brucek@emperor.scs.com (Bruce Klein) (06/19/91)
How are you people achieving this? I have a number of Telebit modems here. For this experiment, I called a T2500 with a T1000 using the PEP protocol. The modems were connected to their hosts through 9600 baud serial ports. Using xmodem and the telebit xmodem support (s111=20) I was able to achieve 500 cps. Next I tried zmodem. With a Sun on one end and a Mac Plus running Zterm on the other (same two modems) I was able to achieve peak thruput of 700 cps. I was unable to sustain this though, and on average it delivered about 400 cps. There may be something wrong with my sz binary--on EVERY file it produces a CRC error at bytes 6144, 13312, 20480, 27648, .... (Anyone have any ideas about that?) Are these figures reasonable? If not, what could be the problem? The two phone lines in this instance were part of the same Centrex system, and thus should be of good quality. Help! -- Bruce Klein brucek@emperor.sjc.mentorg.com ... uunet!emperor!brucek
gandrews@netcom.COM (Greg Andrews) (06/19/91)
In article <1991Jun19.003439.7587@emperor.scs.com> brucek@emperor.scs.com (Bruce Klein) writes: > >How are you people achieving this? > >I have a number of Telebit modems here. For this experiment, I called >a T2500 with a T1000 using the PEP protocol. The modems were connected >to their hosts through 9600 baud serial ports. > >Using xmodem and the telebit xmodem support (s111=20) I was able to achieve >500 cps. > I regularly get 900+ cps with xmodem spoofing through T1000s. I wonder if your modems were REALLY getting the spoofing negotiated... Just turning on a register doesn't do the trick; the modems must negotiate the feature properly or it won't get turned on. > >Next I tried zmodem. With a Sun on one end and a Mac Plus running Zterm >on the other (same two modems) I was able to achieve peak thruput of 700 >cps. I was unable to sustain this though, and on average it delivered >about 400 cps. There may be something wrong with my sz binary--on EVERY >file it produces a CRC error at bytes 6144, 13312, 20480, 27648, .... >(Anyone have any ideas about that?) > Errors in a streaming Zmodem transfer usually point to a flow control problem. The problem could be at your end of the signal chain, or at the Sun's end. Every error caused by bad flow control means data must be transmitted again - slowing down the average throughput numbers. You should also make sure that the modem's spoofing is turned OFF for zmodem transfers. Otherwise, spoofing can cause interference. > >Are these figures reasonable? If not, what could be the problem? The >two phone lines in this instance were part of the same Centrex system, >and thus should be of good quality. > Getting throughputs higher than 950-960 cps isn't possible with a T1000 (unless your transfer protocol performs compression). That's because the T1000's PEP mode is limited to 9600 bps throughput. To get higher rates, you need to use full-speed PEP modems like the T2500, TB+, or T2000. -- .------------------------------------------------------------------------. | Greg Andrews | UUCP: {apple,amdahl,claris}!netcom!gandrews | | | Internet: gandrews@netcom.COM | `------------------------------------------------------------------------'
johnv@tower.actrix.gen.nz (John Veldthuis) (06/20/91)
Quoted from <1991Jun19.003439.7587@emperor.scs.com> by brucek@emperor.scs.com (Bruce Klein): > How are you people achieving this? > > I have a number of Telebit modems here. For this experiment, I called > a T2500 with a T1000 using the PEP protocol. The modems were connected > to their hosts through 9600 baud serial ports. > Well you won't get over 1000 cps with the serial ports stuck at 9600 baud. But I just have my T2500 -> T2500 connection set as it came out of the box and can get up to 1700 cps using ZModem transfer. I can get 910 cps on a V.32 link with V.42 error correction as well. -- *** John Veldthuis, NZAmigaUG. johnv@tower.actrix.gen.nz ***
root@zswamp.uucp (Geoffrey Welsh) (06/21/91)
