is there any modem (mass market v.92) which does not have the 90 kbps
cap? (8/9 of 115 kbps)

since i'm hitting 90 kbps on text, it seems to be the limiting factor.

m.

Strings wrote:
well i see some business modems that are 230 kbps. pricey.

m.

Strings <strings@example.net> wrote:
The extra bucks probably will not gain a significant increase
data transfer rate.

If you are getting 90kb/s with a 115kb/s port speed it appears
that you are limited by compression, not by the port speed.
Otherwise you'd be "hitting better than 110kbps".

It might be that for some very small percentage of time
compression could be higher, but it will max out at something
less than 150kb/s peak, and that will only happen on very rare
occasions. Thus it is very likely that if you move to a modem
which supports a port speed of 230 kbps the affect you'll see on
data transfer rate is to go from 90kb/s to 92kb/s...

--
Floyd L. Davidson <http://www.apaflo.com/floyd_davidson>
Ukpeagvik (Barrow, Alaska) floyd@apaflo.com

~ >> is there any modem (mass market v.92) which does not have the
~ >> 90 kbps cap? (8/9 of 115 kbps)
~ >> since i'm hitting 90 kbps on text, it seems to be the limiting
~ >> factor.
~ >> m.
~ >
~ >well i see some business modems that are 230 kbps. pricey.
~
~ The extra bucks probably will not gain a significant increase
~ data transfer rate.
~
~ If you are getting 90kb/s with a 115kb/s port speed it appears
~ that you are limited by compression, not by the port speed.
~ Otherwise you'd be "hitting better than 110kbps".

115200 bps port speed gives 11520 chars/sec hence 11520*8 = 92160 bps.

~ It might be that for some very small percentage of time
~ compression could be higher, but it will max out at something
~ less than 150kb/s peak, and that will only happen on very rare
~ occasions. Thus it is very likely that if you move to a modem
~ which supports a port speed of 230 kbps the affect you'll see on
~ data transfer rate is to go from 90kb/s to 92kb/s...

One option is to go with a (shudder) Winmodem (preferably one
with a hardware DSP). I've measured 400kbps sustained with
a Cisco AS5400 -> Agere winmodem (V.90 + V.44), transferring
artifically highly compressible text payload.

Aaron

"Aaron Leonard" <Aaron@Cisco.COM> wrote in message
news:bgkl13tjs6fkk2u3oehjh6tvh8mnd87jcc@4ax.com...

i assumed that 8N1 would be 8/9 of 115,200. but i don't know much about
serial I/O.

TCP overhead could explain why i'm right at 90.

i will try a Winmodem. i will have to get a PC with pci slots.

i also use linux, so i'll buy an external too.

m.

thanks Floyd and Aaron.

i will do research on which ISPs have v.44.

Strings <strings@example.net> wrote
[...]
Last time Linux and controllerless modems were discussed there was a
Linux driver for at least one of these so I suggest you check before
you buy an external modem.

--
Mike D

"Michael R N Dolbear" <me@privacy.net> wrote in message
news:01c7850a$87bb1360$LocalHost@default...
i have 2 external modems already, but i want a v.92. i like externals for
non-windows, non-linux operating systems.

upgrading to cable modem would solve my problem. :-)

"Michael R N Dolbear" <me@privacy.net> wrote in message
news:01c7850a$87bb1360$LocalHost@default...
i've seen v.92 externals for $20 new! i don't know how they compare to a
Zoom or USR.

thanks for the info.

~ > 115200 bps port speed gives 11520 chars/sec hence 11520*8 = 92160 bps.
~
~ i assumed that 8N1 would be 8/9 of 115,200. but i don't know much about
~ serial I/O.

"8" is databits, "N" is No (0) parity bits, "1" is stop bit. 8+0+1 = 9,
but this misses the START bit. 1 start bit + 8 data bits + 0 parity bits
+ 1 stop bit = 8 (data bits) / 10 (total bits).

~ TCP overhead could explain why i'm right at 90.

Yep.

Aaron

On Sun, 22 Apr 2007, in the Usenet newsgroup comp.dcom.modems, in article
<462b76db$0$97239$892e7fe2@authen.yellow.readfreen ews.net>, Strings wrote:

8N1 has eight data bits, no parity, one stop bit, but also one start bit.
Thus the maximum port _byte_ speed us a tenth of the port speed. Note
that if you are using CCP on otherwise un-compressed data, the maximum
transfer speed might be somewhat greater. because the compression and
de-compression is done by an application that forms part of the network
stack - downstream of the serial link.

Depends on how you are actually measuring the data transfer speed. If
you are dependent on some speed indicator in your browser, you'd have
to research that appropriately. A more accurate speed indication is to
actually time the download of a compressed data file of adequate size,
but bearing in mind you're not the only one using the Internet, and
your packets may be contesting space on the Internet with Eleventy
Zillion people trying to download their pr0n. Assuming an TCP download.
ftping a tarball of the Linux kernel from kernel.org (43 Megs for a
bzip2 compressed 2.6.x kernel, 55 Megs if gziped) would be a good test.

Normally, with nominal 1500 octet packets, the TCP and IP headers (each
20 to 40 octets in steps of 4) don't amount to much - 3 to 5 percent
would be reasonable. The IP headers _MAY_ be compressed on a dialup
link (Van Jacobson header compression), which reduces the overhead even
more. Note that VJ compression (RFC1144) is separate from the hardware
compression algorithms your modem may support, and separate from any
data compression (RFC1962 updated by RFC2153) that may be negotiated
between the PPP link peers,

The original US Robotics model 5683 WinModem (the one that started it
all) was an ISA card. But - you have CPU cycles to waste?

See the Modem-HOWTO there are many internal modems (ISA and PCI) that
will work in Linux because they are not software modems. There are
even a few USB modems that work with most distributions.

Old guy