JohnButt: No over-clocking that we can see in Chorus connections.
As per Bartender's post earlier - it is an ordering option with Chorus, as the overclocking as part of the Accelerate product offers (e.g Bitstream 2a)
it may be to do with queuing, which can easily make significant differences to international traffic performance with minimal differences to local speedtest performance.
if you get 100 packets coming in and 20 of them are dropped, because they're coming in too fast, then there is a timeout before they are resent - increasing delay significantly. (much higher for single packets, like checking to see if a newer version of a image on a web site is available, which is a pretty common occurance, not so much for data coming through as subsequent packets can trigger shorter timeouts). if a connection can queue more packets instead of waiting to do trips across the world and wait for timeouts it can make significant differences.
getting a 50% improvement in performance from more intelligent queueing whilst keeping the same maximum bandwidth and minimum latency isn't uncommon. (ie 1200 msec average page load time instead of 1800 msec average page load time)
and even 200 msec drops in page load time can be noticed.
it's reasonably difficult to test in a consistent manner though, as you need multiple connections using a significant amount of your bandwidth to test well though - a mixture of cdn content hosted nearby and distant remote locations are one of the easier ways to trigger it though, which things like facebook can do. (although facebook has it's own issues)
Are you subscribed to our RSS feed? You can download the latest headlines and summaries from our stories directly
to your computer or smartphone by using a feed reader.
Trusted
Chorus
