Hi Stephen, On Tue, Jun 12, 2012 at 7:21 PM, Stephen Donecker <[email protected]> wrote: > Chun-Yeow, > > Did you check the rate used during packet transmission? Look at the > rc_stats file in /sys/kernel/debug/ieee80211 for you station. > > If you test using iperf, maybe try to use multiple stream. > > I conducted an iperf test between STA1 and STA2 configured to use a single > stream (MCS7). While the iperf was running I cat the rc_stats from each STA. > Again you can see that due to the large datagram loss the throughput is > ~78Mbps instead of ~125Mbps. I could use help interpreting the rc_stats > output. > > On STA1: > # iperf -c 192.168.11.2 -p 7777 -u -b 150M -t 10 -i 1 > ------------------------------------------------------------ > Client connecting to 192.168.11.2, UDP port 7777 > > Sending 1470 byte datagrams > UDP buffer size: 160 KByte (default) > ------------------------------------------------------------ > [ 3] local 192.168.11.1 port 59448 connected with 192.168.11.2 port 7777 > [ ID] Interval Transfer Bandwidth > [ 3] 0.0- 1.0 sec 15.1 MBytes 127 Mbits/sec > [ 3] 1.0- 2.0 sec 15.2 MBytes 128 Mbits/sec > [ 3] 2.0- 3.0 sec 14.9 MBytes 125 Mbits/sec > [ 3] 3.0- 4.0 sec 15.1 MBytes 126 Mbits/sec > [ 3] 4.0- 5.0 sec 15.0 MBytes 125 Mbits/sec > [ 3] 5.0- 6.0 sec 15.1 MBytes 127 Mbits/sec > [ 3] 6.0- 7.0 sec 15.1 MBytes 126 Mbits/sec > [ 3] 7.0- 8.0 sec 14.9 MBytes 125 Mbits/sec > [ 3] 8.0- 9.0 sec 15.1 MBytes 127 Mbits/sec > [ 3] 9.0-10.0 sec 15.1 MBytes 127 Mbits/sec > [ 3] 0.0-10.0 sec 151 MBytes 126 Mbits/sec > [ 3] Sent 107423 datagrams > [ 3] Server Report: > [ 3] 0.0-10.0 sec 93.0 MBytes 77.8 Mbits/sec 0.259 ms 41090/107424 > (38%) > > # cat > /sys/kernel/debug/ieee80211/phy0/netdev\:wlan0/stations/00\:15\:6d\:85\:0f\:78/rc_stats > type rate throughput ewma prob this prob this succ/attempt > success attempts > HT20/LGI MCS0 6.6 100.0 100.0 0( 0) > 252 252 > HT20/LGI MCS1 13.0 100.0 100.0 0( 0) > 260 260 > HT20/LGI MCS2 19.1 99.9 100.0 0( 0) > 238 239 > HT20/LGI MCS3 25.0 100.0 100.0 0( 0) > 248 248 > HT20/LGI MCS4 36.3 99.9 100.0 0( 0) > 243 244 > HT20/LGI MCS5 46.5 100.0 100.0 0( 0) > 267 267 > HT20/LGI MCS6 51.2 100.0 100.0 0( 0) > 250 250 > HT20/LGI MCS7 57.1 100.0 100.0 0( 0) > 248 248 > HT40/LGI MCS0 13.5 100.0 100.0 0( 0) > 267 267 > HT40/LGI MCS1 26.1 100.0 100.0 0( 0) > 244 244 > HT40/LGI MCS2 37.4 99.9 100.0 0( 0) > 248 249 > HT40/LGI MCS3 48.2 99.9 100.0 0( 0) > 251 252 > HT40/LGI MCS4 68.0 