zlacker

>>BirAda+(OP)
Nice trip down memory lane. I still remember when I popped the 3dfx in and played quake, after playing quake on some ATI or Matrox I had (and a hell of a lot of other games before that lol). It was a transformative experience. I was stunned at how smooth everything was. It was beautiful. It was more incredible than going from 320x200x16(colors) to 640x480x256 and then 1024x768x16.8M, which were all quite marvelous increments. I think Moore's Law was just more visible in the early days. You really felt each iterative change. Going from "PC Speaker" to an Adlib was also a massive transformation.

>>cronix+p4
Interestingly, the transistor density of GPUs has been following a roughly logarithmic curve since 2000, compared to the linear increase in x86 processors [1].

I totally agree that the incremental innovations observed in earlier GPU platforms felt much, much more ‘obvious’ though.

It’s as if the ‘wow factor’ of graphics hardware doesn’t scale at the same rate as density.

Or perhaps releases were more spread out than they are today (compared to the annual release cycle expected today) making the jumps more obvious.

[1] https://www.researchgate.net/figure/Comparison-of-NVIDIA-gra...

>>greggs+Q5
I wonder if part of that is because GPU tasks are ridiculously parallelisable?

If you can split the screen into 64 equal chunks, there's nothing except silicon real estate stopping you splitting it into 128, or 256, or 2048. Think about how SLI worked, in the Voodoo II olden days.