oyster@uwmacc.UUCP (03/27/87)
In article <1592@tekigm2.TEK.COM> jimb@tekigm2.UUCP (Jim Boland) writes: > >If there isn't any idle time, >>the problem is most likely a terminal line which is passing transmit >>data to receive data, using lots of CPU time echoing the same thing >>over and over. ... >We traced it down to a noisy terminal line which had nothing connected at the >other end. Disconnecting unused terminal lines solved the problem. > >There was a software routine printed in DEC PROFESSIONAL(I believe) in >Fall of 85 that checks for terminal activity. I don't remember which >issue and my issues don't go back prior to Dec 1985. I remember the >article,though, and we tried the program. It helped to find our problem. That's the February 1986 issue (Volume 5, Number 2). The basic algorithm is: 1) Examine KB; if logged in, go to next KB, else continue. 2) Determine speed setting of KB. 3) Set speed to zero. 4) Time some arbitrary calculation & display result. 5) Set speed back to normal. 6) Repeat calculation & display results 7) Next KB The author printed results of a sample run where two terminals were causing problems; they ran the calculation at ~60% of the "zero speed." -- - Joel Plutchak, ex-RSTS system manager uucp: {allegra,ihnp4,seismo}!uwvax!uwmacc!oyster ARPA: oyster@unix.macc.wisc.edu BITNET: plutchak@wiscmacc Disclaimer: If subpoenaed, the U of W will disavow all knowledge of my actions.
jfh@killer.UUCP (04/02/87)
[ Line eater at work - munch - munch ] In article <1300@uwmacc.UUCP>, oyster@uwmacc.UUCP (Vicarious Oyster) writes: > In article <1592@tekigm2.TEK.COM> jimb@tekigm2.UUCP (Jim Boland) writes: > > >If there isn't any idle time, > >>the problem is most likely a terminal line which is passing transmit > >>data to receive data, using lots of CPU time echoing the same thing > >>over and over. > ... > >We traced it down to a noisy terminal line which had nothing connected at the > >other end. Disconnecting unused terminal lines solved the problem. When I worked for Pinnacle Systems, we had a number of customers complaining of the same problem - we had it real bad at our office because the serial cables run in the ceiling over noisy stuff like light fixtures. The best fix I ever found was to tie the RxD line to ground with a 10K resistor. If you do this on the computer end of the cable, you don't have to disconnect the unused (or switched off, like with a T-switch) cables. - john. (jfh@killer.UUCP) John F. Haugh II Ex Technical Support Analyst, Pinnacle Systems, Inc. Disclaimer: I don't work for Pinnacle anymore, much less speak for them.