In a letter to All, Bruce Klein (brucek@emperor.scs.com ) wrote: >How are you people achieving this? Actually, if you want to use YMODEM-G, I'm sure that you'll exceed 1000 CPS by a significant margin. >For this experiment, I called a T2500 with a T1000 using >the PEP protocol. The modems were connected >to their hosts through 9600 baud serial ports. Problems: 1) With 9600 bps serial ports, you're limited to 960 CPS, no matter what. Switch to 19,200 bps and you stand a chance of going over 1000 CPS. 2) I'm led to believe that the T1000 is not as fast as the T1800 (TB+) or T2500. >Using xmodem and the telebit xmodem support (s111=20) I was >able to achieve 500 cps. 3) XMODEM is a slow protocol, period... and few programmers bother to code it very efficiently as a result. In fact, even at 115,200 bps over a null modem, I have not been able to reach 1000 CPS with XMODEM! >Next I tried zmodem. With a Sun on one end and a Mac Plus >running Zterm >on the other (same two modems) I was able to achieve peak >thruput of 700 >cps. I was unable to sustain this though, and on average it >delivered >about 400 cps. There may be something wrong with my sz >binary--on EVERY >file it produces a CRC error at bytes 6144, 13312, 20480, >27648, .... >(Anyone have any ideas about that?) Yes, your computer-modem interface was not correctly configured and data was being lost. Consult your manuals on the subject of handshaking. RTS/CTS (hardware) handshaking is preferred, but XON/XOFF will do fine in almost all aplications. Note that each computer and its modem must be using the same handshaking technique (and the cable must support the relevant pins), but the two modems may be using a different apporach (e.g. the Sun and its modem may be using XON/XOFF while the Mac and its modem may be using RTS/CTS). -- Geoffrey Welsh - Operator, Izot's Swamp BBS (FidoNet 1:221/171) root@zswamp.uucp or ..uunet!watmath!xenitec!zswamp!root 602-66 Mooregate Crescent, Kitchener, ON, N2M 5E6 Canada (519)741-9553 "He who claims to know everything can't possibly know much" -me
wtm@uhura.neoucom.EDU (Bill Mayhew) (06/22/91)
Our PBX system uses a TDM bus to switch calls. The system, whose type number I don't know, is made by Harris. Our telecommunications engineer tells me that Harris recommends that 2400 is the highest reliable modem rate that the switch can handle. I have empirically verified that our in-house modem-to-modem calls with v.32 are troublesome. We normally run our modems through direct outside lines and only tested using the Harris PBX switch for use as a bakcup if our computer services lines went out for some reason. I'd imagine that PEP rates are much lower too. Centrex supplied by the telco might behave very differently from our PBX experiences. Centex terminates extensions in the telco's switch doesn't it? If so the grade of equipment doing the switching might be higher. If the switch is a long way away, there could be oportunity for signal degredation. Bill -- Bill Mayhew NEOUCOM Computer Services Department Rootstown, OH 44272-9995 USA phone: 216-325-2511 wtm@uhura.neoucom.edu ....!uunet!aablue!neoucom!wtm via internet: (140.220.001.001)
vances@xenitec.on.ca (Vance Shipley) (06/23/91)
In article <1991Jun22.160247.8304@uhura.neoucom.EDU> wtm@uhura.neoucom.EDU (Bill Mayhew) writes: >Our PBX system uses a TDM bus to switch calls. The system, whose >type number I don't know, is made by Harris. Our >telecommunications engineer tells me that Harris recommends that >2400 is the highest reliable modem rate that the switch can handle. ... >Centrex supplied by the telco might behave very differently from >our PBX experiences. Centex terminates extensions in the telco's >switch doesn't it? If so the grade of equipment doing the >switching might be higher. If the switch is a long way away, there >could be oportunity for signal degredation. Your Central Office supplying the centrex service also uses TDM (Time Division Multiplexing). Digital Central Offices reduce the analog line to a 64 Kbps PCM (Pulse Code Modulation) bit stream and use TDM to switch it. Possibly the Harris PBX is using less than 64Kbps or perhaps there is another problem here. TDM does not preclude baud rates greater than 2400. -- Vance Shipley vances@xenitec vances@ltg ..uunet!watmath!xenitec!vances
caf@omen.COM (Chuck Forsberg WA7KGX) (06/23/91)
In article <244.2861EA83@zswamp.uucp> root@zswamp.uucp (Geoffrey Welsh) writes:
-3) XMODEM is a slow protocol, period... and few programmers bother to code it
-very efficiently as a result. In fact, even at 115,200 bps over a null modem,
-I have not been able to reach 1000 CPS with XMODEM!
