User bus conclusion

[Alexander Neundorf]

On Wednesday 10 November 2010, you wrote:
> Em Quarta-feira, 10 de Novembro de 2010, às 18:14:31, Alexander Neundorf
>
> escreveu:
> > I'm not sure it reaches the "fast" part of this goal...
>
> QDEBUG : tst_QDBusPerformance::oneWay(normal:256-byteArray) 249344 bytes in
> 500 ms (in 974 calls): 0.475586 MB/s
> QDEBUG : tst_QDBusPerformance::oneWay(normal:1024-byteArray) 1167360 bytes
> in 500 ms (in 1140 calls): 2.22656 MB/s
> QDEBUG : tst_QDBusPerformance::oneWay(normal:4096-byteArray) 3551232 bytes
> in 500 ms (in 867 calls): 6.77344 MB/s
> QDEBUG : tst_QDBusPerformance::roundTrip(normal:256-byteArray) 286464 bytes
> in 500 ms (in 1119 calls): 0.546387 MB/s
> QDEBUG : tst_QDBusPerformance::roundTrip(normal:1024-byteArray) 1010688
> bytes in 500 ms (in 987 calls): 1.92773 MB/s
> QDEBUG : tst_QDBusPerformance::roundTrip(normal:4096-byteArray) 3174400
> bytes in 500 ms (in 775 calls): 6.05469 MB/s
>
> Peak performance at 0.44 ms at roundtrip. Throughput increases with amount
> of data transferred per call.
>
> If you're making anywhere near 100 calls per second, you're doing something
> wrong.

6 MB/s for local communication is anything but fast.
100 calls/s is really not much (not in terms of general dbus use-cases, but if 
it claims to be fast).

Alex