home *** CD-ROM | disk | FTP | other *** search
- Path: news.fido.net!fidouk!2-254-235-0!mark.simms
- From: Mark.Simms@f235.n254.z2.fidonet.org (Mark Simms)
- Newsgroups: comp.dcom.modems
- Subject: hypermodem - hype
- Date: 24 Mar 96 19:55:02
- Organization: Barney's Rubble
- Message-ID: <a7a_9603250217@fidouk.fidonet.org>
- NNTP-Posting-Host: rubble.fidouk.org
-
- -=> Quoting RobertN141 to All <=-
-
- Ro> I may be missing something here. A connection of 21,600 although
- Ro> possibly the result of a variety of problems is not considered a good
- Ro> connection today
-
- This was a transatlantic connect to a system which, I'm given to understand
- has a bandwidth restriction resulting in max 24k connects anyway
-
- Ro> 21,600 is still better than 14,400 but is not great compared to most
- Ro> modems.
-
- Stop concentrating on the connect and read the log extract: 10,600cps
-
- Ro> Our AMQUEST Hyper Modem reports consistent 28,800 connections in most
- Ro> normal conditions and usually falls back to 26,400 under adverse
- Ro> conditions.
-
- Rubbish. No modem can do that. Lines' bandwidth and s/n see to that
-
- Ro> I understand it this is an average we seek at 28.8.
-
- Even more rubbish. Top end connects are restricted by the POTs (Plain
- Old Telephone System) No matter how fast your processer, your port;
- if you havn't got the bandwidth to the remote, if you havn't got the
- s/n to the remote, you arn't going to get a 28k8 connect. Period
-
- Ro> Many if not most
- Ro> 28.8 modems top out at around 88,000 throughput.
-
- Then you'll see that 10,600 +cps is well over the top then :-)
-
- Ro> tested by a reputable source. This is due primarily to the horsepower
- Ro> of the microprocessor/microcontroller on the card.
-
- Whilst it's true that the apparent effective thruput could've been upped a
- little with the use of a larger LAP-M frame size and SREJ (selective reject)
-
- Ro> With the HyperModem
- Ro> we have sped up this circuitry and firmware to provide enough
- Ro> horsepower.
-
- You just wont get anywhere near double the effective thruput just by
- having a port at 230k or a different processor.
-
- Ro> other major constraint is the 115,200 dte which is believe it or not a
- Ro> somewhat
- Ro> mythical restriction of the UART. We moved this up to 230,000 which
- Ro> allows us
- Ro> to handle the bursts of compressed data which in some cases are higher
- Ro> than
- Ro> the average 4:1 compression.
-
- C'mon! Let's see verified actual connects, not in the lab, not on the PR
- Agent's desk but _real life_ . What fossil locks at 230k?
-
- Ro> A final word. This can get very technical and controversial.
-
- That'll make a change <bg>
-
- Ro> AMQUEST HyperModem we tell people to fallback to the basic facts that
- Ro> our overall architecture is faster and our DTE rate is 230,000. So
- Ro> fundamentally it
- Ro> follows that we can handle faster data if it is being sent to us.
-
- No, it does not follow. Except mebe where two paralell port modems are
- connected in lease line mode perhaps back to back. And there's no point
- in that since a transsfer program such as contained in Dos 6.2 and up will
- do that
-
- Still waiting on verified log extracts. Btw, what page or frame size do
- you use for either MNP or V42bis. Do you use selective reject? Do you
- use MNP-10? Can you run MNPn with V42bis? Have you Cellular support?
-
- Cheers
- Mark @ City View
- ================
-
- ... mark.simms@p7.f235.n254.z2.fidonet.org
-