View Full Version : anybody compare web speed tests to...

10-16-00, 12:43 PM
had anybody out there actually compared the results from web based speed tests to a more reliable test method?

I know that web based tests are not reliable, but I am wondering how great a margin of error there is.

Here is something that I have noticed and am curious about: at http://www.computingcentral.com/topics/bandwidth/speedtest50.asp

I get a pretty consistant rating (abt 1400-1200 bits/sec, sometimes as low as 900 at peak hours on adsl) the thing is that if I go to that site, and repeatedly hit my back button to retest, the results go crazy. I tend to get a very large result, followed by a small result. Example from a few minutes ago: 1277.9, 566.7, 1283, 412.1, 1457.1, 1075.7

So I'm wondering if the first result is more accurate than the rest, and if repeatedly retesting somehow freaks out the test process, or if their accuracy is that far out of line.

does anybody know?

10-16-00, 03:04 PM
I like DSL reports, combined with a copy of Netstat Live.........


10-16-00, 04:24 PM

The reason you are getting huge number the second time is because you are using cached data. Yes, the first number is the most accurate. If you test multiple times, be sure to empty your cache before each time in order to obtain the most accurate reading.


For unto the Trailer Park is Born a Savior...

10-16-00, 05:51 PM
actually, its the other way around - I get the big number the first time, followed by a small number...

any idea why that is?

10-16-00, 07:22 PM
Originally posted by sol:
actually, its the other way around - I get the big number the first time, followed by a small number...

any idea why that is?

I took the test 3 times,as for my down and up speed, 2 were exactly the same 529/ 64 and the third was higher.

Whitetrashgod is right about clearing your cache. If you want the most accurate way to test the REAL banwith yoiur are receiving

As for your results differing , that could be due to net congestion at that time.

BTW, what is your up and down speeds for your ISP? Just wondering cuase you get a hell of a lot more than me http://www.speedguide.net/ubb/smile.gif I pay 50.00 bucks a month too http://www.speedguide.net/ubb/frown.gif but it beats a 56k anyday! http://www.speedguide.net/ubb/biggrin.gif


[This message has been edited by ***ScrUm*** (edited 10-16-2000).]

10-16-00, 07:43 PM
"Our service is a best-effort 1.5 megabits per second
connection and actual connection speeds may vary."

I can't remember or find any specifics on connection speed other than that. Being new to all of this, I was intimidated by the tweaks on this site. I went to dsl reports, and found a set of instructions targeted at people who didn't know what they were doing, and who didn't want to break their pc's.

I followed the 3 steps and my connection speed DOUBLED! good stuff!

I pay $39/month through earthlink. Well, I called earthlink and dealt with them, but my domain is mindspring and the sercive is provided by covad... the price was set by earthlink though.

I've been happy with the service so far.


10-16-00, 11:51 PM
That site is a little wierd. It's 12:52 EST I did the site 7 times clearing cache after each. 4 times - 4080 kbps,500kbytes/sec; 1 time 2956.5,362.3 ; 1 time 1854.5,227.3, & 1 time 927.3,113.6.

10-17-00, 12:55 PM
I get a pretty consistant rating >>>>>>
Sol, I think this must be a reason for concern. It CANNOT be consistant by definition. There are guys here who can explain to you the reasons, much better, than me. Try to download a file, for a few minutes. There is a meter in the download box. It shows you low speed at first. The speed gradually increases and then FLUCTUATES around some average number. It is never constant. Any commerciual web pages show you the speed at the moment of the test's start, or finish, or average during the test. It never shows you the dynamic speed, or the speed(s) during all the time you are connected. Besides, there is the cache issue. [Although, I hope that a well designed meter will take care of the cache. But, just in case, clear your cache between tests].
And, why do you want to know the exact number? Isn't it like knowing the exact number of electons reaching your computer? The whole thing is like traveling on a highway: you can travel in a bus, sports car, motorbike, etc. During rash hours, at night, at rain, etc. When I surf the Net on Sundays, it's fast. Practically all pages load the moment I click. I have only 364 kbps ADSL. If I had 10MB/s cable, it could not be faster. But now many pages take 1-3-5-10-15 seconds to load, I know that my AVERAGE speeds are below 100kbps (I have BroadBandWizard). What I do not know (and NOBODY here could answer this so far): will cable improve my connection at such "slow" times? Or these slowdowns are for the Net as a whole, POTS, xDSL and cable notwithstanding?

frustrated Yurka :-)

Cable Guy
10-17-00, 02:43 PM
Im usually in the 2.9-3.2 range

Its the cable guy!

