On Wed, Jun 30, 2010 at 05:28:05PM +0200, Clemens Ladisch wrote: > >> latency distribution: > > ... > > 3.1 - 3.2 ms: 1 # > > ... > > 3.9 - 4.0 ms: 1 # > > 4.0 - 4.1 ms: 9903 ################################################## > > ... > > 5.0 - 5.1 ms: 95 # > > The default parameters of this tool are unrealistic; the next MIDI > command is always sent immediately after the previous one has been > received, which tends to align everything to USB frames. Please use > the -w and -r options.
Absolutely. I already suspected 4ms being correlated to USB frames (granularity 1ms, right?) With random wait, it looks like this: > latency distribution: ... 3.0 - 3.1 ms: 86 #### 3.1 - 3.2 ms: 986 ################################################## 3.2 - 3.3 ms: 958 ################################################# 3.3 - 3.4 ms: 967 ################################################# 3.4 - 3.5 ms: 985 ################################################## 3.5 - 3.6 ms: 927 ############################################### 3.6 - 3.7 ms: 966 ################################################# 3.7 - 3.8 ms: 938 ################################################ 3.8 - 3.9 ms: 980 ################################################## 3.9 - 4.0 ms: 904 ############################################## 4.0 - 4.1 ms: 906 ############################################## 4.1 - 4.2 ms: 45 ## 4.2 - 4.3 ms: 44 ## 4.3 - 4.4 ms: 38 ## 4.4 - 4.5 ms: 28 # 4.5 - 4.6 ms: 37 ## 4.6 - 4.7 ms: 34 ## 4.7 - 4.8 ms: 54 ### 4.8 - 4.9 ms: 34 ## 4.9 - 5.0 ms: 42 ## 5.0 - 5.1 ms: 40 ## ... 5.2 - 5.3 ms: 1 # Probably still acceptable, given that the majority of packets deviates by only 1ms. Does USB have a notion of "ticks" (clock?), that is, frames are sent at isochronous distances? Or is the host controller free to send a packet whenever it wants to? Without the latter, 1ms jitter could be the result of queueing up the data for the next time slot. TIA -- mail: [email protected] http://adi.thur.de PGP/GPG: key via keyserver _______________________________________________ Linux-audio-dev mailing list [email protected] http://lists.linuxaudio.org/listinfo/linux-audio-dev
