Terrible perfs with RT-N66U + late-2013 MacBook Pro

[Tallan]

Hi,
I encounter a huge latency problem with every 3 streams MCS speeds on a MBP late-2013 MBP.

I tested firmware 3.0.0.4.260, 3.0.0.4.374.720, 3.0.0.4.374.979, and merlin 3.0.0.4_374.35_4

Everytime same thing, catastrophic ping times with the router :
64 bytes from 192.168.2.1: icmp_seq=1349 ttl=64 time=1.278 ms
64 bytes from 192.168.2.1: icmp_seq=1350 ttl=64 time=1.274 ms
64 bytes from 192.168.2.1: icmp_seq=1351 ttl=64 time=1.141 ms
64 bytes from 192.168.2.1: icmp_seq=1352 ttl=64 time=211.322 ms
64 bytes from 192.168.2.1: icmp_seq=1353 ttl=64 time=131.835 ms
64 bytes from 192.168.2.1: icmp_seq=1354 ttl=64 time=52.215 ms
64 bytes from 192.168.2.1: icmp_seq=1355 ttl=64 time=280.167 ms
64 bytes from 192.168.2.1: icmp_seq=1356 ttl=64 time=200.603 ms
64 bytes from 192.168.2.1: icmp_seq=1357 ttl=64 time=121.299 ms
64 bytes from 192.168.2.1: icmp_seq=1358 ttl=64 time=41.357 ms
64 bytes from 192.168.2.1: icmp_seq=1359 ttl=64 time=269.228 ms


Aside of that, my old 2011 MBP on 2 streams MCS 14 300Mbps, 5GHz never showed any sign of problem :
64 bytes from 192.168.2.1: icmp_seq=0 ttl=64 time=0.874 ms
64 bytes from 192.168.2.1: icmp_seq=1 ttl=64 time=0.759 ms
64 bytes from 192.168.2.1: icmp_seq=2 ttl=64 time=0.782 ms
64 bytes from 192.168.2.1: icmp_seq=3 ttl=64 time=0.785 ms

Is it possible to disable a range of MCS index on the RT-N66U on the merlin firmware ?

 

[Tallan]

rate limited, problem remains the same

OK, Reading the Asus source code, I successfully managed to limit MCS, turning the RT-N66U into 2x3 MIMO on 5GHz (wl1)

nvram set wl1_txchain=5
nvram set wl1_rxchain=5
nvram commit
reboot

Now, both MBP, the new retina and the old one, synchronize in MCS 15 at 300MBps in 5GHz (40MHz channels).

Side-by-side, the problem persists on the Retina MBP while the 2011 one still works like a charm with 1ms ping.

Not knowing exactly which stream to keep, I tried 0-1 and 0-2. Same result.

Does nobody faces this problem ?

What are your ping time on your Retina MacBook Pro Mavericks 10.9.1 ?

I would tend to think it's a mac problem, but could you check on your side and report successful firmware versions on RT-N66U working with Mavericks retina MBP (preferably late-2013) in 300 and 450Mbps with rock solid 1ms ping ?

 

[RMerlin]

This is the first time I see that solution being used with Macs. Usually the fix most people use is enable 802.11d+h support. If you use my firmware I exposed the nvram setting on the Wireless -> Professional page. Might be worth a shot.

 

[Tallan]

solved !

Exploring the Mac way on another wireless infrastructure, I found that the problem is indeed a insane Maverics 10.9 issue.

The OS seems to put the wireless interface into sleep mode after 0.2s.
So with a default delay of 1s between icmp frames, ping is not able to keep the interface up for itself.

If I cut the delay to 0.2s (ping -i 0.2), the problem is gone !!

I put the tx/rx_chains back to 7, the Retina MBP is now synchronized to 450Mbps and ping time is clean around 1ms as it should always have been.

 

[helio58]

Hi,

Please could you explain how you manage to do that

"If I cut the delay to 0.2s (ping -i 0.2), the problem is gone !!

I put the tx/rx_chains back to 7, the Retina MBP is now synchronized to 450Mbps and ping time is clean around 1ms as it should always have been."

Thanks

 

[chercm]

This is the first time I see that solution being used with Macs. Usually the fix most people use is enable 802.11d+h support. If you use my firmware I exposed the nvram setting on the Wireless -> Professional page. Might be worth a shot.

which setting is it ?

 

[RMerlin]

which setting is it ?

Regional Mode. It defaults to "Off", can be set to "802.11d", "802.11h" or "802.11d+h".