GMP 20 years
dclarke at blastwave.org
Thu Mar 24 11:13:58 CET 2011
> The first change log in GMP is dated 1991-03-03, and the initial public
> release happened in September of the same year.
> So finally, GMP is no longer a teenager!
> I wrote GMP because there were several bignum libraries of questionable
> quality at the time, and many research projects in computational number
> theory or computational algebra started with writing a bignum library.
> I wanted to create a library that would allow researchers to start at a
> higher level, and then also give them better performance.
> Back in these days, a 32-bit multiply instruction took from around 20
> cycles to around 40 cycles, and CPU clocks were about as many MHz as we
> have GHz today. Today a 64-bit multiply instruction takes just 2 cycles
> on the best processors. All-in-all, hardware developments have allowed
> GMP to be perhaps 40000 times faster.
> Today GMP has become used in more places than I could imagine back in
> 1991. The usage has grown, but as with any software project, source
> code has also grown substantially--the initial release's tar file was 50
> kbyte while the latest release's tar file was 13 Mbyte.
> About 25 people have contributed to GMP directly over the years--but
> thousands of people have contributed to GMP indirectly by submitting bug
> reports. Thank you all!
> Now, let's look forward to another 20 years of GMP development!
> Extrapolating, it will have become 3 Gbyte of source code at the end of
> that period...
If no one has stepped up to the plate, let me say "thank you" from the
bottom of my very geeky heart which admires truely inspired code.
dclarke at opensolaris.ca <- Email related to the open source Solaris
dclarke at blastwave.org <- Email related to open source for Solaris
ps: I'm surprised that the first release was not in Fortran IV :-)
More information about the gmp-discuss