[dpdk-dev] Ability to/impact of running with smaller page sizes

[Ananyev, Konstantin]

Hi,

> Hi Matt,
> 
> On Mon, Jun 30, 2014 at 05:43:39PM -0500, Matt Laswell wrote:
> > Hey Folks,
> >
> > In my application, I'm seeing some design considerations in a project
> > I'm working on that push me towards the use of smaller memory page
> > sizes.  I'm curious - is it possible in practical terms to run DPDK without hugepages?
> 
> >  If so, does anybody have any practical experience (or a
> > back-of-the-envelop estimate) of how badly such a configuration would
> > hurt performance?  For sake of argument, assume that virtually all of
> > the memory being used is in pre-allocated mempools (e.g lots of
> > rte_mempool_create(), very little rte_malloc().
> >
> 
> There is an case to run DPDK without hugepages in DPDK source codes, which is DPDK supports Xen Dom0. for this, we developed a
> dom0_mm driver.
> Except for Xen Dom0, it is impossible to run DPDK without hugepages without any changes at memory initialization phase, but current
> rte_memzone_reserve_bounded() and
> rte_mempool_xmem_create()  implemention have already support non-hugepage usage in DPDK.
> 

On linux testpmd can run with mempool on 4K pages (though RX/TX HW rings are still on hugepages).
To try it add "--mp-anon" to your testpmd command-line.
Also to get a more 'real' picture you can disable MP caching: "--mbcache=0".   

Konstantin