Please post the output of cat /sys/kernel/debug/ieee80211/phy*/netdev*/stations/*/rc_stats while you're doing a speed test with the broken version
Give me a few minutes. I will have to reflash the broken version. Is it okay if I simply flash the current master branch (so all 4 patches included)?
Yes, that's okay
During upload:
root@LEDE:~# cat /sys/kernel/debug/ieee80211/phy*/netdev*/stations/*/rc_stats
best ____________rate__________ ________statistics________ _____last____ ______sum-of________
mode guard # rate [name idx airtime max_tp] [avg(tp) avg(prob) sd(prob)] [retry|suc|att] [#success | #attempts]
CCK LP 1 1.0M 120 10548 0.0 0.0 100.0 0.0 0 3 3 824 824
CCK LP 1 2.0M 121 5476 0.0 0.0 0.0 0.0 0 0 0 0 0
CCK LP 1 5.5M 122 2411 2.4 0.0 0.0 0.0 0 0 0 0 0
CCK LP 1 11.0M 123 1535 4.8 0.0 0.0 0.0 0 0 0 0 0
CCK SP 1 1.0M 124 10452 0.0 0.0 0.0 0.0 0 0 0 0 0
CCK SP 1 2.0M 125 5380 0.0 0.0 0.0 0.0 0 0 0 0 0
CCK SP 1 5.5M 126 2315 2.4 0.0 0.0 0.0 0 0 0 0 0
CCK SP 1 11.0M 127 1439 4.8 0.0 0.0 0.0 0 0 0 0 0
HT20 LGI 1 MCS0 0 1477 4.8 4.8 100.0 0.0 3 0 0 8 8
HT20 LGI 1 MCS1 1 738 9.7 9.7 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS2 2 492 14.6 14.6 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS3 3 369 21.9 21.9 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS4 4 246 31.7 31.7 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS5 5 185 39.0 39.0 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS6 6 164 43.9 43.9 100.0 0.0 3 0 0 25 25
HT20 LGI 1 MCS7 7 148 48.8 48.8 100.0 0.0 0 0 0 1 1
HT20 LGI 2 MCS8 10 738 9.7 9.7 100.0 0.0 0 0 0 1 1
HT20 LGI 2 MCS9 11 369 21.9 21.9 100.0 0.0 0 0 0 1 1
HT20 LGI 2 MCS10 12 246 31.7 31.7 100.0 0.0 0 0 0 1 1
HT20 LGI 2 MCS11 13 185 39.0 39.0 100.0 0.0 0 0 0 1 1
HT20 LGI 2 MCS12 14 123 56.1 56.1 100.0 0.0 0 0 0 1 1
HT20 LGI 2 MCS13 15 92 68.3 68.3 100.0 0.0 0 0 0 1 1
HT20 LGI 2 MCS14 16 82 75.6 75.6 100.0 0.0 0 0 0 1 1
HT20 LGI 2 MCS15 17 74 80.5 80.5 100.0 0.0 3 0 0 33 33
HT20 SGI 1 MCS0 30 1329 4.8 4.8 100.0 0.0 0 0 0 1 1
HT20 SGI 1 MCS1 31 665 12.2 12.2 100.0 0.0 0 0 0 1 1
HT20 SGI 1 MCS2 32 443 17.0 17.0 100.0 0.0 0 0 0 1 1
HT20 SGI 1 MCS3 33 332 24.4 24.4 100.0 0.0 0 0 0 1 1
HT20 SGI 1 MCS4 34 222 34.1 34.1 100.0 0.0 0 0 0 1 1
HT20 SGI 1 MCS5 35 166 43.9 43.9 100.0 0.0 0 0 0 1 1
HT20 SGI 1 MCS6 36 148 48.8 48.8 100.0 0.0 0 0 0 1 1
HT20 SGI 1 MCS7 37 133 51.2 51.2 100.0 0.0 0 0 0 1 1
HT20 SGI 2 MCS8 40 665 12.2 12.2 100.0 0.0 0 0 0 1 1
HT20 SGI 2 MCS9 41 332 24.4 24.4 100.0 0.0 0 0 0 1 1
HT20 SGI 2 MCS10 42 222 34.1 34.1 100.0 0.0 0 0 0 1 1
HT20 SGI 2 MCS11 43 166 43.9 43.9 100.0 0.0 0 0 0 1 1
HT20 SGI 2 MCS12 44 111 61.0 61.0 100.0 0.0 0 0 0 1 1
HT20 SGI 2 MCS13 45 83 73.2 73.2 100.0 0.0 0 0 0 1 1
HT20 SGI 2 MCS14 46 74 80.5 80.5 100.0 0.0 0 0 0 1 1
HT20 SGI 2 MCS15 47 67 85.4 85.4 100.0 0.0 3 0 0 26 26
HT40 LGI 1 MCS0 60 711 9.7 9.7 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS1 61 356 21.9 21.9 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS2 62 237 31.7 31.7 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS3 63 178 41.5 41.5 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS4 64 119 56.1 56.1 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS5 65 89 70.8 70.8 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS6 66 79 75.6 75.6 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS7 67 71 83.0 83.0 100.0 0.0 0 0 0 1 1
HT40 LGI 2 MCS8 70 356 21.9 21.9 100.0 0.0 0 0 0 1 1
HT40 LGI 2 MCS9 71 178 41.5 41.5 100.0 0.0 0 0 0 1 1
HT40 LGI 2 MCS10 72 119 56.1 56.1 100.0 0.0 3 0 0 17 17
HT40 LGI 2 MCS11 73 89 70.8 70.8 100.0 0.0 0 0 0 1 1
HT40 LGI 2 MCS12 74 59 92.7 92.7 100.0 0.0 0 0 0 1 1
HT40 LGI 2 MCS13 75 44 109.8 109.8 100.0 0.0 0 0 0 1 1
HT40 LGI 2 D MCS14 76 40 117.1 117.1 100.0 0.0 0 0 0 1 1
HT40 LGI 2 B MCS15 77 36 124.5 124.5 100.0 0.0 3 0 0 33 33
HT40 SGI 1 MCS0 90 640 12.2 12.2 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS1 91 320 24.4 24.4 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS2 92 214 34.1 34.1 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS3 93 160 43.9 43.9 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS4 94 107 61.0 61.0 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS5 95 80 75.6 75.6 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS6 96 71 83.0 83.0 100.0 0.0 0 0 0 1 1
HT40 SGI 1 P MCS7 97 64 87.8 87.8 100.0 0.0 0 0 0 1 1
HT40 SGI 2 MCS8 100 320 24.4 24.4 100.0 0.0 0 0 0 1 1
HT40 SGI 2 MCS9 101 160 43.9 43.9 100.0 0.0 0 0 0 1 1
HT40 SGI 2 MCS10 102 107 61.0 61.0 100.0 0.0 3 0 0 43 43
HT40 SGI 2 MCS11 103 80 75.6 75.6 100.0 0.0 0 0 0 65 65
HT40 SGI 2 MCS12 104 54 100.0 100.0 100.0 0.0 0 0 0 3 3
HT40 SGI 2 MCS13 105 40 117.1 117.1 97.0 16.6 0 0 0 72 76
HT40 SGI 2 C MCS14 106 36 124.5 124.5 97.4 15.5 0 0 0 1076 1085
HT40 SGI 2 A MCS15 107 32 131.8 131.8 99.9 0.0 3 170 170 93111 93193
Total packet count:: ideal 105481 lookaround 1008
Average # of aggregated frames per A-MPDU: 3.3
During download:
root@LEDE:~# cat /sys/kernel/debug/ieee80211/phy*/netdev*/stations/*/rc_stats
best ____________rate__________ ________statistics________ _____last____ ______sum-of________
mode guard # rate [name idx airtime max_tp] [avg(tp) avg(prob) sd(prob)] [retry|suc|att] [#success | #attempts]
CCK LP 1 1.0M 120 10548 0.0 0.0 100.0 0.0 0 0 0 222 222
CCK LP 1 2.0M 121 5476 0.0 0.0 0.0 0.0 0 0 0 0 0
CCK LP 1 5.5M 122 2411 2.4 0.0 0.0 0.0 0 0 0 0 0
CCK LP 1 11.0M 123 1535 4.8 0.0 0.0 0.0 0 0 0 0 0
CCK SP 1 1.0M 124 10452 0.0 0.0 0.0 0.0 0 0 0 0 0
CCK SP 1 2.0M 125 5380 0.0 0.0 0.0 0.0 0 0 0 0 0
CCK SP 1 5.5M 126 2315 2.4 0.0 0.0 0.0 0 0 0 0 0
CCK SP 1 11.0M 127 1439 4.8 0.0 0.0 0.0 0 0 0 0 0
HT20 LGI 1 MCS0 0 1477 4.8 4.8 100.0 0.0 3 0 0 8 8
HT20 LGI 1 MCS1 1 738 9.7 9.7 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS2 2 492 17.0 17.0 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS3 3 369 21.9 21.9 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS4 4 246 34.1 34.1 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS5 5 185 46.3 46.3 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS6 6 164 51.2 51.2 100.0 0.0 3 0 0 25 25
