Forums / Support / bandwidth throttle broken in 1.95?

bandwidth throttle broken in 1.95?

by dspbb @
When I throttle to 200 KBps (I normally get 500 KBps), the following occurs:

As reported by my firewall,
The speedIn will spike to about 350, then tapper of to less than 0.05 in about 5 seconds.
This repeats every 33 seconds

When it is not throttled it is a fairly smooth 500
I am using 3 connections

Anyone else having this problem?
 

BinaryBoy's reply to dspbb #921 @

That's pretty much the way it works. It only cares about the average bandwidth.
 

dspbb's reply to BinaryBoy #922 @

That's pretty much the way it works. It only cares about the average bandwidth.

5 seconds * 350 = 1750 (and really, it was like 350 for 1 second, 120 for 1 second, 30 for 1 second, 5 for 1 second, 1 for 1 second, so total = 506)
33 seconds * 200 = 6600
1750/6600 = 73% error
0506/6600 = 92% error
This isn't really even close to a correct average. Is 33 seconds the throttle cycle? Is there a known reason why the throttle is wildly inaccurate?
 

BinaryBoy's reply to dspbb #923 @

I'm looking into it.
 

BinaryBoy's reply to BinaryBoy #924 @

A new build is available from http://www.binaryboy.com/patch.php The previous build had a bug that I thought I'd already fixed. This new build uses a different method to calculate and maintain the average bandwidth. The overall bandwidth is based on the last 6 seconds and the average of the last 2 seconds is displayed by each progress bar.
 

dspbb's reply to BinaryBoy #925 @

Thanks for the quick response. Much appreciated
 

dspbb's reply to BinaryBoy #926 @

That's pretty much the way it works. It only cares about the average bandwidth.

:o
Sorry, weeks ago I had set it to display KBps. So when I set the throttle I was thinking in KBps, even though it does say Kbps right next to the input box. So a 92% error is about right when going from bytes to bits :o :o :o .
Good thing I don't work for NASA
 

BinaryBoy's reply to dspbb #927 @

:) Well the code is cleaner now anyway. It no longer relies on the accuracy of the Windows timer message, which had been bothering me for a while.