DSL ISPs Cable Companies Dialup Internet Service Providers
Greets!
Thought I'd share with you my 56k experiences.
I'm using a Motorola ModemSURFR56 external on a 200mHz AMD K6 system connecting to a Cisco AS5200 access server. I live about two blocks from Ameritech's local phone switch (a NorTel DMS-100). Even in such close proximity to the switch (about 1,000 feet) I almost always connect at 44,000--never higher. Occasionally, I get 42,000 or 40,000 but 44,000 is much more common. The connection is very stable and rarely gets dropped.
Using WS-FTP, I can pull data from a local machine at about 3.8 k/sec. Surfing the web doesn't really seem that much faster but I do notice reduced latency. Server connections are established quicker and pages begin to load much sooner than with a v.34 modem.
My first attempts at connecting with k56flex were pretty sad. I would sometimes connect at 34,000; sometimes 38,000. More often it would fall back to v.34 and connect me at 28,800 or 31,200. I knew the fault was with my Ameritech-supplied subscriber loop. I called Ameritech and explained my
problem. Their script-reading robot told me that data rates are only guaranteed up to 2,400 bps and that there was nothing they could do. I was very irritated with their lack of interest in my problem. The next day I called them back and played stupid, explaining that my line "sounded" bad. I also explained that I had plugged two different phones into their demarc outside the house and that it sounded bad there as well (eliminating house wiring as the culprit). They agreed to send a technician out to look at it.
Two days later their techie shows up, plugs in his buttset and says "sounds fine to me." I insist that it sounds funny and he finally agrees to check it out on the switch side. Within an hour I was connected at 44,000 and have had consistent results ever since.
-- Eric Hendrickson November 16, 1997
Average connect speed is 46000.
Good performance, many dropped connections early in a session, then it smooths out and doesn't disconnect, no idea why. Disconnections are more common early in a session if there is a short period of inactivity. Apparently my modem dislikes inactivity!
Excellent throughput when connected. Compressed files, when connected to a fast server, can be close to connect speed. Compressed files vary wildly, as one would expect. Pure text can DL at 5.7k/sec and higher. FTP for compressed files: 4.8k - 5.0k, 4.85k sustained, on a 46000 connection
I am overall disappointed to learn that X2 users, both in my research via the Net, and in personal experience, typically connect at higher speeds. I am contemplating switching to X2 as a result.
One interesting thing: I got the Flash ROM update and connect speeds suffered, now it's 50% 44000 and 50% 46000, with no other performance parameters affected! I complained to Motorola, no reply yet.
-- Bill Brewer November 11, 1997
Hello to all
Well I have a voicesurfer (motorola), and after some tweaks I have been able to get a stable connection.First, I have the latest flash update and inf's as posted at the Motorola site, Second after many problems with disconnects from my isp (starnetinc.com), and they are using Livingston equipement, I have found an init string that give me good stability. It is as follows... at&f&d3&c1&k0&fw2+ms=56,1,34000,54000 I am able to connect at a speed of 46,000, consistently, and I am not able to connect at a higher speed due to the distance of my phone line at home to the telco switching center.I have rewired my home wiring myself and I have usedtwisted copper wire that is shielded, since I have three lines and I wanted to get rid of any bleed through. Also I have a direct line from the telco box to the modem with no other devices sharing that line, this minimizes the chance of the line picking up any noise from anything else, this is a good idea since the 56k connectons are sensitive to line noise. When download files, I am able to download at a rate of 4.0 k to 7.0 k, and according to net medic, my transfer rate is proceeding at rates up to 60-70 kbs, with compression bieng utilized at rates above 46 kbs. One other thing to mention, I have changed the MTU settings n windows registry from the default 1500 to 576, and guess what.... it really does make a difference. Check out http://www.sns-access.com/%7Enetpro/maxmtu.htm#explain Thats all for now, if I have anything new I will post it!
-- Glyfad November 12, 1997
I have tried two K56Flex modems with my ISP, Erols. With the Zoom internal version, model 2819, I get between 36000 and 42000. With the Motorola internal Modemsurfr, I get between 40000 and 44000. This line gets between 28.8 and 31.2 with regular v.34.
I must say, however, that there is still quite a bit of difference between ISDN over Ethernet via a Cisco router at 64K and the 44K that this analog modem gets. For users who can get flat rate ISDN, that seems like a far better solution. For me, in Bell Atlantic land, the $40/month for ISDN usage is getting to be a bit much, so that's why I'm considering analog.
I hope this is helpful to your readers. (I wasn't able to get an ftp to an Erols server, but I did see downloads of about 4.0K-4.2K/sec using a Web browser, downloading both flash rom updates.)
-- Matt Webster October 22, 1997
These results are with a Motorola Modemsurfr 56k internal on a Pentium 133 with 24MB RAM. Zip file dowloads were using ws_ftp32.zip (~470KB) and text file downloads were the active newsgroup file (~1.12MB) at my ISP. All files downloaded from my home directory on the local server. These downloads were using v1.009 of the firmware on my Motorola Modemsurfr with Livingston Port Masters at my local ISP (MCSNet).
