Re: Recode to Play MP3?

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

mdj wrote:
> On May 28, 5:26 pm, "Michael J. Mahon" <mjma...@aol.com> wrote:
>
>
> > Since you move the data back and forth, the page last written is
> > the page next read--perfect.  Moving from page 0 to a page 0 mod 8K
> > away conflicts perfectly, so if data were *repeatedly* moved from
> > page 0, but not back and forth, every load would be a cache miss!
> >
> > Doing that experiment, with the second copy loop changed to re-copy
> > page 0 to $20xx, results in a time of 8.1sec--almost twice as long.
> >
> > In this case, the cache only benefits code fetching, not data access.
> > But since most 6502 activity is code fetching, it's still pretty fast!
>
>
> 8.1 sec is a big drop. It makes no difference at all to the Transwarp
> (except for the aforementioned speed drop writing to video areas).

It simply corresponds to causing every LDA instruction to miss
on its data reference, resulting in a bus cycle penalty per
iteration, or the equivalent of 8 Zip cycles.

> I would suppose that this would take considerably longer at 4Mhz, so
> my feeling that the Transwarp is slightly faster for block copies than
> the Zip 4Mhz probably holds.

At 4MHz the penalty would be only 4 Zip cycles, so it would actually
perform better than linear extrapolation would predict--about halfway
between the cache hit number and the cache miss number, times two.

I'd estimate about 12.4sec., a little bit slower than the TW at
3.58MHz.

Cache missing on every data reference is, however, quite a worst-case
scenario, almost never seen in real programs.

Usually, the Zip Chip provides a speedup of about 5x.

> One thing is for sure, any delusions I had about a double-speed
> Transwarp being the ideal accelerator performance wise have well and
> truly evaporated :-)

;-)

-michael

NadaNet file server for Apple II computers!
Home page:  http://members.aol.com/MJMahon/

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