I think a lot of these projects don’t use that kind of thing, which makes me wonder if there’s a market/space in the community for a hardware graphics driver off the cheapest FGPA possible that could drive a TFT and enable flappy bird at a bit more than 20FPS
The down side is that it made games super hardware-dependent, and porting games from one console to another often meant re-designing significant parts of the gameplay and artwork. So my first thought is this would make it too much work for the modern hobby gamedev scene, but then again people make games for CHIP-8 so who knows.
Edit: Turns out I was thinking of PICO-8 which is a similar but higher spec virtual console: https://www.lexaloffle.com/pico-8.php
Slightly off-topic, but: it was the most important upside - the game could conceivably be optimized for the characteristics of the controller(s), and in some cases, display (hello, Nintendo DS).
Now that cross-platform portability is the name of the game, most games tend to have bland control schemas, pretty much the lowest common denominator of (XBOne, PS4, PC with mouse and keyboard). Playability suffers. Granted, there are still some hold-overs, among them flight sims, but that's mor of a niche thing.
But by and large, we're in bear part of the cycle of game control innovation.
Remember the Nintendo DS? It used two ARM cores and a 3.7V LiPo battery. And those cheap 'Arduino'-compatible LCDs all have ASICs in them to drive the displays and expose the display RAM over a common communication interface. The actual display drivers support SPI/I2C as a convenience, for fast stuff they have parallel interfaces similar to actual HD 24-bit RGB displys.
In fact a quick search shows that it is already in progress.
https://hackaday.com/2016/01/31/tv-transmitter-uses-esp8266/
https://hackaday.com/2016/03/01/color-tv-broadcasts-are-esp8...
https://www.youtube.com/watch?v=5t1_XNc3vNw
And a VGA signal generated from an esp32 with a few resistors.
It's fast enough to generate a TV signal in software, so yes it's definitely more powerful than an Amstrad.
To be fair, you aren't comparing like with like.
The amstrad has 100+(?) touch 'sensors' in its keyboard. It has digital to analogue converters for the monitor and printer also. It may also have had an additional processor in the floppy drive.
Do those code wheel anti piracy things they used to have count as a hardware crypto accelerator? (sir Alan Sugar would say yes!)
So yeah the ESP32 is still definitely more powerful than the Amstrad.
Edit: I discover now that there was also a Amstrad PCW family of PCs [3], though it don't see a PCW1512 model in there, so I'm not sure which one the parent was talking about.
[1] https://en.wikipedia.org/wiki/PC1512
Writing code for it isn't hard, if you're familiar with the arduino IDE.
I put a brief review here with more details, but I'd recommend it if you're willing to play and experiment:
It is not unlike an arduino in a gameboy form factor. They make a wonderful gift for all ages.
There are downsides to the FPGA approach though. Even the cheapest FPGA can easily end up being more expensive (and more powerful) than the system CPU, putting you in a tail-wagging-the-dog scenario. And if you want your project to have a reasonable lifespan, continued FPGA availability can be a problem.
Many projects just drive video directly from the CPU, but that works better when you have something more powerful than the ESP8266 used here. It would be really nice if there was a simple display chip that this sort of project could take advantage of, but I don't know if FPGAs are the answer.
I have this crazy idea of a 8-bit avr with an xmem interface connected to ESP32 GPIO. Assuming the higher clock rate of the ESP32 is enough for it to bitbang pretend to be an SRAM it could give the AVR of 64k of 'video ram'.
It's a bit weird delegating a more powerful processor to a subsystem, but at this level I don't think it is so much about the practicality as the aesthetic.
Up to the PS3, consoles used to have dedicated hardware which required devs to understand it fully, in order to be able to extract all available performance.
This is one of the reason why late titles were, as a general rule, technically head and shoulders above compared to the launch ones, as the devs acquired experience in that particular hardware.
The Xbox kind of changed that with the "PC-in-a-box" architecture, which made the HW much more predictable but with less margin of improvement over time.
If you want better performance, just buy esp32 for 6 dollars shipped - it has dual cpu, increased speed and many other improvements.
But I guess your point still stands. If you have this huge machine making tiny super fast general chips, what’s the point of trying to go specialized
Plenty of gaming power already available, even if that means Assembly to actually take it to the limit.
Specialized hardware solutions often use less power than another cpu core.
This is important because there are lots of uses involving leaving the esp running on a small battery for a long time