HT20 LGI 1 MCS7 7 148 56.1 56.1 100.0 0.0 0 0 0 1 1
HT20 LGI 2 MCS8 10 738 9.7 9.7 100.0 0.0 0 0 0 1 1
HT20 LGI 2 MCS9 11 369 21.9 21.9 100.0 0.0 0 0 0 1 1
HT20 LGI 2 MCS10 12 246 34.1 34.1 100.0 0.0 0 0 0 1 1
HT20 LGI 2 MCS11 13 185 46.3 46.3 100.0 0.0 0 0 0 1 1
HT20 LGI 2 MCS12 14 123 68.3 68.3 100.0 0.0 0 0 0 1 1
HT20 LGI 2 MCS13 15 92 87.8 87.8 100.0 0.0 0 0 0 1 1
HT20 LGI 2 MCS14 16 82 100.0 100.0 100.0 0.0 0 0 0 1 1
HT20 LGI 2 MCS15 17 74 109.8 109.8 100.0 0.0 3 0 0 33 33
HT20 SGI 1 MCS0 30 1329 4.8 4.8 100.0 0.0 0 0 0 1 1
HT20 SGI 1 MCS1 31 665 12.2 12.2 100.0 0.0 0 0 0 1 1
HT20 SGI 1 MCS2 32 443 19.5 19.5 100.0 0.0 0 0 0 1 1
HT20 SGI 1 MCS3 33 332 24.4 24.4 100.0 0.0 0 0 0 1 1
HT20 SGI 1 MCS4 34 222 39.0 39.0 100.0 0.0 0 0 0 1 1
HT20 SGI 1 MCS5 35 166 51.2 51.2 100.0 0.0 0 0 0 1 1
HT20 SGI 1 MCS6 36 148 56.1 56.1 100.0 0.0 0 0 0 1 1
HT20 SGI 1 MCS7 37 133 63.4 63.4 100.0 0.0 0 0 0 1 1
HT20 SGI 2 MCS8 40 665 12.2 12.2 100.0 0.0 0 0 0 1 1
HT20 SGI 2 MCS9 41 332 24.4 24.4 100.0 0.0 0 0 0 1 1
HT20 SGI 2 MCS10 42 222 39.0 39.0 100.0 0.0 0 0 0 1 1
HT20 SGI 2 MCS11 43 166 51.2 51.2 100.0 0.0 0 0 0 1 1
HT20 SGI 2 MCS12 44 111 75.6 75.6 100.0 0.0 0 0 0 1 1
HT20 SGI 2 MCS13 45 83 97.6 97.6 100.0 0.0 0 0 0 1 1
HT20 SGI 2 MCS14 46 74 109.8 109.8 100.0 0.0 0 0 0 1 1
HT20 SGI 2 MCS15 47 67 119.6 119.6 100.0 0.0 3 0 0 26 26
HT40 LGI 1 MCS0 60 711 12.2 12.2 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS1 61 356 24.4 24.4 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS2 62 237 36.6 36.6 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS3 63 178 46.3 46.3 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS4 64 119 70.8 70.8 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS5 65 89 92.7 92.7 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS6 66 79 102.5 102.5 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS7 67 71 112.3 112.3 100.0 0.0 0 0 0 1 1
HT40 LGI 2 MCS8 70 356 24.4 24.4 100.0 0.0 0 0 0 1 1
HT40 LGI 2 MCS9 71 178 46.3 46.3 100.0 0.0 0 0 0 1 1
HT40 LGI 2 MCS10 72 119 70.8 70.8 100.0 0.0 3 0 0 17 17
HT40 LGI 2 MCS11 73 89 92.7 92.7 100.0 0.0 0 0 0 1 1
HT40 LGI 2 MCS12 74 59 134.2 134.2 100.0 0.0 0 0 0 1 1
HT40 LGI 2 MCS13 75 44 173.3 173.3 100.0 0.0 0 0 0 1 1
HT40 LGI 2 D MCS14 76 40 190.4 190.4 100.0 0.0 0 0 0 1 1
HT40 LGI 2 B MCS15 77 36 207.5 207.5 100.0 0.0 3 0 0 33 33
HT40 SGI 1 MCS0 90 640 12.2 12.2 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS1 91 320 26.8 26.8 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS2 92 214 39.0 39.0 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS3 93 160 53.7 53.7 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS4 94 107 78.1 78.1 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS5 95 80 102.5 102.5 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS6 96 71 112.3 112.3 100.0 0.0 0 0 0 1 1
HT40 SGI 1 P MCS7 97 64 124.5 124.5 100.0 0.0 0 0 0 1 1
HT40 SGI 2 MCS8 100 320 26.8 26.8 100.0 0.0 0 0 0 1 1
HT40 SGI 2 MCS9 101 160 53.7 53.7 100.0 0.0 0 0 0 1 1
HT40 SGI 2 MCS10 102 107 78.1 78.1 100.0 0.0 3 0 0 43 43
