On Tue, Oct 20, 2009 at 09:13:40PM +0900, thus spake 海藻敬之: > Hi, > > I had and still have similar problem as Ignacy reported. > I tried "iwconfig rate" as Nick mentioned , but huge number of the > AR5K_RXERR_CRC > occurred and most of the packets were dropped when the rate was 48Mbps > and 54Mbps. > When set the rate to 36Mbps or less, then then the iperf throughput > got up to 14Mbps. > ( I posted more detail data to this ML just a week before ) > > I used compat-wireless-2009-0820 and did not try yet with the head of the > trunk. > > Takayuki Kaiso
Incidently, I just tested a few times with the ahdemo (only available in madwifi, unfortunately) and observed a few strange things. At some point, one of the machine was able to send at over 20 Mbps, while the others were still stuck at approx. 5 Mbps. Looking at the minstrel stats, the "fast" machine was indeed sending mostly at 48 and 54 Mbps, while the "slow" ones were sending mostly at 11 Mbps, apparently higher bitrates had too high a loss rate for them. When forcing the bit rate to 54 Mbps on the slow machines, the packets were lost altogether. When I tried to repeat the experiment, the fast machine was initially stuck (i.e. slow) too, but suddently started to be fast again (unfortunately I wasn't monitoring the medium at that time). Now I'm letting the experiment run again between the two formerly slow machines and the throughput stays at around 5 Mbps (bitrate mostly at 11 Mbps according to minstrel stats). Another strange thing, though I don't know if it's at all significant, is that the fast machine is AR5424 according to madwifi, but AR5414 according to ath5k. -- /* This is not a comment */ _______________________________________________ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel