Project VGA – Home of the Low Budget, Open Source, VGA Compatible Card (2008) | Hacker News Reader

Project VGA – Home of the Low Budget, Open Source, VGA Compatible Card (2008)

wacco.mveas.com

It’s amazing to see this and the similarities to my Amiga VA2000 graphics card that I did many years later, but I swear I didn’t see this before. VA2000 did HDMI output and some very basic 2D acceleration. It’s now sold out.

I just completed the PCB design for ZZ9000 (firmware is in beta already), a bigger follow-up based on ZYNQ and using the ARM cores for some graphics and general purpose acceleration. It will be open sourced as well.

Do you think there is demand to spin an open HD framebuffer+ARM card for other platforms as well?

ZZ9000: https://shop.mntmn.com/products/zz9000-for-amiga-preorder VA2000: https://github.com/mntmn/amiga2000-gfxcard

How much did the HDMI license cost? Been looking for HDMI output for a retro system, but it DVI would incur no cost, while HDMI would require extra $5k.

You can output DVI using an HDMI connector and not pay the license fee. So "DVI output via HDMI connector" gets you around the license fee. The fee comes in when you use the actual HDMI spec and implement HDMI stuff like Audio, CEC and actually calling it HDMI.

DVI is indeed my approach to get around the issue. Like you mentioned, the downside is losing those nice extras, like audio.

Slightly off topic: why did HDMI become the de facto standard when DP and DVI have no licensing fees?

Some combination of: HDMI consortium is backed by numerous large TV manufacturers and movie studios, DVI lacks audio support, HDMI already had a lot of traction by the time DP was released, DP initially lacked HDCP support, HDMI supports much longer cables than either DVI or DP.

Can we get a [2008]? Interesting work but not a competitor for today's GPUs.

    PCI bus interface (32 bit, 33/66MHz, 3.3/5V compatible)
    Xilinx Spartan 3 s400 FPGA (aiming for ~100MHz)
    16MB SDRAM (aiming for ~166MHz)
    Onboard programmer with USB interface

Tagging the year makes sense, but I would not it was not ever intended to be competitive with today's GPUs, as per the FAQ.

In 2008 it wasn't a competitor for 2008 gpus, not read it intended to be. Still, the date would be good.

It would be nice to see an update for pcie, uefi, and HDMI. Might even be easier to do today than that was back then.

If it supported ISA for the old DOS PCs that didn't have PCI bus it would be a good upgrade for the old school PCs out there. Monochrome MGA used a different address than VGA/CGA/EGA uses. So you could have two monitors in old school PCs.

Regarding the PC monochrome display adapter memory offset vs. CGA/etc. offsets, I wrote some direct memory write display code in Turbo Pascal to take advantage of that, which let me use two monitors in toy ways. But the only real use I recall hearing of the second display was for a debugger (at Microsoft).

i was never a turbo pascal user, but my first real programming environment was borland's turbo c, a close relative. it had a great debugger that would allow you to run your app on your primary color monitor, and display the contents of the turbo debugger on your monochrome display. turbo debugger was also great for single-stepping your way through other people's programs in the same way.

back in the day, i wrote a VGA-based game that, if it detected a secondary monochrome display, would put up a whole bunch of real-time-updating graphics statistics on it. that made performance tuning a lot easier.

I loved Turbo C, too. When TC came out, it put a good C compiler in the hands of some young programmers for whom the main Microsoft compiler was prohibitively expensive. Same with Turbo Assembler and Turbo Debugger.

Not very practical, but I remember a friend doing a demoscene intro or demo, around year 98, using both a VGA and an Hercules card in the same computer.

The Hercules screen showed some animation in the order of "Don't look here, the demo is in the other screen"

I'll have to check pouet.net to find it...

We should try to work towards having viable open source hardware, such as for security, freedom to improve and share, and to prevent slippery slopes like HDCP.

This particular project, for example, could be one piece of a usable open source hardware PC.

(2008)

If the author ever got this project working, there's no sign of it here. One of the last updates says that they got it to produce VGA output, but not under computer control. None of their HDL appears to be online either.

