> From: Neil Horman [mailto:nhorman at tuxdriver.com] > Sent: Thursday, January 29, 2015 8:40 PM > To: Gonzalez Monroy, Sergio > Cc: Thomas Monjalon; Pattan, Reshma; dev at dpdk.org > Subject: Re: [dpdk-dev] [PATCH 1/3] librte_reorder: New reorder library > > On Thu, Jan 29, 2015 at 05:35:09PM +0000, Gonzalez Monroy, Sergio wrote: > > Hi Thomas, > > > > > From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Thomas > Monjalon > > > Sent: Tuesday, January 20, 2015 8:01 AM > > > To: Pattan, Reshma > > > Cc: dev at dpdk.org > > > Subject: Re: [dpdk-dev] [PATCH 1/3] librte_reorder: New reorder > > > library > > > > > > Hi, > > > > > > 2015-01-07 16:39, Reshma Pattan: > > > > 1)New library to provide reordering of out of ordered > > > > mbufs based on sequence number of mbuf. Library uses > > > > reorder > > > buffer structure > > > > which in tern uses two circular buffers called ready and > > > > order > buffers. > > > > *rte_reorder_create API creates instance of reorder buffer. > > > > *rte_reorder_init API initializes given reorder buffer > > > > instance. > > > > *rte_reorder_reset API resets given reorder buffer instance. > > > > *rte_reorder_insert API inserts the mbuf into order circular > buffer. > > > > *rte_reorder_fill_overflow moves mbufs from order > > > > buffer to ready > > > buffer > > > > to accomodate early packets in order buffer. > > > > *rte_reorder_drain API provides draining facility to fetch > > > > out > > > > reordered mbufs from order and ready buffers. > > > > > > > > Signed-off-by: Reshma Pattan <reshma.pattan at intel.com> > > > > Signed-off-by: Richardson Bruce > > > > <bruce.richardson at intel.com> > > > > > > I think 2 things are missing in this patchset: > > > > > > 1) Could you show some performance numbers to compare a simple > > > forwarding with and without this library, in the commit log? > > > > > I'm not allowed to provide specific performance numbers. > Can you elaborate on this? Why can you not provide specific performance > numbers from your testing? Is there some concern over the validity of the > measurements?
Hi Neil, As far as I know, that is exactly the reason. Any Intel specific performance data goes through a performance validation team before being released. Thanks, Sergio