problems/bugs

platform: windows XP, athlon64 X2 4400+.

problem 1:
i seem to be unable to produce more than a 31 bits collision.
this is with hashcash 1.22 (latest windows version)

i have been running hashcash for days attempting to create a large stamp.
dual core, 2 processes, running with -P (progress). it never showed more than 31 
bits found in the progress. i suspected this, and ran a hashcash to produce a 32 
bits stamp. it got to 541% without finishing and i stopped it. a friend of mine 
randomly either has this problem, or did not have it. i think it depends on the 
minting core.
----------------------
problem 2:
the benchmark goes wrong.

D:\temp\hashcash>hashcash -sv
     Rate  Name (* machine default)

   2445103 SHA1 library (hashcash)
     Solution:   1:22:040404:[email protected]::0123456789abcdef:000000Tu1n
     Iterations: 7831665
     Time taken: 3.203

   3231985 ANSI Compact 1-pipe
     Solution:   1:22:040404:[email protected]::0123456789abcdef:0000033LsQ
     Iterations: 51207578
     Time taken: 15.844

ERROR!    ANSI Standard 1-pipe
     Wanted 22 bits, reported 28 bits, got 0 bits.
     End-of-block marker remains intact.
     "1:22:040404:[email protected]::0123456789abcdef:0000000000"
     Time taken: 337.121

ERROR!    ANSI Ultra-Compact 1-pipe
     Wanted 22 bits, reported 28 bits, got 0 bits.
     End-of-block marker remains intact.
     "1:22:040404:[email protected]::0123456789abcdef:0000000000"
     Time taken: 529.824

ERROR!    ANSI Compact 2-pipe
     Wanted 22 bits, reported 28 bits, got 0 bits.
     End-of-block marker remains intact.
     "1:22:040404:[email protected]::0123456789abcdef:0000000000"
     Time taken: 295.669

ERROR!    ANSI Standard 2-pipe
     Wanted 22 bits, reported 28 bits, got 0 bits.
     End-of-block marker remains intact.
     "1:22:040404:[email protected]::0123456789abcdef:0000000000"
     Time taken: 355.812

    ---    PowerPC Altivec Standard 1x4-pipe  (Not available on this machine)
    ---    PowerPC Altivec Compact 2x4-pipe  (Not available on this machine)
    ---    PowerPC Altivec Standard 2x4-pipe  (Not available on this machine)
ERROR!    AMD64/x86 MMX Compact 1x2-pipe
     Wanted 22 bits, reported 28 bits, got 0 bits.
     End-of-block marker remains intact.
     "1:22:040404:[email protected]::0123456789abcdef:0000000000"
     Time taken: 270.480

ERROR!    AMD64/x86 MMX Standard 1x2-pipe
     Wanted 22 bits, reported 28 bits, got 0 bits.
     End-of-block marker remains intact.
     "1:22:040404:[email protected]::0123456789abcdef:0000000000"
     Time taken: 232.172

Best minter: ANSI Compact 1-pipe (3231985 hashes/sec)
Projected average times to mint:
   8 bits:     0.000 seconds (79.2 microseconds)
  10 bits:     0.000 seconds (316.8 microseconds)
  16 bits:     0.020 seconds
  20 bits:     0.324 seconds
  22 bits:     1.298 seconds
  24 bits:     5.191 seconds
  26 bits:    20.764 seconds
  28 bits:    83.056 seconds
  30 bits:   332.224 seconds (5.5 minutes)

---------------------------
problem 3:

this friend tried hashcash and got this problem. i can reproduce it as well. it 
suddenly started to report like 4 28 bits collisions/sec, giving the output 
below. i was able to reproduce this as well. the reported size differs; it 
reports the biggest collision it has found when it reaches 100 million attempts, 
and is likely actually finding 20 bits collisions.

C:\>hashcash -Pmb96 foo -Z1
percent: 0/79228162514264338000000000000 = 0.0000000000000% [0/96 bits]
percent: 69696/79228162514264338000000000000 = 0.0000000000000% [17/96 bits]
percent: 331840/79228162514264338000000000000 = 0.0000000000000% [18/96 bits]
percent: 725056/79228162514264338000000000000 = 0.0000000000000% [19/96 bits]
percent: 921664/79228162514264338000000000000 = 0.0000000000000% [20/96 bits]
percent: 1118272/79228162514264338000000000000 = 0.0000000000000% [21/96 bits]
percent: 3936320/79228162514264338000000000000 = 0.0000000000000% [26/96 bits]
percent: 60756032/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
percent: 100274240/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
percent: 100667456/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
percent: 101060672/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
percent: 101453888/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
percent: 101847104/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
percent: 102240320/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
percent: 102633536/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
percent: 103026752/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
percent: 103419968/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
percent: 103813184/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
percent: 104206400/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
percent: 104599616/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
percent: 104992832/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
percent: 105386048/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
percent: 105779264/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
percent: 106172480/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
percent: 106565696/79228162514264338000000000000 = 0.0000000000000% [28/96 bits]
....

----------------------
problem 4: on windows, for me, hashcash takes a second to do any work. including 
generating a 10 bits collision. on linux it's fast. this can be a problem if 
it's in a library, and it's going to add this delay when calling it from email 
clients and the like.