The whole opengraphics.org thing doesn't seem to be up either. I checked on Archive.org [1] but couldn't find anything.

[1] https://web.archive.org/web/*/opengraphics.org

I dug a little deeper, and it actually looks as though the project got a bit farther than I thought. Here's one of the less bit-rotted captures of their site:

https://web.archive.org/web/20110226000747/http://wiki.openg...

Sadly, the source code still appears to be lost.

I was thinking it was project back on slashdot a decade ago.

https://hardware.slashdot.org/story/08/05/21/2136243/open-so...

It turns out it's possible to drive VGA by "bit-banging", even from quite low-end hardware. This is how the Gigatron drives its display (https://gigatron.io/), but it's been done a few other times: https://hackaday.com/2016/12/08/bitbanging-vga-fits-in-under... https://hackaday.com/2011/05/10/bit-banging-vga-from-an-sd-c...

No mention of bit-bang VGA is complete without a link to https://www.linusakesson.net/scene/craft/

He has also done something far more minimalistic: https://www.linusakesson.net/scene/bitbanger/index.php

Is it possible to make a modern one on commercial FPGAs?

Perhaps simplifying it with a HDMI chip rather than implementing the protocol.

There are various open source FPGA projects that do HDMI output. Some of them probably run on hardware with enough extra LVDS lanes to also do PCIE. The MKR VIDOR 4000 has connectors for both HDMI and Mini-PCIe, and the datasheet for the FPGA on it suggests it might have enough lvds lanes to drive both simultaneously.

The fpga4fun.com website steps through how to create HDMI output code, how to make PCI code, and how to use the Xilinx wizard for PCIe support.

In past projects that actually tried to make desktop graphics cards, it sounded like the hard parts weren't making video output or talking to the host computer, the hard parts were providing VGA bios support for booting.

HDMI is actually pretty straightforward to implement on a modern FPGA. I'm currently working on such an implementation -- I've got low-resolution modes working with fabric serializers; currently working on an OSERDES implementation to make higher resolutions available.

What chip are you using? Kintex or Artix?

Neither! Spartan-6. (Which I suppose isn't quite "modern" either, but it's a big step up from the part being used by this project.) The relevant Xilinx documents [1][2] suggest that it's even possible on Spartan-3A parts.

[1]: https://www.xilinx.com/support/documentation/application_not...

[2]: https://www.xilinx.com/support/documentation/application_not...

Third sounds like a fun FPGA project, does anyone want to work on a similar one today?

Does it make sense? Xilinx AC701 or an older Altera Arria V dev Kit have all the hardware for this. Writing code for them might be interesting learning project, but there are more interesting stuff nowadays to learn.

Not sure what Arria V dev kit you're referring to, but the AC701 retails for ~$1.3k and is overkill for the target application.

What do you want to pay for it? Hardware alone might cost 300$, but you also need somebody that wants work for free. 900$ is realistic price assuming low demand for such niche application. Maybe even higher price like this one: https://numato.com/product/nereid-kintex-7-pci-express-fpga-...

I am not related to Numato, but getting 4k@60Hz requires mid range chip like this Kintex-7.

I also not sure, if somebody wants to open source more than schematics and Gerbers after putting few hundred working hours in such project.

Certainly not anything remotely close to $1k with unnecessary baggage. Where did you pull this imaginary 4K@60Hz performance requirement from anyways?

The original article discussed analog VGA with PCI host control. The GP said something similar. You're proposing high-bandwidth digital video and suggesting native PCIe dev boards...completely out of scope.

Take a look at the i.MX6 Rex project[1] designed by Robert Feranec and Martin Murin before presuming what people out there are willing to release. Comparatively speaking, a VGA over PCI video card is solo undergrad EE fodder.

[1] https://www.imx6rex.com/

Different for sure, but amazing that a $3-5 ESP8266 can output VGA (monochrome) with hardly any additional hardware: https://forum.arduino.cc/index.php?topic=564673.0

The ESP32 can do VGA colour with a few resistors:

https://github.com/fdivitto/fabgl

Reply on news.ycombinator.com