HT40 SGI 2 MCS11 103 80 102.5 102.5 100.0 0.0 0 0 0 1 1
HT40 SGI 2 MCS12 104 54 146.4 146.4 100.0 0.0 0 0 0 1 1
HT40 SGI 2 MCS13 105 40 187.9 187.9 100.0 0.0 0 0 0 1 1
HT40 SGI 2 C MCS14 106 36 207.5 207.5 98.5 11.4 0 0 0 221 222
HT40 SGI 2 A MCS15 107 32 227.0 227.0 99.9 0.0 3 737 737 36270 36280
Total packet count:: ideal 44865 lookaround 307
Average # of aggregated frames per A-MPDU: 15.2
Hm, those stats look normal, rate control and aggregation seems to work fine.
Please do another round of tests by removing the patches again from a recent build. If it works fine again, add back only patch 327.
I've been reading this change a few times and I still don't see how this could be causing performance issues
Now I am getting speeds that are in between the previous good and previous bad results. I do notice that my fresh pull from master includes two new mac80211 commits. Could these be interfering with my test results? Or maybe I am running into Wifi inference from my neighbors that is messing with the reproducibility of my tests. Hmmm...
Edit: I will now flash a previous build before the Minstrel commit itself. Thereby also reverting the latest mac80211 commits. Will double check if the environment has changed in terms of inference.
It is definitely not the RF environment. Unless I am getting extremely unlucky for some reason. The performance on this commit: https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=981cca12b6ce59781d59bda2b42b4ed36f4e37eb
is better than the performance on this commit: https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=fb1be20d63f08a07c63bcd5ffb26d010935dfa61
With the Minstrel patches manually deleted.
How big is the difference in your test there? This still doesn't make any sense to me
My results are now all over the place. Maybe it is RF interference after all. Let's see if I can find a quieter channel to test on for results with a lower variance. If not, it might be worthwhile to rerun my tests on Monday once the Neighbors are gone.
Edit: Up to now, I've been testing with 40mhz channel widths. Maybe it would be a good idea to try out 20mhz instead to reduce the impact of other APs? Or is it possible this issue only shows on 40mhz channels? One way to find out I guess. I'll see if I can come up with a setup with reproducible results.
Okay, sorry for the confusion. The latest commits in master seem to be fine. 20mhz channel width and channel 1 is giving me much tighter results. The Minstrel commit does still seem to affect the throughput.
Pre-minstrel commit:
Latest master commit with minstrels patches manually removed (within margin of error compared to the first screenshot):
Latest master commit with minster patch 326 and 327 included (I tried to only include 327, but that resulted in a compilation error, hence 326 was also included):
For the brave souls who want to test a build with 4.14 kernel and flow offload, here is a test build.
Untested since I am not in the vicinity of the router until tomorrow. Flash at your own risk!
