From patchwork Mon Mar 20 10:08:28 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Hunt, David" X-Patchwork-Id: 22009 Return-Path: X-Original-To: patchwork@dpdk.org Delivered-To: patchwork@dpdk.org Received: from [92.243.14.124] (localhost [IPv6:::1]) by dpdk.org (Postfix) with ESMTP id 0D1CFCF8A; Mon, 20 Mar 2017 18:08:26 +0100 (CET) Received: from mga03.intel.com (mga03.intel.com [134.134.136.65]) by dpdk.org (Postfix) with ESMTP id 2938B37B1 for ; Mon, 20 Mar 2017 18:08:03 +0100 (CET) DKIM-Signature: v=1; a=rsa-sha256; c=simple/simple; d=intel.com; i=@intel.com; q=dns/txt; s=intel; t=1490029683; x=1521565683; h=from:to:cc:subject:date:message-id:in-reply-to: references; bh=TXU8z+Af5WZZLGet2pGFk1ZA8qWdqBesr8K9w4zULGs=; b=V3V3/onXqDmwdgpp5zNeym4gV46D4cocQa5VFkgkX1fiFr55CaCDuJEO UHV7foiQ7FNtnhjZzwImv+lnboaz9Q==; Received: from orsmga004.jf.intel.com ([10.7.209.38]) by orsmga103.jf.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 20 Mar 2017 10:08:02 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.36,195,1486454400"; d="scan'208";a="69068932" Received: from silpixa00397515.ir.intel.com (HELO silpixa00397515.ger.corp.intel.com) ([10.237.223.14]) by orsmga004.jf.intel.com with ESMTP; 20 Mar 2017 10:08:01 -0700 From: David Hunt To: dev@dpdk.org Cc: bruce.richardson@intel.com, David Hunt Date: Mon, 20 Mar 2017 10:08:28 +0000 Message-Id: <1490004522-183515-5-git-send-email-david.hunt@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1490004522-183515-1-git-send-email-david.hunt@intel.com> References: <1489558767-56329-2-git-send-email-david.hunt@intel.com> <1490004522-183515-1-git-send-email-david.hunt@intel.com> Subject: [dpdk-dev] [PATCH v11 04/18] lib: add SIMD flow matching to distributor X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" Add an optimised version of the in-flight flow matching algorithm using SIMD instructions. This should give up to 1.5x over the scalar versions performance. Falls back to scalar version if SSE4.2 not available Signed-off-by: David Hunt Acked-by: Bruce Richardson --- lib/librte_distributor/Makefile | 10 ++ lib/librte_distributor/rte_distributor.c | 16 ++- .../rte_distributor_match_generic.c | 43 ++++++++ lib/librte_distributor/rte_distributor_match_sse.c | 114 +++++++++++++++++++++ lib/librte_distributor/rte_distributor_private.h | 5 + 5 files changed, 186 insertions(+), 2 deletions(-) create mode 100644 lib/librte_distributor/rte_distributor_match_generic.c create mode 100644 lib/librte_distributor/rte_distributor_match_sse.c diff --git a/lib/librte_distributor/Makefile b/lib/librte_distributor/Makefile index 74256ff..a812fe4 100644 --- a/lib/librte_distributor/Makefile +++ b/lib/librte_distributor/Makefile @@ -44,6 +44,16 @@ LIBABIVER := 1 # all source are stored in SRCS-y SRCS-$(CONFIG_RTE_LIBRTE_DISTRIBUTOR) := rte_distributor_v20.c SRCS-$(CONFIG_RTE_LIBRTE_DISTRIBUTOR) += rte_distributor.c +ifeq ($(CONFIG_RTE_ARCH_X86),y) +SRCS-$(CONFIG_RTE_LIBRTE_DISTRIBUTOR) += rte_distributor_match_sse.c +# distributor SIMD algo needs SSE4.2 support +ifeq ($(findstring RTE_MACHINE_CPUFLAG_SSE4_2,$(CFLAGS)),) +CFLAGS_rte_distributor_match_sse.o += -msse4.2 +endif +else +SRCS-$(CONFIG_RTE_LIBRTE_DISTRIBUTOR) += rte_distributor_match_generic.c +endif + # install this header file SYMLINK-$(CONFIG_RTE_LIBRTE_DISTRIBUTOR)-include := rte_distributor.h diff --git a/lib/librte_distributor/rte_distributor.c b/lib/librte_distributor/rte_distributor.c index 75b0d47..6158fa6 100644 --- a/lib/librte_distributor/rte_distributor.c +++ b/lib/librte_distributor/rte_distributor.c @@ -391,7 +391,13 @@ rte_distributor_process_v1705(struct rte_distributor_v1705 *d, for (; i < RTE_DIST_BURST_SIZE; i++) flows[i] = 0; - find_match_scalar(d, &flows[0], &matches[0]); + switch (d->dist_match_fn) { + case RTE_DIST_MATCH_VECTOR: + find_match_vec(d, &flows[0], &matches[0]); + break; + default: + find_match_scalar(d, &flows[0], &matches[0]); + } /* * Matches array now contain the intended worker ID (+1) of @@ -607,7 +613,13 @@ rte_distributor_create_v1705(const char *name, snprintf(d->name, sizeof(d->name), "%s", name); d->num_workers = num_workers; d->alg_type = alg_type; - d->dist_match_fn = RTE_DIST_MATCH_SCALAR; + +#if defined(RTE_ARCH_X86) + if (rte_cpu_get_flag_enabled(RTE_CPUFLAG_SSE4_2)) + d->dist_match_fn = RTE_DIST_MATCH_VECTOR; + else +#endif + d->dist_match_fn = RTE_DIST_MATCH_SCALAR; /* * Set up the backog tags so they're pointing at the second cache diff --git a/lib/librte_distributor/rte_distributor_match_generic.c b/lib/librte_distributor/rte_distributor_match_generic.c new file mode 100644 index 0000000..7c2f9f5 --- /dev/null +++ b/lib/librte_distributor/rte_distributor_match_generic.c @@ -0,0 +1,43 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2017 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include +#include "rte_distributor_private.h" +#include "rte_distributor.h" + +void +find_match_vec(struct rte_distributor_v1705 *d, + uint16_t *data_ptr, + uint16_t *output_ptr) +{ + find_match_scalar(d, data_ptr, output_ptr); +} diff --git a/lib/librte_distributor/rte_distributor_match_sse.c b/lib/librte_distributor/rte_distributor_match_sse.c new file mode 100644 index 0000000..b9f9bb0 --- /dev/null +++ b/lib/librte_distributor/rte_distributor_match_sse.c @@ -0,0 +1,114 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2017 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include +#include "rte_distributor_private.h" +#include "rte_distributor.h" +#include "smmintrin.h" +#include "nmmintrin.h" + + +void +find_match_vec(struct rte_distributor_v1705 *d, + uint16_t *data_ptr, + uint16_t *output_ptr) +{ + /* Setup */ + __m128i incoming_fids; + __m128i inflight_fids; + __m128i preflight_fids; + __m128i wkr; + __m128i mask1; + __m128i mask2; + __m128i output; + struct rte_distributor_backlog *bl; + uint16_t i; + + /* + * Function overview: + * 2. Loop through all worker ID's + * 2a. Load the current inflights for that worker into an xmm reg + * 2b. Load the current backlog for that worker into an xmm reg + * 2c. use cmpestrm to intersect flow_ids with backlog and inflights + * 2d. Add any matches to the output + * 3. Write the output xmm (matching worker ids). + */ + + + output = _mm_set1_epi16(0); + incoming_fids = _mm_load_si128((__m128i *)data_ptr); + + for (i = 0; i < d->num_workers; i++) { + bl = &d->backlog[i]; + + inflight_fids = + _mm_load_si128((__m128i *)&(d->in_flight_tags[i])); + preflight_fids = + _mm_load_si128((__m128i *)(bl->tags)); + + /* + * Any incoming_fid that exists anywhere in inflight_fids will + * have 0xffff in same position of the mask as the incoming fid + * Example (shortened to bytes for brevity): + * incoming_fids 0x01 0x02 0x03 0x04 0x05 0x06 0x07 0x08 + * inflight_fids 0x03 0x05 0x07 0x00 0x00 0x00 0x00 0x00 + * mask 0x00 0x00 0xff 0x00 0xff 0x00 0xff 0x00 + */ + + mask1 = _mm_cmpestrm(inflight_fids, 8, incoming_fids, 8, + _SIDD_UWORD_OPS | + _SIDD_CMP_EQUAL_ANY | + _SIDD_UNIT_MASK); + mask2 = _mm_cmpestrm(preflight_fids, 8, incoming_fids, 8, + _SIDD_UWORD_OPS | + _SIDD_CMP_EQUAL_ANY | + _SIDD_UNIT_MASK); + + mask1 = _mm_or_si128(mask1, mask2); + /* + * Now mask contains 0xffff where there's a match. + * Next we need to store the worker_id in the relevant position + * in the output. + */ + + wkr = _mm_set1_epi16(i+1); + mask1 = _mm_and_si128(mask1, wkr); + output = _mm_or_si128(mask1, output); + } + + /* + * At this stage, the output 128-bit contains 8 16-bit values, with + * each non-zero value containing the worker ID on which the + * corresponding flow is pinned to. + */ + _mm_store_si128((__m128i *)output_ptr, output); +} diff --git a/lib/librte_distributor/rte_distributor_private.h b/lib/librte_distributor/rte_distributor_private.h index 82b0daf..04c9cac 100644 --- a/lib/librte_distributor/rte_distributor_private.h +++ b/lib/librte_distributor/rte_distributor_private.h @@ -190,6 +190,11 @@ find_match_scalar(struct rte_distributor_v1705 *d, uint16_t *data_ptr, uint16_t *output_ptr); +void +find_match_vec(struct rte_distributor_v1705 *d, + uint16_t *data_ptr, + uint16_t *output_ptr); + #ifdef __cplusplus } #endif