I just noticed this email. It sounds beneficial, but no patch was
included. Do you want to send the patch?
On Wed, Sep 07, 2016 at 08:58:02PM +, Rodriguez Betancourt, Esteban wrote:
> The existing implementation uses a switch with
> many conditions, that when compiled is translated
> to a not optimal series of conditional jumps.
>
> With a lookup table the generated code has less conditional jumps,
> that should translate in improving the CPU ability to predict the
> jumps.
>
> Performance Comparison:
> All the timings are in nanoseconds, "OVS Master" corresponds to 13a1d36.
> N is the number of repetitions
>
> Serialize vswitch.ovsschema
> NOVS Master Lookup TableDifferenceDiff per op
> 1 233182200369 3281332813
> 10 2724931 1919168 80576380576.3
> 100 22802794 24406648-1603854 -16038.54
> 1000 253645888 2062597604738612847386.128
> 1 23522457031906839780 44540592344540.5923
> 10 23967770920 19012178655 495559226549555.92265
>
> Serialize echo example
> NOVS Master Lookup TableDifferenceDiff per op
> 1385712565 -8708-8708
> 10 17403 7312 10091 1009.1
> 100 5785956613 1246 12.46
> 1000 592310 528110 64200 64.2
> 1 6096334 5576109520225 52.0225
> 10 60970439 58477626 2492813 24.92813
>
> Serialize mutate example
> NOVS Master Lookup TableDifferenceDiff per op
> 17115 19051 -11936 -11936
> 10 34110 39561 -5451-545.1
> 100296613 298645 -2032-20.32
> 1000 35104992930588 579911579.911
> 133898710 302786313620079362.0079
> 10 305069356 280622992 24446364244.46364
>
>
> Signed-off-by: Esteban Rodriguez Betancourt
> ___
> dev mailing list
> dev@openvswitch.org
> http://openvswitch.org/mailman/listinfo/dev
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev