Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[jaydub]

The new TBB HTML5 tester has always given me poorer single thread speeds than the old Flash based tester.

Having decided to try and confirm that it was the line rather than the iMAC that was causing the issue, I ran some comparative tests on using Chrome on my Android mobile and was somewhat surprised to see that even with a wireless 5G connection, my mobile was giving me much better single thread results than the wired connection on the iMac.

iMAc results from yesterday evening:
https://www.thinkbroadband.com/_assets/speedtest/button/1495481108565477655.png
https://www.thinkbroadband.com/_assets/speedtest/button/1495480953675044055.png
https://www.thinkbroadband.com/_assets/speedtest/button/1495480804254765955.png

Phone results:
https://www.thinkbroadband.com/_assets/speedtest/button/1495484946602490555.png
https://www.thinkbroadband.com/_assets/speedtest/button/1495484868591810355.png

I've just run some more now and seen some similar results.  The phone results are a bit more variable but the average single thread speeds are still almost 10mbps better on the mobile.

I've tried doing a clean up of the iMAC, repairing the OS, running it with all extensions disabled in Chrome and running without the AV, none of which have improved things.

iMac:
https://www.thinkbroadband.com/speedtest/1495548057877274055
https://www.thinkbroadband.com/speedtest/1495548025384990255

Mobile:
https://www.thinkbroadband.com/speedtest/1495547839587675455
https://www.thinkbroadband.com/speedtest/1495547881403639155

Does anyone have any constructive thoughts about what might be causing the difference?

(Not sure whether this is the most appropriate forum, but it does relate to testing a FTTC connection.  Mods, pls feel free to move if not appropriate here.)

[Dray]

Maybe posting on the TBB forums would yield better results?

[jaydub]

Maybe posting on the TBB forums would yield better results?

Maybe, I have already done that.

I thought it worthwhile posting here as well, as there is generally a higher level of technical understanding from the Kitz forumites.

[sevenlayermuddle]

Have you tried using  'Activity monitor' during the speed test, to see if it sheds any light, any process munching up all the CPU, whatever?

https://support.apple.com/en-gb/HT201464

Assuming that's no help, another suggestion... has the Mac been powered up long enough to have settled down and become idle?   If recently awoken, it may be trying to update mail, download updates, or maybe some sort of iCloudy activities, that might be competing for bandwidth with the speed tester.

[tickmike]

there is generally a higher level of technical understanding from the Kitz forumites.

   

[nallar]

I'd hazard a guess that it's something to do with TCP window scaling or the congestion control algorithm.

You could try speed tests while messing with the settings discussed here: https://rolande.wordpress.com/2014/05/17/performance-tuning-the-network-stack-on-mac-os-x-part-2/

One option not covered in that post is net.inet.tcp.use_newreno. By default OS X uses the CUBIC congestion control algorithm. Setting net.inet.tcp.use_newreno to 1 swaps to new reno.