Re: ADTPro 1.1.2 Released

["Michael J. Mahon" <mjmahon@aol.com>]

Polymorph wrote:
>  To: Michael J. Mahon
> Michael J. Mahon wrote:
>
> > Polymorph wrote:
> >
> > > On Feb 20, 12:09 pm, "Michael J. Mahon" <mjma...@aol.com> wrote:
> > >
> > > > Eric Rucker wrote:
>
>
> <snip>
>
> > In all cases, the total transfer time and the total actual write time
> > are the same (except for minor and uncontrollable rotational latency
> > differences at the start of each write).
>
> Yes, it was the disk I/O latency that I was referring to. By reducing 
> the number of writes, I expect some modest reduction in the overall time 
> it takes to transfer a file/image from one machine to another (transfer 
> + write time).
>
> When I said "How much of an increase I'm not sure", I meant I do not 
> know exactly what the latency is, but I expect a few seconds should be 
> shaved off over the course of a 140K transfer, and even more for bigger 
> files/images (I do not want to limit my project to image transfers). The 
> latency will depend largely on what medium/device you're writing to - a 
> hard disk will have a much smaller latency than a floppy disk.

Think of it this way:  once a write starts, there is an optional startup
delay of about a second (if the disk has stopped), followed by a latency
of an average of 1/2 rotation or 0.1sec, followed by the write time,
which will be the same for a whole disk, regardless of buffer size.

Since startup delay is about 10x rotational latency, the latency effects
will be second-order, and the number of startup delays will dominate.

In any case, the time dofference should be at most N-1 seconds, where
the buffer size results in N startup delays.

So it's probably not a big deal.

-michael

NadaNet and AppleCrate II: parallel computing for Apple II computers!
Home page: http://home.comcast.net/~mjmahon

"The wastebasket is our most important design
tool--and it's seriously underused."