redc_n

[Niall Emmart]

Hi everyone, I'm a grad student at the University of Massachusetts.  I work on high precision arithmetic for GPUs.I was looking at the implementation of the redc_n function and was blown away.   It's such a clever useof mpn_mulmod_bnm1.   From reading the docs / change log, it looks like the mpn_mulmod_bnm1 stuff can be used to accelerate other functions as well.    So, here's my question -- is there a main paper tocite for the origin of this idea (i.e. using mulmod_bnm1 to accelerate various computations)?  And further,is there a paper to cite for the specific use of mulmod_bnm1 in REDC?
Many thanks,Niall Emmart