From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Ananyev, Konstantin" Subject: Re: [PATCH] ring: relax alignment constraint on ring structure Date: Thu, 5 Apr 2018 13:49:02 +0000 Message-ID: <2601191342CEEE43887BDE71AB977258A0AB9930@irsmsx105.ger.corp.intel.com> References: <20170630142609.6180-1-olivier.matz@6wind.com> <20180403132644.23729-1-olivier.matz@6wind.com> <20180403150722.GB15937@jerin> <20180403152517.hsjghkj5z6mauze7@platinum> <20180403153703.GA19072@jerin> <20180403155601.rqb7fhu6vggzrh7e@platinum> <20180403163254.GB19072@jerin> <2601191342CEEE43887BDE71AB977258A0AB90E3@irsmsx105.ger.corp.intel.com> <20180405080134.GA2674@jerin> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Cc: Olivier Matz , "dev@dpdk.org" , "Richardson, Bruce" To: Jerin Jacob Return-path: Received: from mga09.intel.com (mga09.intel.com [134.134.136.24]) by dpdk.org (Postfix) with ESMTP id E1EE11CBB4 for ; Thu, 5 Apr 2018 15:49:06 +0200 (CEST) In-Reply-To: <20180405080134.GA2674@jerin> Content-Language: en-US List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" Hi Jerin, > -----Original Message----- > From: Jerin Jacob [mailto:jerin.jacob@caviumnetworks.com] > Sent: Thursday, April 5, 2018 9:02 AM > To: Ananyev, Konstantin > Cc: Olivier Matz ; dev@dpdk.org; Richardson, Bruc= e > Subject: Re: [dpdk-dev] [PATCH] ring: relax alignment constraint on ring = structure >=20 > -----Original Message----- > > Date: Wed, 4 Apr 2018 23:38:41 +0000 > > From: "Ananyev, Konstantin" > > To: Jerin Jacob , Olivier Matz > > > > CC: "dev@dpdk.org" , "Richardson, Bruce" > > > > Subject: RE: [dpdk-dev] [PATCH] ring: relax alignment constraint on rin= g > > structure > > > > Hi lads, > > > > > -----Original Message----- > > > From: Jerin Jacob [mailto:jerin.jacob@caviumnetworks.com] > > > Sent: Tuesday, April 3, 2018 5:43 PM > > > To: Olivier Matz > > > Cc: dev@dpdk.org; Ananyev, Konstantin ;= Richardson, Bruce > > > Subject: Re: [dpdk-dev] [PATCH] ring: relax alignment constraint on r= ing structure > > > > > > -----Original Message----- > > > > Date: Tue, 3 Apr 2018 17:56:01 +0200 > > > > From: Olivier Matz > > > > To: Jerin Jacob > > > > CC: dev@dpdk.org, konstantin.ananyev@intel.com, bruce.richardson@in= tel.com > > > > Subject: Re: [dpdk-dev] [PATCH] ring: relax alignment constraint on= ring > > > > structure > > > > User-Agent: NeoMutt/20170113 (1.7.2) > > > > > > > > On Tue, Apr 03, 2018 at 09:07:04PM +0530, Jerin Jacob wrote: > > > > > -----Original Message----- > > > > > > Date: Tue, 3 Apr 2018 17:25:17 +0200 > > > > > > From: Olivier Matz > > > > > > To: Jerin Jacob > > > > > > CC: dev@dpdk.org, konstantin.ananyev@intel.com, bruce.richardso= n@intel.com > > > > > > Subject: Re: [dpdk-dev] [PATCH] ring: relax alignment constrain= t on ring > > > > > > structure > > > > > > User-Agent: NeoMutt/20170113 (1.7.2) > > > > > > > > > > > > On Tue, Apr 03, 2018 at 08:37:23PM +0530, Jerin Jacob wrote: > > > > > > > -----Original Message----- > > > > > > > > Date: Tue, 3 Apr 2018 15:26:44 +0200 > > > > > > > > From: Olivier Matz > > > > > > > > To: dev@dpdk.org > > > > > > > > Subject: [dpdk-dev] [PATCH] ring: relax alignment constrain= t on ring > > > > > > > > structure > > > > > > > > X-Mailer: git-send-email 2.11.0 > > > > > > > > > > > > > > > > The initial objective of > > > > > > > > commit d9f0d3a1ffd4 ("ring: remove split cacheline build se= tting") > > > > > > > > was to add an empty cache line betwee, the producer and con= sumer > > > > > > > > data (on platform with cache line size =3D 64B), preventing= from > > > > > > > > having them on adjacent cache lines. > > > > > > > > > > > > > > > > Following discussion on the mailing list, it appears that t= his > > > > > > > > also imposes an alignment constraint that is not required. > > > > > > > > > > > > > > > > This patch removes the extra alignment constraint and adds = the > > > > > > > > empty cache lines using padding fields in the structure. Th= e > > > > > > > > size of rte_ring structure and the offset of the fields rem= ain > > > > > > > > the same on platforms with cache line size =3D 64B: > > > > > > > > > > > > > > > > rte_ring =3D 384 > > > > > > > > rte_ring.name =3D 0 > > > > > > > > rte_ring.flags =3D 32 > > > > > > > > rte_ring.memzone =3D 40 > > > > > > > > rte_ring.size =3D 48 > > > > > > > > rte_ring.mask =3D 52 > > > > > > > > rte_ring.prod =3D 128 > > > > > > > > rte_ring.cons =3D 256 > > > > > > > > > > > > > > > > But it has an impact on platform where cache line size is 1= 28B: > > > > > > > > > > > > > > > > rte_ring =3D 384 -> 768 > > > > > > > > rte_ring.name =3D 0 > > > > > > > > rte_ring.flags =3D 32 > > > > > > > > rte_ring.memzone =3D 40 > > > > > > > > rte_ring.size =3D 48 > > > > > > > > rte_ring.mask =3D 52 > > > > > > > > rte_ring.prod =3D 128 -> 256 > > > > > > > > rte_ring.cons =3D 256 -> 512 > > > > > > > > > > > > > > Are we leaving TWO cacheline to make sure, HW prefetch don't = load > > > > > > > the adjust cacheline(consumer)? > > > > > > > > > > > > > > If so, Will it have impact on those machine where it is 128B = Cache line > > > > > > > and the HW prefetcher is not loading the next caching explici= tly. Right? > > > > > > > > > > > > The impact on machines that have a 128B cache line is that an u= nused > > > > > > cache line will be added between the producer and consumer data= . I > > > > > > expect that the impact is positive in case there is a hw prefet= cher, and > > > > > > null in case there is no such prefetcher. > > > > > > > > > > It is not NULL, Right? You are loosing 256B for each ring. > > > > > > > > Is it really that important? > > > > > > Pipeline or eventdev SW cases there could more rings in the system. > > > I don't see any downside of having config option which is enabled > > > default. > > > > > > In my view, such config options are good, as in embedded usecases, cu= stomers > > > can really fine tune the target for the need. In server usecases, let= the default > > > of option be enabled, no harm. > > > > But that would mean we have to maintain two layouts for the rte_ring st= ructure. >=20 > Is there any downside of having two configurable layout? meaning, we are = not > transferring rte_ring structure over network etc(ie no interoperability > issue). Does it really matter? May I am missing something here. My concern about potential compatibility problems we are introducing - library build with 'y', while app wit 'n', or visa-versa. I wonder are there really a lot of users who would be interested in such sa= vings? Could it happen that this new option would sit here unused and untested? Konstantin >=20 > I was thinking like this: >=20 > in config/common_base: > CONFIG_RTE_EAL_HAS_HW_PREFETCH=3Dy >=20 > #ifdef RTE_EAL_HAS_HW_PREFETCH > #define EMPTY_CACHE_LINE char pad0 __rte_cache_aligned > #else > #define EMPTY_CACHE_LINE > #endif >=20 > struct rte_ring_headtail { > .. > .. > char pad0 __rte_cache_aligned; /**< empty cache line */ > struct rte_ring_headtail prod __rte_cache_aligned; > EMPTY_CACHE_LINE > struct rte_ring_headtail cons __rte_cache_aligned; > EMPTY_CACHE_LINE > }