|
Posted by Bill M. on December 9, 2006, 1:44 am
If you were Registered and logged in, you could reply and use other advanced thread options
On Sat, 09 Dec 2006 04:24:01 GMT, "Timothy Daniels"
>"Bill M." wrote:
>> "Timothy Daniels" wrote:
>>> What you measure is probably bps sent or received - NOT
>>>what bits are received without a request to retransmit.
>>>When packets are received with errors, a request is sent back
>>>for a retransmit. The InternetFrog site measures how many
>>>bits are received unmangled.
>>>
>>>*TimDaniels*
>>
>> Huh? No way. Like all so-called speed test sites, this site is a poor
>> indicator of actual performance. In fact, this one seems to be worse
>> than most others I've tried.
>
>
> On what do you base your opinions?
Just personal experience over the years. Speed test sites aren't
sophisticated enough to be trusted, other than as a novelty. They have
relatively small files available for download/upload, they typically
don't allow multiple connections, and they're unlikely to saturate
your pipe. Instead, get a program that monitors your transfer rate in
real time, (I use DUMeter but there are others), and find a way to
open multiple connections (either to the same server or to multiple
servers). I use ReGet Pro and have it set to open 8 simultaneous
connections. Some people are able to max out their pipe with fewer
connections, but 8 does it for me, any time of day or night.
The site you provided tells me I'm getting 11.9 Mbps down, but its
files are simply too small for it to realize that I only get 6.5-6.8
Mbps down after you factor out the initial SpeedBoost surge. I'm on a
6 Mbps plan, BTW.
>For some reason, Tier 3
>RoadRunner tech support seems to think the InternetFrog site is
>a useful measure (more useful than TimeWarner's site and more
>useful than DSLreports and Reviews.CNET) of actual information
>transfer, i.e. error-free packets received. That is what the % factor
>represents - percent of packets received error-free. As corroborated
>by a treaceroute, these errors were occurring in the routers just
>outside RoadRunner's network. In other words, their backbone
>provider was screwing up, and RoadRunner has to get together with
>them to get it solved.
For the vast majority of us, TCP packet retransmissions aren't enough
of a problem to affect throughput speeds, and the problem would have
to be pretty darn severe to drop throughput from 5 Mbps to 1.5 Mbps.
--
Bill
|