Date   : Tue, 24 Mar 1981203:00:00-MST
From   : DAN@MIT-ML
Subject: Speedup patch for Morrow Designs Disk Jockey 2D Board

While looking through my Disk Jockey 2D manual, I discovered an interesting
"feature" in the FIRMB software which is a disadvantage to those who have
fast (3 ms track-to-track) single-sided disk drives.  In the "PREP" routine
(around 0E254H or so in the standard FIRMB software; it depends on the
FIRMB version that you have) is the code:

       .
       .
       .

       LDA     DSTAT           ;get the double
       ANI     DSIDE           ;-sided flag
       STA     DSFLAG          ;save for status
       RAR                     ;shift for
       RAR                     ;3/6 ms step
       RAR                     ;-rate constant
       ADI     SKCMD           ;do a

       .
       .
       .

This code checks to see if a single sided or double sided disk is being
used by isolating b3 of the DSTAT flag byte.  If b3 is "1", then the disk
is single-sided.  This is shifted over to b0, and is added to the Seek
Command value (18H) to form the final 3 millisecond Track-to-Track (TTT)
Seek Command of 18H, or the 6 ms TTT Seek Command of 19H.

Unfortunately, if you are using single-sided 3 ms TTT drives (such as
the Remex drives that Morrow supplies), you don't get the advantage of
the speedier seek time because the above software assumes that all single
sided drives (e.g. older Shugarts) can't hack 3 ms TTT.

If you have 3 ms TTT drives, you can modify your FIRBM software to
use a 3 ms TTT time by replacing the three "RAR" instructions with:

       XRA     A               ;always use
       NOP                     ;3 ms track-
       NOP                     ;to-track time

Of course, this will require a new EPROM to be burned.  Be careful, as
the EPROM on the DJ2D board has INVERTED data (i.e. logical TRUE = 0,
logical FALSE = 1).

After making this patch, I did some disk-intensive benchmarks so see if
this improved performance.  Depending on what I did (Z80 assembly, Pascal
compilation, etc.), I got between 5% to 17% speed improvement, which is not
bad for changing three bytes of code.

                                                       - Dan