> It should take about:
> 
>     2 passes * (101632 * 1024 kbytes / 13884 kbytes/sec) /pass
>          = 14990 sec
>    or  about    4 hours 10 min.
>   + some time to rewind and stop/start at filemarks.
> 
> The stop/start at filemarks is what makes it slower than expected.  The 
> more files it has to write, the slower.
> The important parameter is a good guess of the length:
> 
>     amtatype -s 100g
> 
> would do it fine.  The program writes 100 files of 1 Gbyte each.  If you 
> let it the default size estimate of 1Gbyte, then the program will write 
> files of 100 Mbyte hoping it will have about 100 when hitting end of 
> tape.  But now it needs to write 1000 files.  The second pass writes 
> files of half the size, and will need to write 2000 files.  Now the 
> start/stop time becomes sigificant in the total runtime, and the program 
> becomes slow.
> 
> Paul

Paul,

That makes perfect sense, because in the beginning, I was using small
file sizes, thinking that I would get a better resolution (so that I
could squeeze a few more bytes out of the end of the tape)... but it
turned out backfiring on me considering it was taking days to complete.

Whether the file size would have had any bearing at all on the amount of
unclaimed space at the end of the tape is unknown to me, perhaps it
doesn't matter.  Maybe you know the answer to that.

As for taking one hour, I must be remembering wrong.

-Dave