10-17-00, 04:16 PM
I still think that this site is weird - or that there is something weird with my computer or connection.

by cleaning my cache between tests, I get more consistant results, but they are still all over the board

2:15pm pst - I returned the result 927.3 3 times, but also returned 6 other results ranging from 491.6 to 1236.4

are there other web based tests besides this one and dslreports that I can compare to? I can't get the dslreports test working - something to do with my browser's java not working...


10-18-00, 12:44 PM
Sol, relax. It's possible, of course, get the same, to the decimal point, speed, three times in a row, but I'd be very suspicious!
491.6 to 1236.4 is more likely. I'm sure that you didn't notice any diffrence in actual speeds, the download was fast. It's fast above 200 kbps, most pages load instanteneously.
If you want to check the other sides, do the search on this and other BBs, you'll find plenty of URLs.

10-18-00, 03:49 PM
Thanks, Yurka.

Don't worry, I'm not hyperventilating. I realize that speeds fluctuate, and I understand your point. However, I had assumed that a test would be designed with a large enough data sample to smooth out those fluctuations to a reasonable average. Getting a 412 and 1457 as responses within a very short time period seems like an UNreasonable average. So my curiosity got the better of me. Is it a poorly designed test? is there a reason (like cleaning the cach between tests) that the numbers are all over the board? Are other people's experiences similar? these were the kind of things that I was wondering.

Also, and nobody seemed to know this one, I was wondering if anybody had compared web based speed tests to a more reliable testing method. Just wondering, because I get such a range speed ratings.

Yurka, you are right about one thing: the "number" or exact speed means nothing, especially for me. I don't play games, or even download large files very often. Mostly I just surf, and I can't see any difference in the time it takes to load web pages between times when I am connecting at 600 bps and times when I am connecting at 1200.

Thanks for the input


10-20-00, 03:46 PM
Sol, sorry. I was busy, also got the RR cable, instead of my 364/128kbps ADSL. It's faster, but not dramatically. While on this BB, some threads load the moment I click, some take 1-3 secs, rare one takes 5secs or longer. If I type a message, it may take more or less to upload, too. I'm still stidying, but I know that ADSL was always faster on weekends. Occasional time-out occured with cable, too.
Ansering your main question: I have BroadBandWizard from Kissco. It periodically measures the speed to its site. I can measure it at any time. So, then I know that on "fast" days my speeds were higher than on "slow" days. A free gizmo called AnalogXNetStatLive measures my speeds to the given page, at real time. But it's so poorly designed that I can't read it. And it does the same thing: it gives high numbers for "fast pages" and low numbers for "slow" pages. It's like having a personal speedometer on a bus: you look out the window and know that the bus is going fast. You look at the dial and see that it's 70 mph. Whoa!
I was told that with NetStatLive, I will be able to tell whether the speed dropped because of my PC or the Net, but I don't know how to do that, even if I could read its face, which I can't, because it's thin green lines on black background.

frustrated Yurka :-)

10-24-00, 02:18 PM
Yes, now I know that cable is faster in general, it was also faster during the weekend.
But sometimes, I click a thread (or my email in Yahoo!) and it hangs for ... long. I guess it's in the nature of the Net, cable, xDSL or POTS. Take it or leave it.

frustrated Yurka :-)

11-01-00, 08:52 AM
i get what seems to be acurate results from free net mon http://www.kissco.com/
reason i seems to be
is i get about the same speed depending on time of day im on cable they also make a progy called broadband wizard and it seems to realy speed things up i whent from 512k a sec or so to um well acording to freenetmon 1.8 to 2.3 k a sec on cable isp so heck give it a try
i sugest backing up the registry for any one not familiar with the registry and editing manualy or do it right before a planned reformat then you have nothing to loose by trying it and a hole lot to possibly gain