How does bc work

[Bill Rugolsky Jr.]

On Wed, Oct 11, 2006 at 10:41:01AM -0400, Bill Rugolsky Jr. wrote:
> On Wed, Oct 11, 2006 at 09:30:13AM +0100, Dan Track wrote:
> > Hi
> > 
> > Does anybody know where bc stores its calucation while calculating a
> > number. Basically I'm running 36000000^36000000.
>  
> Rewriting it as (36^36000000) * (10 ^ 216000000)
> will save calculating, storing, and printing the 216M zeros,
> and will cut the first factor down to a "mere" 56M digits.

Just for kicks, I did this on a 4GB dual Opteron 246 running x86_64 Rawhide.
It took about 9 hours, and used about 200MB of VM (sorry, lost the output
of "time"), but the end result is:

(10656 ... <56026881 digits> ... 09376) * (10 ^ 216000000)

I wonder how the various other packages supporting multiple precision
(e.g., Pari) stack up.

Regards,

	Bill Rugolsky