@AmpleApricots have you ever used a vector monitor? It's niche as hell but it needs to be seen because it's revelatory, a real holy shit moment.
Old oscilloscopes used these technologies, then there's also the
Tectronix 4010 which is a terminal that does vector graphics. Little known thing - the terminal emulator xterm for linux to this day has a tectronix emulation mode that for example can be seen in action with
gnuplot. These things are a bit before my computing time though. Tectronix emulation was quite widespread in some Amiga, DOS, and Mac terminal software and that's where I saw it most really, which was more for legacy reasons and the fact that these computers were cheaper to aquire than a tectronix than anything else. Interesting thing about the phosphor in the genuine Tectronix screens is that it only needs to be hit by the ray gun once and then will glow for a relatively long time. (vs. many refreshes with the normal phosphor) The reasoning for this once again was technical necessity - if you have a raster graphics based display with a high refresh, you need memory as screen buffer to put the pixels into and memory used to be easily the most complicated (read: expensive) part of modern computer technology. If you could eek away with not needing that buffer, you could save quite a bit of money. Also at least for a while after drawing, the graphics were very sharp and wouldn't flicker because there was no refresh per se and you just looked at glowing phosphor, which is easier on the eyes. You also could go really high-res and intricate which would've been prohibitively expensive with anything raster-based because of the memory required. That's why they were very popular for CAD stations for example for doing PCB design which lends to that whole vector graphics thing. Downside is, screen burn-in and also the lines would eventually get blurrier, you needed a screen refresh then. The screen refresh would include erasing the entire screen first and then redrawing it, which means you couldn't use this for moving pictures. Where did you see one?
In a wider and only somewhat related sense, there's also a gaming console doing vector-based graphics, the
Vectrex. It wasn't popular in my corner (is a brit thing) of the world but I actually saw one. Something the youtube videos on this will not be able to show you is how smooth the output of these technologies is when doing animation. We kinda lost smooth, stable output in the last thirty years and it doesn't look like it'll ever be back. (The vectrex didn't use any fancy phosphor or special tubes though, it was all in the control circuity) I think using the technology in this console was inspired by the Asteroids arcade game, but I frankly don't really know.
TV that didn't have billions of tiny dots comprising its screen
Since I got really interested in small high-DPI screens (everything 220+ PPI) I'd say it comes close enough. A lot of problems you have in lower DPI screens (visible pixels, aliasing etc.) aren't really a problem anymore at these pixel densities because while still technically "there", these effects are just too tiny to be visible to the naked eye. It's surprisingly difficult to buy such screens though. Not that they're expensive (and they are) they're just actually rare. It is almost like screen manufacturers decided 94-100 PPI is enough because that's where most screens lie. It's probably the most cost-effective density is my guess. A 16:10 or 3:2 oled monitor with 250+ PPI of max 14-16" would be my dream, but such a thing literally does not exist as finished, stand-alone product.
The closest to the tectronix storage tube idea of just storing the picture inside the screen is eink is my guess. While inherently raster-based, it also has this memory effect just with eink it doesn't fade. Also some modern eink screens are really fast with erasing/refreshing and can reach framerates of up to 40 FPS.
GPUs themselves might have ways to be more efficient if they were returning drawing instructions rather than said bitmaps
Old GPUs (think ISA vintage) already had "GUI acceleration" which basically was simple line/box/filled box etc. drawing functions, so that the CPU didn't have to do it all itself and push it to the graphics cards memory pixel-per-pixel. Because everything in the PC world happened bitmap-based already back then it just rarely was effectively used. I can't imagine it'd matter much nowadays, at any rate. An e.g. 16000x16000 picture (that's 256 megapixels!) would be a 768 mb 8-bit BMP. For 1979 that'd be an insane size, for today - meh.