> There is only a 1.7% speed reduction for
> read_sector_address(dword in
> address). We'll use this one because of it's advantage for
> MSD, and we
> don't need to use start_read/stop_read.

That's fine :-)
 
> PS.. Just for kicks... if you put a constant in the array
> locations
> and forget about the "for loop" you can get the bits one by
> one. You
> add 4kb to your program space and get a speed of:
> 1,590,272

Yes, I've thought of that one ;-) Maybe there could be a constant 
PATA_HD_PIO_TURBO=TRUE or the like to use the codespace wasting variant, but it 
can be added any time, no hurry. We would have to closely watch the timing, 
with higher clock frequencies the code has to be blown up even more with nop's. 
And, there would be the question if such a lump of code should be there twice, 
as we would need the same amount of code space for turbo writing.

> I don't think your script is that accurate however (maybe
> it's my OSC
> speed). My scope shows about 2.4MBps!

I haven't checked with the scope yet... My first goal has been to distinct fast 
from slow algorithms, in fact I did not care to much about the absolute values. 
As you used a different clock frequency, did you change the hardcoded 
TMR0-preset in the interrupt procedure? If you had not, the readings however 
should have been greater than expected, not smaller... Is TMR0 in your device a 
16 bit timer?

Greets,
Kiste




-- 
You received this message because you are subscribed to the Google Groups 
"jallib" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/jallib?hl=en.

Reply via email to