Amazingly Fast Dialup Throughput!
Steve
advantage of V.44 compression. I was amazed at the dramatic speed increase!
Web pages loaded much faster even though both modems connect around 50k
bits/sec. This fast throughput couldn't just be the result of V.44
compression so I did some throughput experiments.
I repeatedly downloaded different types of files using both modems and
averaged the results. I used system monitor to measure bytes received/sec
and multiplied by 8 to convert to bits per second (bps). Both modems
connected to my ISP between 49333 and 53333 bps. Here are the average
results:
File Type USR 56k PCI Zoom V.92 PCI
Zoom/USR Relative Speed
News Group Headers 115,200 bps 358,659 bps
311 %
Yahoo Web Page 111,776 bps 161,961 bps
145 %
jpeg files 52,408 bps 52,784
bps 100 %
Notice that the dramatic speed increase occurs on compressible data such as
HTML. Also notice that there was no speed difference when downloading data
that is already compressed (jpeg).
Why the dramatic speed increase from the Zoom modem? The Zoom modem is an
HCP modem; its modem controller function is performed by the host computer's
processor. The USR modem is a controller modem; its controller function is
performed on the modem card. Even though the USR modem sits on the fast PCI
bus, its onboard controller limits maximum dataflow to 115,200 bps. When
data is compressed higher than 2:1, the controller becomes a bottleneck.
Don't get me wrong, the USR modem is a well constructed, reliable modem.
Its speed limitation has nothing to do with the modem's quality. This
experiment shows that HCP modems provide a dramatic web browsing benefit
over controller modems (internal or external). HCP modems (winmodems) do
not deserve the badmouthing they have received in the past.
I searched the web and found this page that describes the same benefits that
I found.
http://www.toast.net/support/00298/
Good Surfing,
Steve
David 01
"Steve" <NoS...@NoAddress.com> wrote in message
news:VKjW9.1855$zF6.1...@bgtnsc04-news.ops.worldnet.att.net...
David
Paul Murphy
"Steve" <NoS...@NoAddress.com> wrote in message
news:VKjW9.1855$zF6.1...@bgtnsc04-news.ops.worldnet.att.net...
> not deserve the badmouthing they have received in the past.
>
> I searched the web and found this page that describes the same benefits
that
> I found.
> http://www.toast.net/support/00298/
>
> Good Surfing,
>
> Steve
point with these software modems (compared to a very limited range of
controller based modems). Unfortunately though there are still those
disadvantages with software modems that are listed on that website, as well
as a very important one that isn't there but makes many of them very
"badmouth worthy" - like most things that are done through software as
opposed to dedicated hardware, some of them tend to be far less reliable as
well as finiky regarding OS service patch levels etc.
Perhaps the ideal solution is for PCI full hardware modem developers (eg
Multitech, Actiontech and those using Lucent Venus or Topic/IC Plus
chipsets) to develop faster synthetic serial ports (if they haven't already
done so - and I have a feeling at least one of them has).
I'd rather have a much more reliable modem that always does what I want
under a multitude of OSs and is slower at 1 particular aspect of some
downloads than vice versa. I'd be interested to see how these tests compare
if a range of true full hardware controller based PCI modems as above were
used as the control samples instead of the USR External. That would be a
much fairer comparison. Of course some winmodems are better than other
winmodems and if those results showed a consistant speed advantage to
winmodems over full hardware products, then (and only then), I might
consider these "cutdown" controllerless products as being worthy of
installation into anything other than the cheapest of PCs. I might also
reconsider my very strong desire to badmouth them too.
Paul
S. J. Shestokes
"Steve" <NoS...@NoAddress.com> wrote in message
news:VKjW9.1855$zF6.1...@bgtnsc04-news.ops.worldnet.att.net...
> advantage of V.44 compression. I was amazed at the dramatic speed
increase!
> Web pages loaded much faster even though both modems connect around 50k
> bits/sec. This fast throughput couldn't just be the result of V.44
> compression so I did some throughput experiments.
Well, Ok it was a slow weekend, the weather was bad, so boogied over to Best
Buy and got that Amazingly Fast Zoom V92 PCI modem.
I mean I am a guy who will try anything to get faster speeds and anyway if
it didnt work BB is good for returns. Alas after much playing around with
it, I didnt find it any faster than my External Courier modem. I tested and
played with it but the Zoom just isnt amazing.
Note that right now I dont have V92 only V90. I had to to dump my V92
provider because of busies all the time.
