though several PCI sound implementations have been tamed for DOS now, so not all that necessary a thing.

Lot of stuff on there too, that's useful to integrate into one chip systems, not really so useful in a desktop PC when you can get $10 USB dongle for it.

Lot of stuff that would have been amaaaaaazing 25 years ago too like the hardware mp3 player, but anything that's all PCI enough to justify needing it probably don't show a blip on the CPU use graph to do it in software now.

Other things like PS/2 ports and floppy drive controller, if those aren't onboard then your bigger problem is lack of BIOS support, which add in boards aren't necessarily going to cure, and it's probably been cured better by using USB versions.

So yeah, remaining a halfassed solution in search of a unicorn problem 🤣

Aha, found something it might be adaptable for, but that is also true of other dual parallel port cards that have a wide range of address options. ... does not necessarily use it's FPGA abilities, but it's parallel port genes...

Found in a copy of Electronic Expermenters Handbook 1994 Winter edition, but probably appearing in other issues also. No direct link due to ambiguous copyright status and Vogon's policy on that. Let's just say it can be dug up. "SRAM II Add a bootable RAM drive to your PC" subheader "solid state disk drive" It uses dual 8255 PIAs in basically parallel port modes to interface multiple chips of SRAM (Which is how it came for cheap in 1994) Now, one chip does all the addressing, and the other chip does chip select, which won't be needed if just one large device is used, and also handles "data lines" which I am still getting my head around it's function in that regard. I'm not really seeing it as super necessary, merely a way to use up the rest of the 8255.... maybe switched buffer of some kind will do, if you even really really need one or it's a belt and braces thing. Anyhoo, seems to indicate, you could maybe do a 2MB flash drive with a dual parallel. Also seems to indicate, you might manage a 64KB bootable dongle on a regular parallel port if you are a genius.... Then thinking a second, maybe a arduino type board on the parallel that's a tad smarter and can page a SD card to the 64KB buffer... there's maybe a path to that in here but it's very half baked at the moment, need a full patch pastry chef.

Anyhoo, wasn't in the first place suggesting these were worth buying at new prices, but they are turning up at surplus for peanuts and at eWaste facilities, where corps had them in newer systems because they had speciality form printers etc working on parallel ports when they disappeared from commodity PCs.

If you're into fully reconfigurable I/O you might want to look at "Anything I/O FPGA Cards" by Mesa Electronics, particularly the 5i24 series. It's a PCI card, hosting reasonably big Spartan-6 fpga and level shifters. They provide precompiled configuration streams for a dozens of interfaces typically associated with CNC machines, and they shared the designs sources. Unfortunately that board has no ram onboard, so it limits the possibilities (no fancy GUS for you).