dillon@PAVEPAWS.BERKELEY.EDU (Matt Dillon) (05/31/86)
I wrote a little test program for the serial port... it writes 128 bytes (0-127), and reads them back.... I wanted to see how fast the serial port would go without loosing characters (had RAD_BOOGIE set as well). No protocols, NO 7-wire. Well, I set the thing to 292000 baud, and it worked! (8-bit, no parity, 1 stop) = 10 bit transmission = 29200 bytes/sec, or one byte every 34.2 uS. Did I make a mistake, or can the serial port *really* receive that fast without loosing characters. The 128-byte packets were mere blips on the RD and SD L.E.D.'s for my modem (in local echo mode). Even with a loaded system there didn't seem to be any problem. Whether I'm right or wrong, I would like to complement the designers of the Amiga for the excellent serial port! Here's a technical question (please don't answer unless you really know the answer... no answer is better than the wrong answer): What is the overhead for interrupt service routines (say, the first service routine in a given chain)? -Matt
neil@amiga.UUCP (Neil Katin) (06/03/86)
In article <8605311901.AA02278@pavepaws> dillon@PAVEPAWS.BERKELEY.EDU (Matt Dillon) writes: > Here's a technical question (please don't answer unless you really >know the answer... no answer is better than the wrong answer): What is >the overhead for interrupt service routines (say, the first service routine >in a given chain)? > > -Matt These are the timings that I made a while ago for the 1.2 release. They are significantly faster than for 1.1 (about 20%) time from 68K interrupt acknowledge to processing of int handler: 33.5 time from int handler to first int server: 14 (only applies if this int is processed via an int server chain. Serial ints are NOT processed this way, but I thought I'd give the time anyway. The times are in microseconds. Hope this helps. Neil