Ask HN: Moore's Law historical data but in FLOPS?

I'm curious what the historical growth of computing looks like if you count the total compute of all operational processors. Has anyone come across research that attempts to estimate this? I assume this would be far more exponential than Moore's law