Driving oleds at 2000-10,000 hz

[Pyrolistical]

With oleds pixel response times being marketed at 0.1ms, that means we could achieve refresh rates of 2,000-10,000 hz, if it weren't for 2 peaky problems.

1. Display interface (ie. Display circuits/DisplayPort) is too slow
2. Graphics processor fill rate is too slow

We can eliminate both problems by directly driving pixels with a FPGA. One FPGA isn't fast enough, we use more than one to drive rows in parallel (back to SLI). We would need to synchronize the FPGAs at a higher rate than the refresh rate.

To achieve 4k 10-bit color at 10,000 hz, we would need to drive pixels at:
3840 width x 2160 height x 10,000 hz = 82.944 ghz
At 10 bits color, it would be pushing 3840 width x 2160 height x 3 color components x 10 bits per color x 10,000 hz = 2,488.32 gigabits per second

To bring everything down to a more reasonable 100 mhz 30 output pins FPGA, we would get a pixel rate of:
(30 pins / 10 bits per color / 3 color components) x 100 mhz = 1 million pixels
Then we could drive 1 million pixels / 2160 height = 463 rows per FPGA, or about 5 FPGAs for the whole display.

Of course this isn't a general purpose display and would only be suitable for demos that fit in a FPGA. But imagine the ufo test at 10,000 hz. Or maybe we could fit an entire game in a FPGA?

Did I math right?

[punpcklbw]

It should be possible for some special uses, but are there any actual prototypes built driven this way though?

All I can think of is those fast-rotating DIY LED displays like ones shown on the "Maker Electronics Projects" YT channel (can't post a link here due to restrictions)
To form an image, you have to switch the LED array really fast, on the order of thousands of Hz. Specialized LED drivers or a microcontroller with some amplifying circuitry could handle this.

I wouldn't dare to drive a display panel at these speeds for anything but experimental setups. The point is, if you're fast (and bright) enough, you might not even need an actual display panel to produce an image, a single scanline may be enough and it's easier to drive than a whole 2D display. OLEDs generally suck due to their low brightness and lifespan, so this approach requires micro LEDs or just bright SMD LEDs.