Re: Wanted: a faster protocol for IEEE devices

On Sep 17, 2013, at 8:02 PM, gsteemso@commodore128.org wrote:

> On 2013/09/17 7:02 pm, Julian Perry wrote:
> [Jim Brain wrote, attribution did not survive:]
>>> d82 was created from 8250 disk in 12 minutes.
>> That's quite respectable, considering it's shovelling a meg of data.
> 
> I'm not sure I agree with that. That's not even 1.5 KB/s. Considering the IEEE-488 bus is capable of over 1MB/s with a faster machine driving it, I have to say I'm a little underwhelmed.
> 
> Anyone have any ideas as to how much of the sluggishness is due to the ancient and slow floppy drive mechanism? Of course, a good bit would have to be due to the inadequate buffering capacity, as well. Not sure about the 8250 without looking it up, but I seem to recall the 2040/4040 only has a couple of KiB on board. Those RAM chips cost a lot in the late 1970s.


If you look at the source to libd82copy/std.c, you'll see why it's slow. For every block, it does:

1. U1:2 0 track sector
2. read status
3. B-P2 0 (read block DOS)
4. talk, read actual data, untalk

You need a speeder routine like d64copy that at least reads blocks as they show up under the drive head, keeping track of sectors that have been seen. 

Even with handshaking, you should easily be able to get whole disk reads down to less than 5 minutes. With no head seeking or other delays, the minimum at 300 rpm is about 2 minutes.

-Nate


       Message was sent through the cbm-hackers mailing list