The chart below shows results from some tests of Windows95 Software compression. In addition to checking the Software Compression Enable box, I had to add the init string "%C0" to my modem to get it to work properly. Column 1 is obviously the initial connect rate as reported by DUN. Column 2 indicates if the file is a zip or text file/if software compression was used (YES=software compression, NO=hardware compression)/and if it was an Up or Download. Column 3 indicates the range of download speeds as reported by system monitor throughout the transfer. Column 4 reports the average speed given by the FTP client as well as the overall time in seconds to complete the transfer. It appears that System Monitor results only take into account the effect of hardware compression, while the results given by the FTP client take into account both hardware and software compression, minus any overhead used by error correction, etc. Column 5 lists the ping results to the dialin server, averaged over about 50 "Pings".
Connect Type/Comp/ Speed Speed/Time PING Speed Up-Download Sys. Monitor WS_FTP Time ------------------------------------------------------------------ 48000 Zip/NO/Down 5.5-5.7k 5.3k/87.8s 100ms 48000 Zip/NO/Up 2.8-3.2 2.81/166.9 100 48000 Txt/NO/Down 11.4-11.8 11.13/98.6 100 48000 Txt/NO/Up 7.9-8.6 7.83/140.1 100 48000 Zip/YES/Down 5.5-5.7 5.33/88.0 85ms 48000 Zip/YES/Up 2.4-3.2 2.70/174.0 85 48000 Txt/YES/Down 5.5-5.7 13.65/80.3 85 48000 Txt/YES/Up 2.7-3.3 6.83/160.6 85
-- name withheld by request, October 3, 1997
hey all -
i have had more helpful info. come off of your page - so i thought i'd share my 56k modem speeds and how i had to tweak it.
My isp is running accend 56k banks , digital lines from them all the way to the trunk in my neighborhood. I wired my house line myself, with cheapo wire. i have used the flash upgrade usflash.exe from motorola, as well as the inf. they had at their site.
Using tcp tester - i see 60-85k/sec between my isp and home.
On a FTP transfer ,, say when i upload a www.page or something to my server, ranges from 40k- 60k/sec (I think he means 4.0 to 6.0 K - Les).
While going to www.microsoft.com, netscape showed 7.4k/sec at bottom where
it displays status....if that is a true figure i do not know. My connect speed in the modem status icon says 26400 , and my upstream speed according to my providers stats page is 28800.
Here is the modem init i have best luck with, even though it supposedly sets a limit of 33600(i upped this # to 57600 and it wouldnt connect)...
AT+MS=11,0,21600,33600
Also - against what my isp says ,, i MUST put the DNS #'s as DNS and Gateway, and turn off compression.My isp says this is all wrong , but it works for me.What can i tell ya ?Any tweakz are always helpful....
Good luck and keep up th great site!
-- Saylor October 1, 1997
ati3 -> V1.009-K56_DLP
The test file was compressed(Zip) file, of file size: 1.33MB.
Avg. Initial Connection Speed: 48Kb/s
Avg. ftp get speed: 4.48 Kbytes/s
Note: Previously with the V0.519D-K56_DLP code I was getting frequent disconnects, when connected at high speed. After I upgraded to V1.009-K56_DLP code the disconnects ceased.
-- Harold Woodward September 19, 1997
Here are some additional results from my Modemsurfr 56k internal modem. My previous results were with version 0.516 from Motorola. I have recently flashed my modem to v1.009 (possibly a beta flash upgrade). I have since resubscribed to my original (MCSNet) Livingston-based ISP. As I still have an account on my newer Ascend-based ISP (MCHenry Com), I did some additional tests there as well. Of course, my original tests with v0.516 were at McHenry com. Much to my surprise, I was able to get 56k-type connections and downloads with the Ascend based ISP as well. I was under the impression that after the flash upgrade to v1.009, I would only connect at v34 speeds to older k56 "plus" equipment. Apparently, this is not the case. Here is a summary of the old and new test results. I now typically get 31.2k connects with V34 and 48k connects at k56 speeds with both ISPs.
The downloaded file was ws_ftp.zip, located on the local FTP server in both cases and downloaded via WS_FTP LE. For the v1.009 tests, I connected a few different times and averaged the results. The McHenry 56k results are from Naperville, IL to the Elgin, IL POP. The McHenry V34 results are from Naperville to the Naperville POP. The MCS results are from Naperville to the Naperville POP. Tests were between 6PM and 6:30 PM on 9/11/97.