Depends on the computers and the software. A pair of DOS machines running
ZCOMM or Pro-YAM get up to 4000 CPS using xmodem-crc. (1359 sending to an
XT equivalent). Using 1k blocks raises the 4000 to over 9000 CPS.
--
Chuck Forsberg WA7KGX ...!tektronix!reed!omen!caf
Author of YMODEM, ZMODEM, Professional-YAM, ZCOMM, and DSZ
Omen Technology Inc "The High Reliability Software"
17505-V NW Sauvie IS RD Portland OR 97231 503-621-3406nolan@helios.unl.edu (Michael Nolan) (06/23/91)
caf@omen.COM (Chuck Forsberg WA7KGX) writes: >In article <244.2861EA83@zswamp.uucp> root@zswamp.uucp (Geoffrey Welsh) writes: >-3) XMODEM is a slow protocol, period... and few programmers bother to code it >-very efficiently as a result. In fact, even at 115,200 bps over a null modem, >-I have not been able to reach 1000 CPS with XMODEM! >Depends on the computers and the software. A pair of DOS machines running >ZCOMM or Pro-YAM get up to 4000 CPS using xmodem-crc. (1359 sending to an >XT equivalent). Using 1k blocks raises the 4000 to over 9000 CPS. I get about 1800cps on a 19200 hard-wired connection between my unix system and a 20mhz '386 PC. X/Y/Zmodem run at pretty much the same speed on this setup, too, with Zmodem being just a little bit faster than Ymodem, which is just a little bit faser than Xmodem. ------- Michael Nolan This is my .sig Internet: nolan@helios.unl.edu T*His_iS#MY%.SIg oN DrUGs!@%#@% UUCP: tssi!nolan Any questions?
brucek@emperor.scs.com (Bruce Klein) (06/28/91)
A while back I posted to the net wondering how so many of you were achieving over 1000 cps throughput between telebits when I had trouble getting 500 cps. Many of the responders pointed out that I should have been using a 19.2K port instead of a 9600 port. That is certainly true in general, but I believe I should be able to achieve over 500 cps even on the 9600 baud port. The most promising lead came from uunet!rwing!pat (Pat Myrto), who suggested that my problems sounded indicative of a flow control problem: >No they aren't. I suspect you have flow control problems, because of >the CRC error pattern. It might take that long to fill up the buffer, >it overflows, things stop for the resend, and the cycle starts over again. >Be sure your port and modem agree on the flow control used, and that both >SUPPORT it - many drivers don't support RTS/CTS control, or don't >do it right. You should be getting around 900 or more cps, depending >on several factors. Its either that, or the port is losing characters >causing errors. I *KNOW* the modems will do better than that, if things >are correct, I do it all the time. Uucp is delivering peaks of 1400 cps. >The fix, without being there and going over your hdwe AND software setup >and settings, is not obvious - it could be several or a combination of >things. All relating to overruns or other similar forms of character >loss introducing errors, and thus resends. Thank you to those who responded: uunet!pacdata!jimh (Jim Harkins) mpd@anomaly.sbs.com (Michael P. Deignan) loft386!dpi@cs.utexas.edu (Doug Ingraham) dspeed@sactoh0.SAC.CA.US (Dave Speed) uunet!rwing!pat (Pat Myrto) and any others who may have posted to the net that I haven't got yet (our feed is a week+ behind). -- Bruce Klein brucek@emperor.sjc.mentorg.com ... uunet!emperor!brucek