
 ----====== HUGE_FLOAT ver 20-august-1997 ======---- 
    author: Joerg Arndt, email: arndt@jjj.de 
 compiled using GNU C++ version "2.7.2.1"
 at Feb  8 1999, 13:58:57
    hfloat is online at http://www.jjj.de/hfloat/
    (check my homepage for the latest version of hfloat) 
 radix set to 10000

 size of workspace is 32768 bytes  (maxsize= 4194304 bytes) 
 max precision set to 1024 LIMBs =4096 decimal digits   =13606 bits 
 set mul_convolution to FHT_CNVL
 set sqr_convolution to FHT_CNVL

 usage: ../bin/pi.exe ldn [alg [radix]]
   (or  ../bin/pi.exe --help)
   ldn is the (base 2) log of the desired precision in digits of base <radix>
   alg   (optional) is the desired algorithm (see list below) 
   radix (optional) gives radix, e.g. 10, 16, 65536, 10000 (=def.)

   e.g.: pi 11 3 100 means:
   compute pi to 2048 x 2 = 4096 digits with Gauss-AGM algorithm

   0 => pi: Borwein's quartic (r=4)    10 => pi: AGM cubic
   1 => pi: Borwein's quartic (r=16)   13 => pi: AGM quartic
   2 => pi: Borwein's quadratic        14 => pi: AGM3, quartic fast
   3 => pi: Gauss-AGM                  15 => pi: AGM3, quartic slower
   4 => pi: AGM3 fast                  5N => pi: N-term arctan series (N=2...7)
   5 => pi: AGM3 slower               100 => e : (2.7182818...) via exp(1)   
   6 => pi: Borwein's Derived AGM     101 => e : (2.7182818...) via series
   7 => pi: Borwein's quintic         111 => log(2):  (0.693147...)
   8 => pi: Borwein's cubic           163 => exp(pi*sqrt(163)) 
   9 => pi: Borwein's nonic           999 => (your) stuff in yourstuf.cc
