On Wed, Mar 17, 2021 at 10:33 PM Timothy McDaniel <timothy.mcdan...@intel.com> wrote: > > Convert code to use x86 vector instructions, thereby significantly > improving dequeue performance. > > Signed-off-by: Timothy McDaniel <timothy.mcdan...@intel.com> > --- > config/rte_config.h | 1 + > drivers/event/dlb2/dlb2.c | 607 > ++++++++++++++++++++++++++++++++++++---- > drivers/event/dlb2/dlb2_priv.h | 19 +- > 3 files changed, 574 insertions(+), 53 deletions(-) > > diff --git a/config/rte_config.h b/config/rte_config.h > index aedb68c..133ca35 100644 > --- a/config/rte_config.h > +++ b/config/rte_config.h > @@ -144,5 +144,6 @@ > #undef RTE_LIBRTE_PMD_DLB2_QUELL_STATS > #define RTE_LIBRTE_PMD_DLB2_SW_CREDIT_QUANTA 32 > #define RTE_PMD_DLB2_DEFAULT_DEPTH_THRESH 256 > +#define RTE_LIBRTE_PMD_DLB2_VECTOR_CODE 1
This is not required. Please expose as devargs option.
