ndt-users list created

[Aaron Brown <>]

Hey Chris,

This seems to be a property of either the OS or the browser buffering some 
amount of data, and not telling us about it. So, from the websocket client 
side, it looks like we’ve sent more than we have. The final result is the 
from the server’s perspective so it’s the correct one. We’re looking into 
ways around this.

I’ve not seen it on the download side, but it could be some timing 
discrepancy between when the server thinks a test started/finished, and when 
the client thinks it started/finished.

Cheers,
Aaron

> On May 4, 2015, at 1:42 PM, Florio, Christopher N 
> <>
>  wrote:
> 
> Here's an interesting user report, seems that either the graphical dials 
> are reporting wrong of the numeric is....   
> 
> 
> 
> 
> On May 4, 2015, at 1:35 PM, Greenwood, Liam 
> <>
>  wrote:
> 
>> I tested it in Firefox 38.0, MacOS 10.10.2, on a Retina Macbook Pro using 
>> the Thunderbolt->Ethernet connection trying WebSockets.
>> 
>> Upload dial pegged at maximum immediately, and showed no speed less than 
>> 7xx mb/s - for a final of 641.89.
>> 
>> Reran test without reloading page and again the dial pegged to max, this 
>> time display showed 8xx or 9xx speeds the whole test, and a final speed of 
>> 641.
>> 
>> Download test had a similar discrepancy, but not so exaggerated. Showed a 
>> final number of 797.20.
>> 
>> Third test was high 800s, very low 900s all test for a final 637.78.  
>> Download didn’t show a discrepancy I could notice.
>> 
>> Safari 8.0.5
>> 
>> Same discrepancy in upload display - slower results - 614 upload, 577 
>> download.
>> 
>> Cheers, Liam
>