home *** CD-ROM | disk | FTP | other *** search
- Path: sobt.accessorl.net!user
- From: eric@accessorl.net (Eric Shaw)
- Newsgroups: comp.dcom.modems
- Subject: Re: Is 33.6 only available for USR?
- Date: Tue, 23 Jan 1996 21:57:26 -0500
- Organization: Access Orlando
- Message-ID: <eric-2301962157260001@sobt.accessorl.net>
- References: <4dd14b$odp@grid.direct.ca> <4ddchf$c4i@shellx.best.com> <30FA9B32.4862@fishnet.net> <4dh7j3$81i@shellx.best.com> <30FCE339.36B9@fishnet.net> <4dlt8f$1qom@seminole.gate.net> <DLGM5w.H8o@freenet.carleton.ca> <4dqj68$dcak@navajo.gate.net>
- NNTP-Posting-Host: sobt.accessorl.net
-
- In article <4dqj68$dcak@navajo.gate.net>, dhaire@gate.net (doug haire) wrote:
-
- >Anthony Hill (an171@FreeNet.Carleton.CA) wrote:
- >:
- >: doug haire (dhaire@gate.net) writes:
- >: > John White (jwhite@fishnet.net) wrote:
- >: > : better. Many hundreds of hours of my testing with Courier, Sportster
- >: > : PPI, Supra etc. show this is not the case. My Courier with the latest
- >: > : 33.6 code only marginally outperforms my PPI LCD. On a connection to a
- >: > : provider using Courier upgraded to 33k I achieve 28.8 about 80% of the
- >: > : time. With the LCD I achieve 28.8 50% of the time. Now 33k is really a
- >: > : big deal isn't it?
- That's right, *IF* you have a near *PERFECT* line to connect above 28.8,
- the Courier might perform just slightly better than the PPI or Supra on
- compressed data. For uncompressed data, a 33.6 connection on the Courier
- may likely be slower than a 26.4K connection on most other modems. The
- compression in there is SLOW. If you make a large file of just a few
- bytes repeated over and over and send it between 2 USR's, you won't get
- more than 6-7K/s, even if its all one byte repeating, because of either
- the processing speed of the modem, or because of inefficient algorithms.
- If the same file is transferred with a PPI or Supra or most other modems,
- the transfer rate will be limited by your DTE rate, i.e. 115200 limits you
- to a little above 11K/s, even if you lock the modems at 14.4K.
- >
- >I believe there was a change included with the V.34+ upgrade (at least in
- >the Courier) that improved speeds with the lower symbol rates, allowing a
- >28.8k connect (stable) at a lower symbol rate. I could be wrong about
- >this but I don't think so. I do know, subjectively speaking, that my
- >unstable 28.8k connects became more stable when I went to the V.34+ and
- >that I started getting more 28.8k connects than I had. Admittedly, this
- >is subjective since I did not run control tests.
- From what I've read, this is the case, which would help if your line has
- limited bandwidth. Haven't actually seen how much this helps - with the
- lines here, a 3429 or 3200 symbol rate is almost always used, but the SNR
- usually limits connections below 28.8K.
-
- >: > There is more to that improvement than just the connect speed. Look into
- >: > that Courier sometime and examine how it handles V.42 error correction.
- Yeah, they went up to 244 byte LAPM packets (a wierd size, not a power of
- 2), when a lot of Rockwell modems (and others) have long been using 256
- byte packets, if not 512.
-
- The v.42 negotiation between the USR and the newer, crappier PPI modems
- has similar problems to the negotiation between it and the older PPI
- modems. The new ones use an AT&T chipset instead of a Rockwell one. I
- guess that must mean that AT&T and Rockwell have the same bugs <g>.
-