Many readers whine to me about the lack of speed increase in their web browsing after installing a daughterboard upgrade or graphics accelerator. They forget they are still using a 28.8K modem. When you finish reading this report, you're going to be green with envy. Meet...

..."Super Modem"!
(Previewed on 16 Jul '97--This series of tests completed 1 Dec '97)
© 1997, 1998, 1999 Rob Art Morgan, editor of BARE FEATS
Gotta Question? Comment? Want to advertise?
Email Rob Art at barefoot@hawaii.rr.com

The local cable television provider (Oceanic Cable) installed my "RoadRunner" Motorola CyberSurfer cable modem two weeks ago. The first thing I noticed is that the time it takes to start up an internet session is shorter. Even including the time it takes to login with the cable modem after restart (you are always connected), I can launch Netscape Communicator and finish loading my home page using the cable modem in less than HALF the time it takes to connect, launch and load with a conventional 33.6K modem.

Then I went to visit 5 typical sites (Netscape, BareFeats, MacUser, CNN, ESPN). It took the cable modem ONE THIRD the time to visit those sites as it did with the 33.6K modem.
(First time loads with cache set to zero.)

The third test I performed was to upload a 1.1MB document from my local hard drive to my shell account on the local internet service provider. The cable modem achieved the feat 22.5 times as fast as the 33.6K modem (20 seconds versus 450 seconds).

Then I tried downloading the same document. The results were even more impressive. The cable modem finished the job in 3 seconds! That's 150 times faster than the 33.6K modem or a transfer rate of around 350K bytes per second!!! I couldn't believe it. I tried it 3 more times and checked the size of the document received to make certain. Wow!

One last test. I wanted to dramatize the variance in download speed you will experience even with a cable modem depending on what server you use, where it is and how busy it is. I downloaded the same 1.5MB hqx document (MacAttack.hqx) from MacWorld's Mac Download archives on the continental U.S. and from the University of Hawaii Info-Mac archives on the same island on which I reside (Oahu).

---

REFLECTIONS ON AN "internet accelerator"
Are you one of those web surfers who wonders why you're web browsing in "molasses mode" after installing a daughterboard upgrade or graphics accelerator? If you're an internet junkie like me, this is a job for "super modem"... that is... if it's available in your area (bug your local cable T.V. provider).

• In the last graph above, I attempt to dramatize this point: your results will vary depending on:
  1) The processing power of the remote site.
  2) The user load on the remote web site.
  3) The slowest data path between your machine and the remote site.
• You'll notice the web page load speed gain wasn't nearly as dramatic as the download/upload speed of shareware files. Still, an average cut in first time page loading of 1/3 is just great.
• I haven't had a chance to play any internet based network games, but I can hardly wait....
• Cost? $40 a month for unlimited access. EXPENSIVE? Not when you consider it includes web page space, free use of modem, and no dedicated phone lines. In my case, when you figure I've been buying a new modem each year, paying for a dedicated phone line, and coughing up premium monthly ISP fees, it's a 50% per month savings for me! It's definitely cheaper than ISDN and much faster. There is installation cost. It varies depending on whether you already have cable service, and/or own your own Ethernet interface and transceiver.

• Test bed was an Apple Power Macintosh 7500 running a 233MHz 604e daughterboard with 48MB of RAM, and 512K RapidCache L2 cache.
• System was Mac OS 8.0
• Cable Modem was a Motorola CyberSurfer provided by Hawaii's Oceanic Cable's RoadRunner service.
• 33.6K modem was a Global Village Platinum
• Web Browser was Netscape Communicator 4.04
• For this test, I zeroed out the cache for two reasons. First, the RoadRunner ISP caches the frequently accessed web pages URL's, making a local cache unnecessary in many cases. Secondly, I have decided to address various methods of caching (hard drive versus ram disk versus no cache) in a future page.