The Zoom DID give a good showing tho, for $40 this is a damn good modem. I
have a 1gig AMD athlon computer and can monitor cpu usage with Ram Optimizer
The Zoom took about 20% of the cpu usuage vs hardly any with the external
modem. But I didnt notice any slow down at all because of the Zoom. Its been
many years since I have tried a soft modem - the Zoom is VERY good indeed,
but hardly amazing.
Ok, you want amazing performance you can see - try Propel Accelerator -
www.propel.com
Free to try, with no BS or registration or credit card and no Spyware or
banner ads. No it aint free after 10 days - it costs $60 a year.
I havent bought it yet - I am a pretty cheap guy ( poor too -retired). BUT
this Propel seems to be worth the $$. I just may get it.
Yeah, I can hear the snickering in the background. I laughed too at first,
but having played with Propel for a week, I aint laughing any more. It makes
web browsing fly. IT is AMAZING and you wont have to play, or test it very
long to like it.
It doesnt work on downloads or email, but it is wonderfully fast on the web.
Stan S.
I just might keep the Zoom too..........
Steve
My ISP's POP uses V.92/V.44. After posting the results of my original test,
I wondered how much of my observed download speed increase was due to V.44
compression and how much was due to the HCP modem's internal processing. I
therefore dialed into another POP that uses V.90/V.42bis and measured
browsing performance with the Zoom V.92 PCI modem. I also repeated the
experiment on the V.92/V.44 POP to compare download speed on the same day
and minimize day-to-day internet differences.
Here are the new test results for the Zoom V.92 PCI Modem:
File Type V.90/V.42bis POP V.92/V.44
POP
News Group Headers 227,299 bps 307,075 bps
Yahoo Web Page 118,720 bps 169,928 bps
Notice that the V.92/V.44 connection provides a significantly faster
download. (I didn't repeat the jpeg test because there is no speed
improvement on data that is already compressed.)
Conclusion: The significantly faster web browsing that I observed when
switching to the Zoom modem was a result of the HCP modem's processing
combined with V.44 compression.
Good Surfing,
Steve
"Steve" <NoS...@NoAddress.com> wrote in message
news:VKjW9.1855$zF6.1...@bgtnsc04-news.ops.worldnet.att.net...
Franc Zabkar
finger to keyboard and composed:
>Test Follow-up:
>
>My ISP's POP uses V.92/V.44. After posting the results of my original test,
>I wondered how much of my observed download speed increase was due to V.44
>compression and how much was due to the HCP modem's internal processing. I
>therefore dialed into another POP that uses V.90/V.42bis and measured
>browsing performance with the Zoom V.92 PCI modem. I also repeated the
>experiment on the V.92/V.44 POP to compare download speed on the same day
>and minimize day-to-day internet differences.
I'm a big believer in empirical data and your post is the first I've
seen that actually *measures* V.44 performance. Thanks for making the
effort.
-- Franc Zabkar
Please remove one 's' from my address when replying by email.
Ed Schulz
TIA TR-30.1 and ITU-T would not have approved V.44 had there not been
some good evidence of advantages over V.42bis. See these contributions:
ftp://ftp.tiaonline.org/TR-30/TR301/Public/TR-30.1/2000-01-BocaRaton/10001008.doc
ftp://ftp.tiaonline.org/TR-30/TR301/Public/TR-30.1/2000-01-BocaRaton/10001014.doc
I agree that measurement of implementations on real modem connections
is useful, too. To make sure that you're measuring V.44 data compression
and not something else, I'd suggest:
- Force the connect rate to something low that you know will experience
few LAPM frame errors, no retrains, and no rate renegotiations.
- After your test, check the connection using AT#UD or similar command
to make sure that there were few LAPM errors, no retrains, and no
rate renegotiations.
- Take care not to compare throughputs made at different DCE rates, at
different round-trip delays, at different LAPM error rates, to/from
different server modems, etc.
- Make a new call for each measurement. Otherwise, the compression
history (dictionary) might affect the results.
- Know the data you're sending/receiving, and recognize that your
result only applies to the specific data stream tested.
--
Ed Schulz
Agere Systems
edsc...@agere.com
Rick
> I agree that measurement of implementations on real modem connections
> is useful, too. To make sure that you're measuring V.44 data compression
> and not something else, I'd suggest:
Very good recommendations, Ed. Which is to say "it's not as easy as it
may first appear to make meaningful measurements in uncontrolled
conditions".
:-)