I cannot comment on the offload stuff, but I flashed my DIR-860L with a 4.14 kernel a few days ago and I haven't seen anything strange. Haven't run any performance tests though.
Good to know. My build flashes fine and I haven't noticed anything funky yet.
@All, OP updated with a fresh new build. See OP for details, enjoy!
If you want to try out Flow Offload you can do so by running the following command in an ssh terminal:
iptables -I FORWARD 1 -m conntrack --ctstate RELATED,ESTABLISHED -j FLOWOFFLOAD
If you want to disable Flow Offload again, you can revert the change by running the following command in an ssh terminal:
iptables -D FORWARD -m conntrack --ctstate RELATED,ESTABLISHED -j FLOWOFFLOAD
However, you shouldn't use Flow Oflload in a production environment yet for the reasons listed here
Are you also noticing worse WiFi performance compared to your previous build? As mentioned in a few posts earlier, my 2.4 ghz WiFi performance was cut in halve. Removing the Minstrel patches fixed the issue. Are you seeing the same thing? Or is this issue specific to my particular setup?
Yes, 2.4G is only a half speed of the previous version. I tested on my wndr3800ch and wndr3700 v4.
With minstrel patch on wndr3800ch:
best ____________rate__________ ________statistics________ _____last____ ______sum-of________
mode guard # rate [name idx airtime max_tp] [avg(tp) avg(prob) sd(prob)] [retry|suc|att] [#success | #attempts]
CCK LP 1 1.0M 120 10548 0.0 0.0 99.1 5.6 0 3 3 10711 11661
CCK LP 1 2.0M 121 5476 0.0 0.0 0.0 0.0 0 0 0 0 0
CCK LP 1 5.5M 122 2411 2.4 0.0 0.0 0.0 0 0 0 0 0
CCK LP 1 11.0M 123 1535 4.8 0.0 0.0 0.0 0 0 0 0 0
CCK SP 1 2.0M 125 5380 0.0 0.0 0.0 0.0 0 0 0 0 0
CCK SP 1 5.5M 126 2315 2.4 0.0 0.0 0.0 0 0 0 0 0
CCK SP 1 11.0M 127 1439 4.8 0.0 0.0 0.0 0 0 0 0 0
HT20 LGI 1 MCS0 0 1477 4.8 4.8 100.0 0.0 1 0 0 1 1
HT20 LGI 1 MCS1 1 738 9.7 9.7 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS2 2 492 14.6 14.6 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS3 3 369 19.5 19.5 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS4 4 246 29.2 29.2 100.0 0.0 5 0 0 1 1
HT20 LGI 1 MCS5 5 185 36.6 36.6 95.5 20.4 0 0 0 34 40
HT20 LGI 1 MCS6 6 164 39.0 39.0 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS7 7 148 43.9 43.9 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS0 60 711 9.7 9.7 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS1 61 356 19.5 19.5 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS2 62 237 29.2 29.2 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS3 63 178 36.6 36.6 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS4 64 119 51.2 51.2 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS5 65 89 61.0 61.0 100.0 0.0 6 0 0 1 1
HT40 LGI 1 D MCS6 66 79 65.9 65.9 98.1 12.1 5 0 0 145 255
HT40 LGI 1 B MCS7 67 71 70.8 70.8 99.9 0.0 5 0 0 5257 5769
HT40 SGI 1 MCS0 90 640 12.2 12.2 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS1 91 320 21.9 21.9 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS2 92 214 31.7 31.7 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS3 93 160 41.5 41.5 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS4 94 107 53.7 53.7 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS5 95 80 65.9 65.9 95.2 19.9 5 0 0 707 833
HT40 SGI 1 C MCS6 96 71 70.8 70.8 95.4 18.9 5 0 0 3986 4533
HT40 SGI 1 A P MCS7 97 64 75.6 75.6 97.3 2.2 6 87 87 487390 513133
Total packet count:: ideal 496366 lookaround 11920
Average # of aggregated frames per A-MPDU: 2.8
Without minstrel patch on wndr3800ch:
best ____________rate__________ ________statistics________ _____last____ ______sum-of________
mode guard # rate [name idx airtime max_tp] [avg(tp) avg(prob) sd(prob)] [retry|suc|att] [#success | #attempts]
CCK LP 1 1.0M 120 10548 0.0 0.0 79.7 25.2 0 0 0 76 87
CCK LP 1 2.0M 121 5476 0.0 0.0 100.0 0.0 0 0 0 1 1
CCK LP 1 5.5M 122 2411 2.4 2.4 100.0 0.0 0 0 0 1 1
CCK LP 1 11.0M 123 1535 4.8 4.8 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS0 0 1477 4.8 4.8 100.0 0.0 1 0 0 1 1
HT20 LGI 1 MCS1 1 739 9.7 9.7 95.5 14.4 0 0 0 39 42
HT20 LGI 1 MCS2 2 493 17.0 17.0 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS3 3 369 21.9 21.9 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS4 4 246 34.1 34.1 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS5 5 185 43.9 43.9 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS6 6 164 48.8 48.8 100.0 0.0 0 0 0 1 1
HT20 LGI 1 MCS7 7 148 56.1 56.1 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS0 60 711 12.2 12.2 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS1 61 356 24.4 24.4 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS2 62 237 34.1 34.1 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS3 63 178 46.3 46.3 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS4 64 119 68.3 68.3 100.0 0.0 0 0 0 1 1
HT40 LGI 1 MCS5 65 89 87.8 87.8 100.0 0.0 0 0 0 1 1
HT40 LGI 1 D MCS6 66 79 97.6 97.6 99.5 0.0 5 0 0 221 224
HT40 LGI 1 B MCS7 67 71 107.4 107.4 99.2 3.1 5 0 0 4322 4810
HT40 SGI 1 MCS0 90 640 12.2 12.2 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS1 91 320 26.8 26.8 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS2 92 214 39.0 39.0 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS3 93 160 51.2 51.2 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS4 94 107 73.2 73.2 100.0 0.0 0 0 0 1 1
HT40 SGI 1 MCS5 95 80 97.6 97.6 95.4 20.4 5 0 0 213 219
HT40 SGI 1 C MCS6 96 71 107.4 107.4 95.7 12.5 5 0 0 2029 2315
HT40 SGI 1 A P MCS7 97 64 117.1 117.1 96.0 3.4 5 369 370 1027261 1107439
Total packet count:: ideal 1033663 lookaround 559
Average # of aggregated frames per A-MPDU: 9.2
Testing wireless performance on r6302 with iperf3 running for 60 seconds yields the following results:
2.4 GHz band connecting at 144 Mbps:
[ ID] Interval Transfer Bandwidth
[ 4] 0.00-60.00 sec 418 MBytes 58.4 Mbits/sec sender
[ 4] 0.00-60.00 sec 418 MBytes 58.4 Mbits/sec receiver
5 GHz band connecting at 866.7 Mbps:
[ ID] Interval Transfer Bandwidth
[ 4] 0.00-60.00 sec 657 MBytes 91.8 Mbits/sec sender
[ 4] 0.00-60.00 sec 657 MBytes 91.8 Mbits/sec receiver
Compared to other builds:
Build | Band (GHz) | Speed (Mbits/sec) |
---|---|---|
r6302 | 2.4 | 58.4 |
r6302 | 5 | 91.8 |
r6150 | 2.4 | 56.7 |
r6150 | 5 | 107 |
r6009 | 2.4 | 43.7 |
r6009 | 5 | 40.7 |
r5442 | 2.4 | 73.5 |
r5442 | 5 | 57.4 |
To keep it short, 2.4 GHz performance seems unchanged but down from r5442. 5 GHz performs worse then on previous build (r6150) but performs better then previous builds.
Since there have been more mac80211 commits what is the build number of the build you are testing?
@nbd Another person reporting a roughly 50% speed cut with the Minstrel patches. Yet @Bartvz reports unchanged performance. Perhaps it is client dependent? Is there anything else we can do to help pinpoint the issue further?
It's my custom build with the latest master source (2018-02-25).
I applied the fast path patch and some small modification.
Please refer my github repository, use the master-custbuild branch:
@sotux
Hi,I'm interesting to test your custom build for fast patch
so I just select kmod-ipt-offload then put "iptables -D FORWARD -m conntrack --ctstate RELATED,ESTABLISHED -j FLOWOFFLOAD" to firewall.user ? is that right ?
thank you