inter-record gaps

[Don Speck]

    Occasionally one has to know how much data will fit on a
magtape, such as for tar or dump.  This is easy to calculate
from the blocksize, density, IRG, and useable tape length;
unfortunately, I frequently found these calculations giving
over-optimistic answers, wasting my time with tapes that ran
off the reel.

    So, I measured some tapes with various tape drives.  I
counted how many write()s it takes to get an EOT error, once
for large blocksize (10K) and again for small (1K).  This
gave me two equations in two unknowns (the interrecord gap
and tape length) to solve.

    Calculated tape lengths varied from 2315 to 2360 feet;
no doubt the leader and trailer account for this deduction
from the advertised length.  IRG's varied on streaming tape
drives depending on how much time I allowed between writes
(as the drive switched between its various modes).  Results:

					_____________mode_____________
    configuration	       density	start/stop  25-ips  100/75-ips
750 + DEC TU80			1600	0.84-0.87  0.66-0.73   0.85
750 + Emulex TC13 + CDC 92185	1600	0.74-0.78    0.67      0.70
750 + Emulex TC13 + CDC 92185	6250	   0.41      0.33	-
Sun 2/170 + Tapemaster + CDC	1600	0.75-0.88     - 	-
780 + TU77			1600	   0.73
780 + TU77			 800	   0.71

    Summary:  all of the drives seemed to need more than the
minimum allowed interrecord gap.  CDC streamers (including
the TU80 and the drive that comes with Suns) can need nearly
50% more than the minimum gap.	See the definition of "extended
short gap" mode in the CDC 92181/85 manuals.

	Don Speck	speck at cit-vax.arpa