Code ISP connect Download Speed Download Time 0.516 McH 48000 5.0 to 5.6 87 s 0.516 McH 31200 3.5 to 3.9 135 s 1.009 McH 48000 5.3 to 5.6 (5.22 avg) 89.9 1.009 McH 31200 3.2 to 3.9 138.8 1.009 MCS 48000 5.4 to 5.7 (5.35 avg) 87.7
McHenry's own test results are at www.mc.net/addan/results.htm
As you can see, the k56 results are respectable and essentially identical for both versions and ISPs. I will try to do some more detailed and "scientific" tests if I have time.
-- name withheld by request September 19, 1997
Initial connect speed: 28.8
FTP transfer rate: Recieved 103247 bytes in 31.3 secs. 3.22 kbps
-- Michael L. Read September 19, 1997
My original survey message complained about the poor performance of my Motorola ModemSURFR. Well, I upgraded it to the v1.009 firmware a few days ago and the device now actually works! Imagine that!!
Here are the results:
Modem: Motorola ModemSURFR 56K External
Initial Connect Speed: 46,000bps
Sustained FTP Transfer Rate: 4.7k/sec
-- Earl Hall September 6, 1997
(Compare Earl's earlier results, below. - Leslie)
ati3 -> V0.516-K56_DLS
The test file is a zip compress file of size 106,400 bytes.
Avg. Initial Conneciton Speed: 48Kb/s
Avg. ftp put speed: 3.21 Kbytes/s
Avg. ftp get speed: 4.73 Kbytes/s
Good performance if the line would only stay up.
-- LaRCNET Network Support Office August 28, 1997
ISP-side: Ascend MAX 4004 on ISDN PRIs
Connects at 44-46K
Downloads around 4.1-4.5KBytes/sec
ISP is running Rockwell V1.xxx code, Motorola is runing V0.519D-K56_DLS with datapump RC56DPF L8570A Rev 13.2/13.2. Pretty pleased considering we're on the bleeding edge with all this... Waiting patiently for V1.xxx for the Motorola...
-- Steve Bour, August 21, 1997
Connect Speed: Usually 42,000; sometimes 44,000
Transfer rate: only apx. 2.0Kbps
Not only does my ISP's 56KFLex line give me this poor performance, but so does Motorola's special test line. Pretty disappointing.
Earl Hall July 27, 1997
(See Earl's later message, above, after upgrading his firmware. - Leslie)
Initial Connect Speed Varies between 48,000bps - 50,000bps
FTP Transfer Rate on a 770kb (.zip) file is 4.7kbps - 5.2kbps
Thanks...
-- Mike Talbert July 25, 1997
I recently noticed a link at your site to the modem tests at McHenry Com. I have been using their k56flex site in Elgin for a few weeks. It is about 15 miles from my local exchange (they only have a standard 33.6 line in my local exchange at this time). Feel free to post any of this information, but PLEASE do not post my e-mail address. Here are the results to compare with McHenry's tests:
Motorola Modemsurfer Internal (McHenry used an external for their tests)
RCV56DPF L8570A Rev 13.1/13.1
Note: My household internal wiring is very good. Speeds as reported by System Monitor.
56k connect 56k download speed 56k download time 48000 5.0 to 5.6 kbps 87 seconds 33.6 k connect 33.6k download speed 33.6 k download time 31200 3.5k to 3.9kbps 135 s
I have downloaded their test file a few times over the last month with nearly identical results. I have always connected at 48000 to the McHenry site.
Below is an excerpt of a post I made to a newsgroup on my other ISP on 5/8/97 with some additional data for my modem:
I've recently tested my Motorola Modemsurfr 56K on the Moto test line. It's worth a long distance call. I connected about 10 times, getting 48K connects 5 times, 50K connects 4 times and 52K once. I connected at 31.2K once before I remembered to remove the 56k disable string. They have some test files to download. With a 50K connection, I downloaded a 1.31 MB "Graphics" file in 1 min, 50 sec. System Monitor reported a constant download speed of about 11400 bytes/second, so I assume it was easily compressible. I downloaded a 1.25 MB "executable" file in 3 min, 15 sec. Sysmon reported varying trasfer rates between 4600 bytes/sec and 8700 bytes/s. I have no idea what the files were as I haven't looked at them yet. I got the number from Motorola Tech Support.
I posted this update on 6/4/97, before I signed on with McHenry:
As a quick update, I just spent a few minutes being the scourge of other local ISPs by using their k56 lines to test connect speeds again. Two connects to McHenry's Elgin line, both at 48,000. Two connects to Megsinet's Downers Grove line, one at 46k, one at 48k. Two connects to Megsinet's Aurora line, both at 48k connect speed. Their guest access did not seem to be working, so I could not try any downloads. Two connects to Motorola's test line, one at 48k, one at 50k. All these are as reported by Windows95 DUN window. No downloads, so the tests aren't particularly good for anything but speculation. All these connections are from Naperville. The Moto test line is in the southeast (Alabama?), yet I've consistently had the best connect speeds there, and also had very good dowload throughputs and transfer times as well.
-- name withheld by request
