> -----Original Message-----
> From: Jerin Jacob <[email protected]>
> Sent: Wednesday, June 7, 2023 4:15 PM
> To: Yan, Zhirun <[email protected]>
> Cc: [email protected]; [email protected]; [email protected];
> [email protected]; [email protected];
> [email protected]; Liang, Cunming <[email protected]>; Wang,
> Haiyue <[email protected]>; mattias.ronnblom
> <[email protected]>
> Subject: Re: [PATCH v9 13/17] graph: enable graph multicore dispatch scheduler
> model
>
> On Wed, Jun 7, 2023 at 9:30 AM Zhirun Yan <[email protected]> wrote:
> >
> > This patch enables to chose new scheduler model. Must define
> > RTE_GRAPH_MODEL_SELECT before including rte_graph_worker.h to enable
> > specific model choosing.
> >
> > Signed-off-by: Haiyue Wang <[email protected]>
> > Signed-off-by: Cunming Liang <[email protected]>
> > Signed-off-by: Zhirun Yan <[email protected]>
> > ---
> > lib/graph/rte_graph_worker.h | 12 ++++++++++++
> > 1 file changed, 12 insertions(+)
> >
>
> > rte_graph_walk(struct rte_graph *graph) {
> > +#if !defined(RTE_GRAPH_MODEL_SELECT) || RTE_GRAPH_MODEL_SELECT
> ==
> > +RTE_GRAPH_MODEL_RTC
>
> Is nt defined instead of !defined?
>
!defined(XX) means not defined XX.
What is nt defined means?
> Use bracket around RTE_GRAPH_MODEL_SELECT == RTE_GRAPH_MODEL_RTC.
>
Ok.
>
> > rte_graph_walk_rtc(graph);
> > +#elif defined(RTE_GRAPH_MODEL_SELECT) && RTE_GRAPH_MODEL_SELECT
> ==
> > +RTE_GRAPH_MODEL_MCORE_DISPATCH
>
> Use bracket around RTE_GRAPH_MODEL_SELECT ==
Ok.
>
> > + rte_graph_walk_mcore_dispatch(graph);
> > +#else
> > + int model = rte_graph_worker_model_get(graph);
>
> Introduce rte_graph_worker_model_no_check_get() as commented earlier.
Got it.
> > +
> > + if (model == RTE_GRAPH_MODEL_RTC)
> > + rte_graph_walk_rtc(graph);
> > + else if (model == RTE_GRAPH_MODEL_MCORE_DISPATCH)
> > + rte_graph_walk_mcore_dispatch(graph);
>
> I think, switch case better to support new model in future. Please check the
> performance before changing.
>
Yes, I agree.
And I checked the performance with switch case and
rte_graph_worker_model_no_check_get()
I get very similar performance. The improvements is <0.1%. I guess the
performance impact will
be less if the workerload goes more complicated, cause the node process in walk
will spent more
time.
And I will change in next version.
> i.e
>
> switch ( rte_graph_worker_model_no_check_get())
> {
> case RTE_GRAPH_MODEL_MCORE_DISPATCH:
> rte_graph_walk_mcore_dispatch(graph)
> break;
> default:
> rte_graph_walk_rtc(graph);
> }
