[Coco] 16550 wasRe: RS232 paks

[Mark Marlette]

Gene,

I am glad that your memory has returned on this. :)

Darn it, being arrogant...Again......

Regards,

Mark
Cloud-9


At 3/5/2009 10:54 PM, you wrote:

>On Thursday 05 March 2009, Bob Devries wrote:

> >Of course, I have no idea exactly what the THROUGHPUT was in CPS with that

> >setup.

> >

>When I was working on rzsz-3.36 last, the optimized version everybody should

>have by now, using hardware flow control, could transmit to an amiga 

>at around

>740 cps with both sides set to 9600 baud.  In fairness to the flow control

>problems in the 6551, the amiga could of course keep up with the coco so the

>flow control never activated to trigger the bug under those 

>conditions.  So of

>course that bug never stuck up its hand in my tests.  That would only show up

>when the target machine was slower than the coco.

>

>Conversely, with the amiga sending to the coco, the flow controls 

>were used to

>slow the amiga down, and again the average speed was in the 735 cps 

>area.  And

>it was sustained over files of nominally a megabyte.  Only 

>noticeable slowdown

>was those created by the no-halt controller if I was writing to a floppy, but

>usually I used my hard drive or the ramdisk I wrote since my coco3 

>has a 2 meg

>kit in it and I can instantly (a few hundred milliseconds) set it up at up to

>about 1.7 megs max but usually for 1.5 megs.

>

>x and y modem, with their much simpler and faster error checking, 

>can run lots

>faster than that.  The speed killer in rzsz is the crc checking, 

>done on a per

>byte basis as the byte comes in or goes out.  Someday, someone should convert

>that to checking just one 256 byte buffer at once, it should at least triple

>rzsz's speed in cps.

>

>Those speeds above were obtained with the crc table lookup method.  Using

>rzsz's original crc method slowed it by a bit more than half, to around 330

>cps both ways, so that table lookup was a huge improvement right there.  Now

>it needs to be converted to doing that loop once, over a full or partial

>buffer.  I believe that would get it nicely above 960 cps (9600 baud).

>

>Anyway, those are the figures to shoot at, have fun, the code _is_ out there.

>

>--

>Cheers, Gene

>"There are four boxes to be used in defense of liberty:

>  soap, ballot, jury, and ammo. Please use in that order."

>-Ed Howdershelt (Author)

><``Erik> 18,446,744,073,709,551,616 is a big number

>

>

>--

>Coco mailing list

>Coco at maltedmedia.com

>http://five.pairlist.net/mailman/listinfo/coco

>

>No virus found in this incoming message.

>Checked by AVG - www.avg.com

>Version: 8.0.237 / Virus Database: 270.11.8/1985 - Release Date: 

>03/05/09 07:54:00