From: "Ananyev, Konstantin" <konstantin.ananyev@intel.com>
To: Neil Horman <nhorman@tuxdriver.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>
Subject: Re: [dpdk-dev] [PATCH 10/17] librte_acl: add AVX2 as new rte_acl_classify() method
Date: Mon, 15 Dec 2014 16:33:47 +0000 [thread overview]
Message-ID: <2601191342CEEE43887BDE71AB977258213C0D9C@IRSMSX105.ger.corp.intel.com> (raw)
In-Reply-To: <20141215160009.GC3803@hmsreliant.think-freely.org>
Hi Neil,
> -----Original Message-----
> From: Neil Horman [mailto:nhorman@tuxdriver.com]
> Sent: Monday, December 15, 2014 4:00 PM
> To: Ananyev, Konstantin
> Cc: dev@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH 10/17] librte_acl: add AVX2 as new rte_acl_classify() method
>
> On Sun, Dec 14, 2014 at 06:10:52PM +0000, Konstantin Ananyev wrote:
> > Introduce new classify() method that uses AVX2 instructions.
> > From my measurements:
> > On HSW boards when processing >= 16 packets per call,
> > AVX2 method outperforms it's SSE counterpart by 10-25%,
> > (depending on the ruleset).
> > At runtime, this method is selected as default one on HW that supports AVX2.
> >
> > Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
> > ---
> > lib/librte_acl/Makefile | 9 +
> > lib/librte_acl/acl.h | 4 +
> > lib/librte_acl/acl_run.h | 2 +-
> > lib/librte_acl/acl_run_avx2.c | 58 +++++
> > lib/librte_acl/acl_run_avx2.h | 305 ++++++++++++++++++++++++
> > lib/librte_acl/acl_run_sse.c | 537 +-----------------------------------------
> > lib/librte_acl/acl_run_sse.h | 533 +++++++++++++++++++++++++++++++++++++++++
> > lib/librte_acl/rte_acl.c | 5 +-
> > lib/librte_acl/rte_acl.h | 2 +
> > 9 files changed, 917 insertions(+), 538 deletions(-)
> > create mode 100644 lib/librte_acl/acl_run_avx2.c
> > create mode 100644 lib/librte_acl/acl_run_avx2.h
> > create mode 100644 lib/librte_acl/acl_run_sse.h
> >
> > diff --git a/lib/librte_acl/Makefile b/lib/librte_acl/Makefile
> > index 65e566d..223ec31 100644
> > --- a/lib/librte_acl/Makefile
> > +++ b/lib/librte_acl/Makefile
> > @@ -45,8 +45,17 @@ SRCS-$(CONFIG_RTE_LIBRTE_ACL) += acl_bld.c
> > SRCS-$(CONFIG_RTE_LIBRTE_ACL) += acl_gen.c
> > SRCS-$(CONFIG_RTE_LIBRTE_ACL) += acl_run_scalar.c
> > SRCS-$(CONFIG_RTE_LIBRTE_ACL) += acl_run_sse.c
> > +SRCS-$(CONFIG_RTE_LIBRTE_ACL) += acl_run_avx2.c
> >
> > CFLAGS_acl_run_sse.o += -msse4.1
> > +ifeq ($(CC), icc)
> > +CFLAGS_acl_run_avx2.o += -march=core-avx2
> > +else ifneq ($(shell \
> > +test $(GCC_MAJOR_VERSION) -le 4 -a $(GCC_MINOR_VERSION) -le 6 && echo 1), 1)
> > +CFLAGS_acl_run_avx2.o += -mavx2
> > +else
> > +CFLAGS_acl_run_avx2.o += -msse4.1
> > +endif
> >
> This seems broken. You've unilaterally included acl_run_avx2.c in the build
> list above, but only enable -mavx2 if the compiler is at least gcc 4.6.
Actually 4.7 (before that version, as I know, gcc doesn't support avx2)
> Unless
> you want to make gcc 4.6 a requirement for building,
I believe DPDK is required to be buildable by gcc 4.6
As I remember, we have to support it all way down to gcc 4.3.
> you need to also exclude
> the file above from the build list.
That means that for gcc 4.6 and below rte_acl_classify_avx2() would not be defined.
And then at runtime, I have to check for that somehow and (re)populate classify_fns[].
Doesn't seems like a good way to me.
