[rt2x00-users] rt2800 & minstrel
Ivo Van Doorn
ivdoorn at gmail.com
Mon Aug 9 18:37:15 UTC 2010
Hi,
>> /sys/kernel/debug/ieee80211/phy*/stations/*/rc_stats
>>
>> contain for information? And what does it contain when the link
>> quality decreases and different rates are being tried?
>> During my test with rt2800usb it seems that only rates
>> 1Mbs, 18Mbs and 54Mbs seem to work reliable, while the rates
>> between 18Mbs and 54Mbs show large failures (1 frame out of 80 succeeds)
>>
>> Just an impression of the data from my test:
>>
>> rate throughput ewma prob this prob this succ/attempt
>> success attempts
>> t 1 0.8 93.7 100.0 0( 0) 19 20
>> 2 0.0 0.0 0.0 0( 0) 0 0
>> 5.5 0.0 0.0 0.0 0( 0) 0 0
>> 11 0.0 0.0 0.0 0( 0) 0 3
>> 6 0.0 0.0 0.0 0( 0) 0 3
>> 9 0.0 0.0 0.0 0( 0) 0 3
>> 12 0.0 0.0 0.0 0( 0) 0 3
>> 18 0.0 0.0 0.0 0( 0) 153 236
>> 24 0.0 0.0 0.0 0( 0) 1 81
>> 36 0.0 0.0 0.0 0( 0) 1 81
>> 48 0.0 0.0 0.0 0( 0) 1 81
>> T P 54 43.3 99.9 100.0 0( 0) 1082 1162
>>
>> Total packet count:: ideal 1202 lookaround 133
>
>
> That looks quite strange. I mean 24, 36 and 48 were tried 81 times but only 1
> success?
Yeah, it is quite weird, especially since 54 and 18 behave reasonably well.
>> It sounds to me that it is a bit weird, that 24, 36 and 48Mbs show that
>> high failure counts while 18Mbs is suddenly performing so well again..
>>
>> I also got a report from somebody else which states that for 54Mbs
>> 27190 out of 27381 frames succeeded, for 48Mbs it was 756 out of
>> 931. Every rate below that always failed (11Mbs was attempted 76 times
>> without success, while 1Mbs was tried 18 times without success).
>
> Indeed, I get similar results, not as strange as yours but a trend is visible:
>
> rate throughput ewma prob this prob this succ/attempt success attempts
> 1 0.9 95.7 100.0 0( 0) 44 44
> 2 0.0 0.0 0.0 0( 0) 0 0
> 5.5 1.2 25.0 100.0 0( 0) 1 1
> 11 0.0 0.0 0.0 0( 0) 0 0
> 6 0.0 0.0 0.0 0( 0) 0 0
> 9 0.0 0.0 0.0 0( 0) 0 5
> 12 0.1 1.4 0.0 0( 0) 1 15
> 18 1.0 6.3 0.0 0( 0) 6 68
> 24 2.4 11.7 0.0 0( 0) 160 248
> 36 10.6 34.8 0.0 0( 0) 185 318
> t 48 37.2 95.2 100.0 0( 0) 1543 1806
> T P 54 43.2 99.7 100.0 146(146) 128718 129436
>
> However, I cannot spot such an issue with minstrel_ht:
>
> type rate throughput ewma prob this prob this succ/attempt success attempts
> HT20/LGI MCS0 6.2 100.0 100.0 0( 0) 23 23
> HT20/LGI MCS1 11.7 100.0 100.0 0( 0) 20 20
> HT20/LGI MCS2 16.5 100.0 100.0 0( 0) 23 23
> HT20/LGI MCS3 20.8 100.0 100.0 0( 0) 24 24
> HT20/LGI MCS4 28.0 99.9 100.0 0( 0) 38 39
> HT20/LGI MCS5 33.7 100.0 100.0 0( 0) 24 24
> HT20/LGI MCS6 36.2 100.0 100.0 0( 0) 21 21
> HT20/LGI MCS7 39.0 100.0 100.0 0( 0) 23 23
> HT20/LGI MCS8 11.7 100.0 100.0 0( 0) 24 24
> HT20/LGI MCS9 20.7 99.8 100.0 0( 0) 21 22
> HT20/LGI MCS10 28.0 99.9 100.0 0( 0) 25 26
> HT20/LGI MCS11 33.7 100.0 100.0 0( 0) 22 22
> HT20/LGI MCS12 43.1 100.0 100.0 1( 1) 24 24
> HT20/LGI MCS13 36.7 75.0 0.0 0( 1) 31 32
> HT20/LGI MCS14 52.0 100.0 100.0 0( 0) 23 23
> HT20/LGI MCS15 54.3 100.0 100.0 1( 1) 30 30
> HT20/SGI MCS0 6.9 100.0 100.0 0( 0) 24 24
> HT20/SGI MCS1 12.9 100.0 100.0 0( 0) 20 20
> HT20/SGI MCS2 18.0 100.0 100.0 0( 0) 24 24
> HT20/SGI MCS3 22.5 100.0 100.0 0( 0) 24 24
> HT20/SGI MCS4 30.1 100.0 100.0 0( 0) 23 23
> HT20/SGI MCS5 35.9 100.0 100.0 0( 0) 25 25
> HT20/SGI MCS6 38.4 99.8 100.0 0( 0) 21 22
> HT20/SGI PMCS7 41.3 100.0 100.0 1( 1) 23 23
> HT20/SGI MCS8 12.9 100.0 100.0 0( 0) 23 23
> HT20/SGI MCS9 22.5 100.0 100.0 0( 0) 19 19
> HT20/SGI MCS10 30.1 100.0 100.0 0( 0) 24 24
> HT20/SGI MCS11 35.9 100.0 100.0 0( 0) 23 23
> HT20/SGI MCS12 44.3 97.4 100.0 0( 0) 20 21
> HT20/SGI MCS13 49.9 97.4 100.0 0( 0) 29 30
> HT20/SGI t MCS14 53.7 98.8 100.0 0( 0) 46 48
> HT20/SGI T MCS15 55.8 98.8 100.0 18( 18) 5568 5593
Ok so maybe we are not handling the 11bg rates correctly yet. Even
though there is
little what can be wrong with it (the MCS to CCK/OFDM mapping is
simple, and that
is the only difference). What do you normally use for testing?
minstrel or minstrel_ht?
Thanks for the quick test.
Ivo
More information about the users
mailing list