On Tue, 2003-01-14 at 13:56, Jon LaBadie wrote:
> Because something is wrong or you have a very old version of tapetype.
> 
> Your drive specs will include a speed rating, in MB/sec probably.
> Divide that into your tape's rated capacity to get an approximation
> of how long it will take to write a complete tape.  Tapetype should
> take about twice as long as it makes 2 passes.
> 
> Be sure in your calculation that you use matching speed and capacity
> numbers.  Matching in the sense of both refering to HW compression on
> or both refering to HW compression off.

Jon,
I am using the tapetype that came with 2.4.2p2 (from the Debian
package).

I am using GNU mt 2.5, which does not have a "comp" command like the BSD
version.  The closest I can see is "datcompression", but this is not a
DAT drive.  Nevertheless, querying my drive with:

# mt -f /dev/nst0 datcompression

yielded:

Compression on.
Compression capable.
Decompression capable.

I set datcompression off, but I don't know what effect that will have
since this really isn't DAT drive.

The Compaq SDLT320 is rated at 16 MB/s native, but with SDLT 110/220
tapes, I'm assuming the speed will drop back to that of the SDLT220
drive, which is 11 MB/s.

So with a 110GB native capacity (112640 MB; 1GB=1024MB), that computes
to 10240 seconds (2h 50m) to write a full tape with no H/W compression. 
Doubling that for two passes means tapetype *should* finish in 5h 40m.  

However, the first pass of tapetype took 304793 seconds to write
"3513152 32Kb blocks in 103328 files", which is a paltry 368 KB/s.  That
is a large discrepancy in expected vs. actual performance.

Eric