miller@rcf.rsmas.miami.edu (03/23/91)
As a followup to the Mac+ accelerator thread, I posted an article regarding the speed of terminal emulators (e.g., Versaterm Pro) and whether or not they would run faster on an accelerated Mac+. I stated that my Mac+ running Versaterm was much slower than a "real" terminal connected to the same line. After further (more careful) comparisons, I see that this is not true. I have listed a file containing 500 80-character lines from a mainframe to the Mac+'s screen and to a couple of terminals. Herewith is the data: TERMINAL CHARACTERS/SECOND DEC VT240 460 Human Designs HDS3200 889 MAC+ with Versaterm emulating a DEC VT220 615 So, there are two surprises here. First, the Mac+/Versaterm VT220 is *faster* than DEC's VT240 which is only about half as fast as a third party terminal. Second, all 3 are much slower than I would have guessed.... How many characters per second should my nominal 19.2k baud connection yield??? P.S. The numbers shown above remain the same regardless of whether I am doing the listing directly from the DEC VAX to which I am connected via a LAT server or whether i am doing the listing from a DEC ULTRIX machine across an ethernet. ------------------------------------------------------------------------------- Jerry L. Miller /\ Rosenstiel School of Marine and Atmospheric Science #\ o O \/ /\ University of Miami #/ /\/\/\/\/\/\/\/\/\ / 4600 Rickenbacker Cwsy. / / o O \/ Miami, FL 33149 USA Phone: (305) 361-4762 INTERNET: miller@rcf.rsmas.miami.edu
minich@unx2.ucc.okstate.edu (Robert Minich) (03/24/91)
by miller@rcf.rsmas.miami.edu: |As a followup to the Mac+ accelerator thread, I posted an article regarding |the speed of terminal emulators (e.g., Versaterm Pro) and whether or not they |would run faster on an accelerated Mac+. I stated that my Mac+ running |Versaterm was much slower than a "real" terminal connected to the same line. |After further (more careful) comparisons, I see that this is not true. | |I have listed a file containing 500 80-character lines from a mainframe to the |Mac+'s screen and to a couple of terminals. Herewith is the data: |TERMINAL CHARACTERS/SECOND |DEC VT240 460 |Human Designs HDS3200 889 |MAC+ with Versaterm emulating a DEC VT220 615 As far as scrolling speed, I have found that people here often perceive terminals that use jump scrolling as being faster than the same term with smooth scroll on, even though I've found that at the typical 2400bps rate here there is no difference except for very short line (approximately 20 characters or less.) I can actually read the stuff on my screen (when I'm stuck away from my Mac on a dumb term) but others have to do ^S and ^Q flow control... |So, there are two surprises here. First, the Mac+/Versaterm VT220 is *faster* |than DEC's VT240 which is only about half as fast as a third party terminal. |Second, all 3 are much slower than I would have guessed.... | |How many characters per second should my nominal 19.2k baud connection yield? | |P.S. The numbers shown above remain the same regardless of whether I am |doing the listing directly from the DEC VAX to which I am connected via |a LAT server or whether i am doing the listing from a DEC ULTRIX machine |across an ethernet. Some things you might try to test effects on transfer rates: 1) resize the window to a be short vertically 2) time an ascii file capture (assuming you can choose to NOT have the data echoed to the screen) I have a for-all-intents-and-purposes-but-not-really direct serial line that is 9600bps and I DO get text rates of ~950cps if the screens are dense or I minimize the scrolling that occurs by applying #1 above. If you find that scrolling is slowing down your term emulator, then an accelerator should help out. I find my SE/30 significantly better at keeping up with scrolling than SEs and Pluses, although all Macs do just fine on the serial ports for the data rates in question. Ideally, a graphics coprocessor would do all the work, but then Apple would have to change their philosophy on the Macintosh architecture. -- |_ /| | Robert Minich | |\'o.O' | Oklahoma State University| "I'm not discouraging others from using |=(___)= | minich@d.cs.okstate.edu | their power of the pen, but mine will | U | - "Ackphtth" | continue to do the crossword." M. Ho
tagreen@bronze.ucs.indiana.edu (Todd A. Green) (03/25/91)
In article <1991Mar22.155813.11505@rcf.rsmas.miami.edu> miller@rcf.rsmas.miami.edu writes: >As a followup to the Mac+ accelerator thread, I posted an article regarding >the speed of terminal emulators (e.g., Versaterm Pro) and whether or not they >would run faster on an accelerated Mac+. I stated that my Mac+ running >Versaterm was much slower than a "real" terminal connected to the same line. >After further (more careful) comparisons, I see that this is not true. Let's think of where the bottle necks are occuring here. Is it the speed of the CPU, or is it your modem, or is it how fast your machine can write to the screen or is it a combination of the 3 (or are there other factors). I can say (rather safely) that the speed of your modem is the deciding factor and that the speed of your CPU has very little to do with it, and the screen comes in the middle somewhere. (Try chaning your bit depth from 1 to 24 and watch the difference ). >I have listed a file containing 500 80-character lines from a mainframe to the >Mac+'s screen and to a couple of terminals. Herewith is the data: >TERMINAL CHARACTERS/SECOND >DEC VT240 460 >Human Designs HDS3200 889 >MAC+ with Versaterm emulating a DEC VT220 615 Yes but what you do not state is the connection speeds between the various terminals. I'll assume they are all the same for a fair comparison. >So, there are two surprises here. First, the Mac+/Versaterm VT220 is *faster* >than DEC's VT240 which is only about half as fast as a third party terminal. >Second, all 3 are much slower than I would have guessed.... > >How many characters per second should my nominal 19.2k baud connection yield??? Don't forget that you are NOT at 19.2K "baud" but rather 19.2K bps. (bits per second). There is a big difference. You're modem is running at 2,400 baud carrying 4 bits per signal (Most likely :) ) = 9,600 bps. Using compression such as MNP-5 you get throughput of 19.2K using 2:1 compression. V42.bis gives you 4:1 compression thus allowing throughput of 38.4k. Now, you must convert bps to characters per second. You cannot just divide by bps/8 to get cps (characters per second) You must realize that besides sending the data bits you also have the parity and stop bits. (Which add up to 10 on my configuration.) So, if you take the cps that you got 615 X 10 = 6150 bps. When I use my Hayes UltraSmartmodem 9600 to connect to IU's NEC 9631's (which are only MNP5 and not v.42 :( ) I usually get throughput of 924-980 cps. I've not taken the time to tweak my registers, so it could possibly get better. When I call Hayes 1-800 number I usually get throughput of 22-30K. Gotta love v42!! Hope this helps....if there are any errors forgive me I've only had 3 hours of sleep... > [junk deleted] Todd ============================================================================== Todd A. Green "<_CyberWolf_>" ---> Pascal <- tagreen@ucs.indiana.edu Unix Systems Administration ---> Unix <--- tagreen@silver.ucs.indiana.edu Macintosh Systems Administration ---> VMS <---- tagreen@bronze.ucs.indiana.edu WCC Office:136.04 phone:855-0949 ---> C <------ tagreen@lothario.ucs.indiana "Friends don't let friends ---> Mac <---- tagreen@iubacs.BITNET Use DOS" - Scott Ostrander ---> SunOS <-- tagreen@lykos (FTP only) ==============================================================================