100.0 100.0 0( 0) > 250 250 > HT40/LGI MCS5 84.0 100.0 100.0 0( 0) > 246 246 > HT40/LGI MCS6 93.4 99.9 100.0 0( 0) > 237 238 > HT40/LGI t MCS7 101.0 100.0 100.0 0( 0) > 256 256 > HT40/SGI MCS0 14.9 100.0 100.0 0( 0) > 250 250 > HT40/SGI MCS1 28.7 100.0 100.0 0( 0) > 241 241 > HT40/SGI MCS2 41.1 100.0 100.0 0( 0) > 245 245 > HT40/SGI MCS3 52.9 99.9 100.0 0( 0) > 552 553 > HT40/SGI MCS4 74.0 100.0 100.0 0( 0) > 241 241 > HT40/SGI MCS5 90.9 100.0 100.0 0( 0) > 258 258 > HT40/SGI MCS6 97.6 96.6 100.0 0( 0) > 238 239 > HT40/SGI T PMCS7 108.6 99.9 100.0 405(405) > 1686665 1690372 > > Total packet count:: ideal 1687108 lookaround 5730 > Average A-MPDU length: 4.9 > > On STA2: > # cat > /sys/kernel/debug/ieee80211/phy0/netdev\:wlan0/stations/00\:15\:6d\:85\:02\:de/rc_stats > type rate throughput ewma prob this prob this succ/attempt > success attempts > HT20/LGI MCS0 6.2 100.0 100.0 0( 0) > 9 9 > HT20/LGI MCS1 11.7 100.0 100.0 0( 0) > 8 8 > HT20/LGI MCS2 16.5 100.0 100.0 0( 0) > 9 9 > HT20/LGI MCS3 20.8 100.0 100.0 0( 0) > 8 8 > HT20/LGI MCS4 28.0 100.0 100.0 0( 0) > 9 9 > HT20/LGI MCS5 33.7 100.0 100.0 0( 0) > 9 9 > HT20/LGI MCS6 36.2 100.0 100.0 0( 0) > 10 10 > HT20/LGI MCS7 39.0 100.0 100.0 0( 0) > 296 296 > HT40/LGI MCS0 12.1 100.0 100.0 0( 0) > 8 8 > HT40/LGI MCS1 21.5 100.0 100.0 0( 0) > 10 10 > HT40/LGI MCS2 28.7 100.0 100.0 0( 0) > 8 8 > HT40/LGI MCS3 34.7 100.0 100.0 0( 0) > 8 8 > HT40/LGI MCS4 43.8 100.0 100.0 0( 0) > 10 10 > HT40/LGI MCS5 50.0 100.0 100.0 0( 0) > 9 9 > HT40/LGI MCS6 53.1 100.0 100.0 0( 0) > 10 10 > HT40/LGI t MCS7 55.5 100.0 100.0 0( 0) > 9 9 > HT40/SGI MCS0 13.3 100.0 100.0 0( 0) > 8 8 > HT40/SGI MCS1 23.3 100.0 100.0 0( 0) > 10 10 > HT40/SGI MCS2 30.8 100.0 100.0 0( 0) > 8 8 > HT40/SGI MCS3 37.0 100.0 100.0 0( 0) > 8 8 > HT40/SGI MCS4 46.2 100.0 100.0 0( 0) > 8 8 > HT40/SGI MCS5 52.3 100.0 100.0 0( 0) > 8 8 > HT40/SGI MCS6 55.5 99.9 100.0 0( 0) > 559 561 > HT40/SGI T PMCS7 57.7 99.8 100.0 1( 1) > 64887 65261 > > Total packet count:: ideal 65723 lookaround 203 > Average A-MPDU length: 1.3
Based on this format, it looks like you are using the minstrel_ht rate control algorithm (cat /<debugfs>/<phy>/rc_name to confirm). At least as far as throughput is concerned, the ath9k rate control seems to perform better. Enable with CONFIG_ATH9K_RATE_CONTROL=y. Thomas _______________________________________________ Devel mailing list [email protected] http://lists.open80211s.org/cgi-bin/mailman/listinfo/devel
