Comment by saagarjha

Supercomputers have very different needs. They want 64-bit floating point which nobody has been focusing on for a while