From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mails.dpdk.org (mails.dpdk.org [217.70.189.124]) by inbox.dpdk.org (Postfix) with ESMTP id 0854AA0C4B; Tue, 2 Nov 2021 19:38:32 +0100 (CET) Received: from [217.70.189.124] (localhost [127.0.0.1]) by mails.dpdk.org (Postfix) with ESMTP id C20BC40A4B; Tue, 2 Nov 2021 19:38:32 +0100 (CET) Received: from mga12.intel.com (mga12.intel.com [192.55.52.136]) by mails.dpdk.org (Postfix) with ESMTP id CFD4140689 for ; Tue, 2 Nov 2021 19:38:30 +0100 (CET) X-IronPort-AV: E=McAfee;i="6200,9189,10156"; a="211397018" X-IronPort-AV: E=Sophos;i="5.87,203,1631602800"; d="scan'208";a="211397018" Received: from orsmga007.jf.intel.com ([10.7.209.58]) by fmsmga106.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 02 Nov 2021 11:38:29 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.87,203,1631602800"; d="scan'208";a="489227673" Received: from silpixa00400072.ir.intel.com ([10.237.222.213]) by orsmga007.jf.intel.com with ESMTP; 02 Nov 2021 11:38:27 -0700 From: Vladimir Medvedkin To: dev@dpdk.org Cc: yipeng1.wang@intel.com, sameh.gobriel@intel.com, bruce.richardson@intel.com, konstantin.ananyev@intel.com, stephen@networkplumber.org, thomas@monjalon.net Date: Tue, 2 Nov 2021 18:38:21 +0000 Message-Id: <1635878305-102888-1-git-send-email-vladimir.medvedkin@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1630944239-363648-1-git-send-email-vladimir.medvedkin@intel.com> References: <1630944239-363648-1-git-send-email-vladimir.medvedkin@intel.com> Subject: [dpdk-dev] [PATCH v8 0/4] optimized Toeplitz hash implementation X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.29 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" This patch series adds a new optimized implementation for the Toeplitz hash function using Galois Fields New instruction (GFNI). The main use case of this function is to calculate the hash value for a single data, so there is no bulk implementation. For performance reasons, the implementation was placed in a public header. It is the responsibility of the user to ensure the platform supports GFNI (by doing runtime checks of rte_thash_gfni_supported variable) before calling these functions. v8: - rebased on the latest main - fixed buffer overflow reported by ASAN v7: - reworked performance tests code v6: - addressed Thomas and Stephen's comments - squash the doc and test commith with the relevant code addition - split for 2 patches, each per implementation - single and bulk v5: - rebase on the latest main - fix spelling v4: - included rte_log.h inside the rte_thash_gfni.h v3: - implementation moved to x86 specific header - added rte_thash_gfni_supported() instead of the variable - removed RTE_INIT section, due to adding rte_thash_gfni_supported() - reworked rte_thash_complete_matrix() to make it easier tor read v2: - fixed typos - made big_rss_key static const and indented - addressed Konstantin's comments Vladimir Medvedkin (4): hash: add new toeplitz hash implementation hash: add bulk toeplitz hash implementation hash: enable gfni thash implementation test/thash: add performance tests for the Toeplitz hash app/test/meson.build | 2 + app/test/test_thash.c | 237 ++++++++++++++++++++++++++++ app/test/test_thash_perf.c | 135 ++++++++++++++++ doc/api/doxy-api-index.md | 1 + doc/guides/prog_guide/toeplitz_hash_lib.rst | 38 ++++- doc/guides/rel_notes/release_21_11.rst | 4 + lib/hash/meson.build | 6 +- lib/hash/rte_thash.c | 72 ++++++++- lib/hash/rte_thash.h | 54 +++++++ lib/hash/rte_thash_gfni.h | 87 ++++++++++ lib/hash/rte_thash_x86_gfni.h | 223 ++++++++++++++++++++++++++ lib/hash/version.map | 6 + 12 files changed, 856 insertions(+), 9 deletions(-) create mode 100644 app/test/test_thash_perf.c create mode 100644 lib/hash/rte_thash_gfni.h create mode 100644 lib/hash/rte_thash_x86_gfni.h -- 2.7.4