Re: FDD speed and capacity

[mjmahon@aol.com (Michael J. Mahon)]

Allen Bong asked:

>Something had bugged me for a long time about what are the factors that
>control the speed of transferring a file to a FDD and the capacity of the
>floppy disk drive in DOS, ProDos and GS/OS.
>
>If Apple II was designed with a faster cpu, would it end up with a DOS
>system that transfers a large file to the FDD faster or a disk with a higher
>capacity?  As the file was written to FDD using the RWTS subroutine which
>uses software timing, does it determine the transfer rate?  What did
>Pronto-DOS and Derversi DOS do to make the transfer faster?

The data transfer is timed by software loops, but to the specifications of
the SA400 (single-density) drive, at 250Kbps.  Regardless of the speed of
the processor, single-density drives cannot transfer at a faster rate.

The faster _throughput_ of the disk drive under Pronto-DOS, DiversiDOS,
ProDOS, and many other "accelerated" DOS's is achieved by changing
the buffering scheme during reading so that less data copying has to
be done.  DOS 3.3 reads the nibbilized data into a buffer, converts it
to 8-bit data in place, and then moves the data to the destination address
in memory.  ProDOS (and the others) read (most) of the nibbles into the
final memory location, then translate them into the final data in place,
eliminating the copy.

>Which has the faster transfer speed ?   An Apple IIgs (2.8MHz) with 3.5"
>800K using GCR format disk or a 720K DSDD using MFM format running on a PIII
>1GHz windoz machine?

The raw data transfer rates are the same.

The invariance of I/O data rates in the face of exponentially increasing
processor
speed is what creates the "I/O bottleneck", a fundamental problem of computer
system design which must be solved to achieve system performance that
scales with processor speed.  (Lately, exponentially expanding memory sizes
and multi-level memory hierarchies have provided the solution for most business
systems.)

>If the 6502B was selling at $10 in 1976, Would Woz come out with a DOS that
>runs faster or a bigger capacity disk drive?

Although it is possible that RWTS could be re-designed to handle higher data
rates for a higher-density drive, the point is moot.  As data rates increase,
it
becomes much more advantageous to use more sophisticated hardware disk
controllers to achieve both higher density and higher speeds.

In the case of the Apple II, going to a higher memory/bus speed would have
been a fundamental change in the design of the machine, going far beyond the
clock speed of the processor.  This is why all processor "speedups" to the
Apple II must still preserve the 1MHz bus speed when interacting with memory
or peripherals.


-michael

 Email:  mjmahon@aol.com
 Home page:  http://members.aol.com/MJMahon/