macmillan%wnre.aecl.cdn%ubc.CSNET@RELAY.CS.NET.UUCP (01/17/87)
1. Why does a SHOW TERM on a terminal connected to a DECserver have the characteristic "No DMA"? I thought the purpose of DECservers was to off-load character processing.. a. Is it really DMA, but show as non-DMA? b. OR are our DECservers connected incorrectly? 2. Has anyone noticed the VERY SLOW response of a VT240, running in Tektronix emulation, over a DECserver? Compare the speed of a terminal connected to a VAX 750 at 4800 b/s on a DZ-11, with an identical terminal connected to a VAX 8650 at 19200 b/s on a DECserver. The DECserver-connected terminal ran about 25% of the speed. ie. the DZ-11 connected terminal was 4 times as fast, when producing a plot in Tektronix 4010 mode. Perhaps #2 is related to #1. Note: The 750 and 8650 were lightly loaded. Response on each system was good. John MacMillan Atomic Energy of Canada Whiteshell Nuclear Research Pinawa, Manitoba, Canada R0E 1L0 (204) 753-2311 x2539
LEICHTER-JERRY@YALE.ARPA.UUCP (01/20/87)
1. Why does a SHOW TERM on a terminal connected to a DECserver have the characteristic "No DMA"? I thought the purpose of DECservers was to off-load character processing.. a. Is it really DMA, but show as non-DMA? b. OR are our DECservers connected incorrectly? c. Neither of the above. The DMA characteristic is only meaningful when applied to a hardware terminal line interface; it then says that this hard- ware interface can send bursts of characters out to a terminal via DMA. The hardware interface a terminal connects to on a DECServer is not really of con- cern to VMS; all VMS sees is an Ethernet interface that LTDRIVER sends the data to the terminals over. An Ethernet interface IS, indeed, a DMA device - but that fact isn't even visible to LTDRIVER, which doesn't talk directly to it, but uses an interface provided by some port driver like XEDRIVER. An LT device will ALWAYS show up as "No DMA". 2. Has anyone noticed the VERY SLOW response of a VT240, running in Tektronix emulation, over a DECserver? Compare the speed of a terminal connected to a VAX 750 at 4800 b/s on a DZ-11, with an identical terminal connected to a VAX 8650 at 19200 b/s on a DECserver. The DECserver-connected terminal ran about 25% of the speed. ie. the DZ-11 connected terminal was 4 times as fast, when producing a plot in Tektronix 4010 mode. I've never heard of this before. A quick guess is that the terminal at 19200 is sending a lot of XOFF's and XON's, resulting in unnecessary traffic. Try the following paradoxical approach: Run the DECserver-connected terminal at 4800 bps and see what happens. If that helps, try 9600. There is actually nothing to be gained by running at 19,200 bps. A VT240 can't process characters that fast - it will just use XON/XOFF to throttle the rate back to something it CAN handle, which is somewhere around 9600 bps, usually somewhat less. BTW, a VT220 is faster - it can't quite manage 19,200 bps either, but it CAN run faster than 9600 - though not by much. (The pre- ceeding conflates short-term line speed with long-term character throughput, but I don't want to get into the details here.) I'm assuming that, when you say it's "an identical terminal", that you've checked at that it really IS identical - same rev firmware (hit SETUP), save comm parameters (like XOFF at xxxx setting) except for speed, etc. Also, that the DECserver port is set up correctly - flow control enabled, local and switch characters set to something that won't appear in any reply data from the terminal, and so on. -- Jerry -------