Instead, I prefer to always build acl_run_avx2.c,
but for old compilers that don't support AVX2 -
rte_acl_classify_avx2() would simply be identical to rte_acl_classify_sse().
> That in turn I think allows you to remove a
> bunch of the ifdeffing that you've done in some of the avx2 specific files.
Actually there are not many of them.
One in acl_run_avx2.h and another in acl_run_avx2.c.
Konstantin
>
> Neil
next prev parent reply other threads:[~2014-12-15 16:41 UTC|newest]
Thread overview: 31+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-12-14 18:10 [dpdk-dev] [PATCH 00/17] ACL: New AVX2 classify method and several other enhancements Konstantin Ananyev
2014-12-14 18:10 ` [dpdk-dev] [PATCH 01/17] app/test: few small fixes fot test_acl.c Konstantin Ananyev
2014-12-14 18:10 ` [dpdk-dev] [PATCH 02/17] librte_acl: make data_indexes long enough to survive idle transitions Konstantin Ananyev
2014-12-14 18:10 ` [dpdk-dev] [PATCH 03/17] librte_acl: remove build phase heuristsic with negative perfomance effect Konstantin Ananyev
2014-12-14 18:10 ` [dpdk-dev] [PATCH 04/17] librte_acl: fix a bug at build phase that can cause matches beeing overwirtten Konstantin Ananyev
2014-12-14 18:10 ` [dpdk-dev] [PATCH 05/17] librte_acl: introduce DFA nodes compression (group64) for identical entries Konstantin Ananyev
2014-12-14 18:10 ` [dpdk-dev] [PATCH 06/17] librte_acl: build/gen phase - simplify the way match nodes are allocated Konstantin Ananyev
2014-12-14 18:10 ` [dpdk-dev] [PATCH 07/17] librte_acl: make scalar RT code to be more similar to vector one Konstantin Ananyev
2014-12-14 18:10 ` [dpdk-dev] [PATCH 08/17] librte_acl: a bit of RT code deduplication Konstantin Ananyev
2014-12-14 18:10 ` [dpdk-dev] [PATCH 09/17] EAL: introduce rte_ymm and relatives in rte_common_vect.h Konstantin Ananyev
2014-12-15 15:56 ` Neil Horman
2014-12-14 18:10 ` [dpdk-dev] [PATCH 10/17] librte_acl: add AVX2 as new rte_acl_classify() method Konstantin Ananyev
2014-12-15 16:00 ` Neil Horman
2014-12-15 16:33 ` Ananyev, Konstantin [this message]
2014-12-15 20:20 ` Neil Horman
2014-12-16 16:16 ` Ananyev, Konstantin
2014-12-17 15:32 ` Neil Horman
2014-12-17 19:22 ` Ananyev, Konstantin
2014-12-17 20:27 ` Neil Horman
2014-12-18 15:01 ` Ananyev, Konstantin
2015-01-06 9:57 ` Ananyev, Konstantin
2015-01-06 12:40 ` Neil Horman
2014-12-17 0:38 ` Ananyev, Konstantin
2014-12-14 18:10 ` [dpdk-dev] [PATCH 11/17] test-acl: add ability to manually select RT method Konstantin Ananyev
2014-12-14 18:10 ` [dpdk-dev] [PATCH 12/17] librte_acl: Remove search_sse_2 and relatives Konstantin Ananyev
2014-12-14 18:10 ` [dpdk-dev] [PATCH 13/17] libter_acl: move lo/hi dwords shuffle out from calc_addr Konstantin Ananyev
2014-12-14 18:10 ` [dpdk-dev] [PATCH 14/17] libte_acl: make calc_addr a define to deduplicate the code Konstantin Ananyev
2014-12-14 18:10 ` [dpdk-dev] [PATCH 15/17] libte_acl: introduce max_size into rte_acl_config Konstantin Ananyev
2014-12-14 18:10 ` [dpdk-dev] [PATCH 16/17] libte_acl: remove unused macros Konstantin Ananyev
2014-12-14 18:10 ` [dpdk-dev] [PATCH 17/17] libte_acl: fix compilation issues with RTE_LIBRTE_ACL_STANDALONE=y Konstantin Ananyev
2014-12-16 13:51 ` Neil Horman
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=2601191342CEEE43887BDE71AB977258213C0D9C@IRSMSX105.ger.corp.intel.com \
--to=konstantin.ananyev@intel.com \
--cc=dev@dpdk.org \
--cc=nhorman@tuxdriver.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).