Hi All,
> On Nov 6, 2016, at 11:47, Kevin Darbyshire-Bryant > <ke...@darbyshire-bryant.me.uk> wrote: > > > > On 06/11/16 10:12, Xander Shelley wrote: >> Hi there, >> >> I've installed LEDE Nov5 snapshot onto my TP-Link TD-W8970 router to use >> with a UK fibre ISP (Plus.net) >> >> Out of the box most things work great. >> >> If I connect via Ethernet to the router from my xubuntu T410 laptop, and >> run a speed test using (speedtest.net) I get close to line speed (31Mb/s >> Line Speed, measured speed 29MB/s) which is great. > > VDSL2 overheads would account for the drop there. Assuming TCP/IPV4 goodput measurements the best you can expect on a VDSL2 line is: sync bandwidth * coding factor * per packet overhead factor with coding factot: 64/65 payload factor: variable depending on encapsulation, typically PPPoE with MTU 1500 and VLAN PPPoE: 2 Byte PPP + 6 Byte PPPoE VLAN: 4 Byte VLAN VDSL2 (IEEE 802.3-2012 61.3 relevant fuer VDSL2): 1 Byte Start of Frame (S), 1 Byte End of Frame (Ck), 2 Byte TC-CRC (PTM-FCS), = 4 Byte COMMON ethernet: 4 Byte Frame Check Sequence (FCS) + 6 (dest MAC) + 6 (src MAC) + 2 (ethertype) = 18 byte Sum: 8 + 4 + 4 + 18 = 34 Bytes effective MTU 1500 - PPPoE - PPP = 1500 - 8 = 1492 Bytes payload factor: (1492 - 20 - 20) / (1492 + 8 + 4 + 4 + 18) = 0.951507208388 PPPoE goodput: 31 * (1492/1526) * (64/65) = 29.8430083678 IPv4 goodput: 31 * ((1492 -20)/1526) * (64/65) = 29.4429680411 TCP/IPv4 goodput: 31 * ((1492 - 20 - 20)/1526) * (64/65) = 29.0429277145 TCP/IPv4 goodput (assuming MTU 1508 baby jumbo frames): 31 * ((1500 - 20 - 20)/1534) * (64/65) = 29.0506468759 So 29 out of 31 is to be expected… About the wifi problem, I would try to test against a local machine connected to the wired LAN ports of the router to reduce the internet introduced variability… Best Regards Sebastian > >> >> However, when I switch over to using wifi, this measured speed drops to >> 12-16MB/s.... >> >> Looking at the LuCI interface it shows that my laptop is connected >> around 113MB/s which is as expected. >> >> Am I doing anything incorrectly/missed a step in my setup? > > I can't offer any magic bullet. Wifi is notoriously variable in terms of > bandwidth and I have to say my 2.4Ghz is slower (similar to your measurement) > than 5Ghz even though rates would suggest plenty of spare capacity. > > It's worth proving the router isn't running out of CPU though...a top during > wired/wireless tests would be helpful. > > Kevin > > _______________________________________________ > Lede-dev mailing list > Lede-dev@lists.infradead.org > http://lists.infradead.org/mailman/listinfo/lede-dev _______________________________________________ Lede-dev mailing list Lede-dev@lists.infradead.org http://lists.infradead.org/mailman/listinfo/lede-dev