search

hpc-carpentry / coordination

Delocalized issues relevant to the HPC Carpentry organization overall
https://hpc-carpentry.github.io/
6 stars 7 forks source link

Spinoff "Pi" calculation to hpc-parallel novice #103

Closed wirawan0 closed 1 year ago

wirawan0 commented 2 years ago

This is per 2022-02-17 discussion on

https://codimd.carpentries.org/eT62gOnGTquEfaqLozRB0g?view

There was this mention:

There is a negative review at http://pracrand.sourceforge.net/PractRand.txt of https://csrc.nist.gov/Projects/Random-Bit-Generation/Documentation-and-Software - Would be good to discuss inclusion of this kind of material in the lessons.

I would like to know what the "negative review" mentioned in these two links. I read them, couldn't make out what this negative review is. Except probably the unreliability of doing "just any" random number generators for highly scalable parallel computation. Which is really true and will bite people who scale up to many many number of threads or processes w/ independent random streams.

bkmgit commented 2 years ago

The most used statistical tests currently are http://pracrand.sourceforge.net/ and TestU01. The NIST tests seem not to be as demanding.