ADTPro performance

["schmidtd" <schmidtd@my-deja.com>]

My other thread kinda got hijacked, so I'll post some quick results
over here. :-)

I was curious to see how the ProDOS-ification of ADT would affect
performance.  The meat of where the code spends its time - data
transfer, CRC algorithm, etc. are very much the same between the two
variants.  The main differences are the underlying OS (DOS vs. ProDOS),
a little difference in the buffer size (28k for ADT vs. 20k for
ADTPro), and of course my own clumsy, baroque assembly skills.

The following are timings I took for a few different scenarios.  My
conclusions are that ProDOS outperforms DOS (really ESDOS) on
later-model hardware, and that the transfer buffer size probably
doesn't have nearly as much bearing on performance as the DOS/driver
combination does.  I originally started goofing with DOS ADT to see if
I could squeeze better performance out of it by doing a disk transfer
in 4 passes instead of 5.  I could never quite get it to fit, but I
think the results would ultimately have been disappointing.  Anyway, on
to the results:

P = ADTPro 0.0.4
A = ADT 1.23
All transfers were conducted at 115,200 rate.

Writing speed, in minutes
-------------------------

Apple ][+, 5-1/4" Disk ][
P 1:21.7
A 0:38.9

IIgs ROM03, 5-1/4" A9M0107 Disk
P 0:28.5
A 0:32.4

IIgs ROM03, 3-1/2" 800k Disk
P 2:41.7


Reading speed, in minutes
-------------------------

Apple ][+, 5-1/4" Disk ][
P 0:34.8
A 0:33.7

IIgs ROM03, 5-1/4" A9M0107 Disk
P 0:26.6
A 0:29.7

IIgs ROM03, 3-1/2" 800k Disk
P 2:12.2

IIgs ROM03, CFFA 32MB partition (fairly sparse)
P 53:23.0

5-1/4 diskette was the ADTPro 0.0.4 distribution .DSK (fairly sparse)
3-1/2 diskette was the GS/OS 6.0.1 Install disk (farily full)