As of December 2021, it seems that numpy on macOS/M1 doesn't support quad-precision floating point
numpy