From patchwork Sun Jul 16 15:13:52 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: =?utf-8?q?R=C3=A9mi_Denis-Courmont?= X-Patchwork-Id: 42770 Delivered-To: ffmpegpatchwork2@gmail.com Received: by 2002:a05:6a20:6da1:b0:131:a7d0:bc6d with SMTP id gl33csp4887110pzb; Sun, 16 Jul 2023 08:14:10 -0700 (PDT) X-Google-Smtp-Source: APBJJlH0a4EAfGelFrl2LZXx7r0IZCngVp7SfLcQQ3Az6fxyHVWGm3T3svATQpHgdgUhKjZOUD5j X-Received: by 2002:a05:6402:b2e:b0:51b:dcb4:a9b3 with SMTP id bo14-20020a0564020b2e00b0051bdcb4a9b3mr10773542edb.24.1689520450468; Sun, 16 Jul 2023 08:14:10 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1689520450; cv=none; d=google.com; s=arc-20160816; b=mUzfT8Yct/njOzA5+ATRRVXmOTHS1Xb3MP7S8VG+x9QWrRqxOrpfcQHWTfVi1IuFKx Xlm83qY2BBsL3dL+IpPwfKfzVWthgydqkyupC2Q2A0nC4dL3bcjiQRA5TT5cRa5/A/2g n2Lg3JlZcqVfApil31EeFHO5Tytac07t3/i98snw+hisRcwxtoixNyNFNbgmhW9hbTA/ NtC9/Dwnb5ginbN5ZOebB4OvIupCqUHn4RPV5vo45OAhl7RhIbvnWPJeJtP0m38PcKJi Gt0b+bYBtKEsMDYhpP8ivK05yfkDzSoS9XRjYlYrTcxY2pIwp7pvzz1im4aog7EPXPOX QZQQ== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=sender:errors-to:content-transfer-encoding:reply-to:list-subscribe :list-help:list-post:list-archive:list-unsubscribe:list-id :precedence:subject:mime-version:message-id:date:to:from :delivered-to; bh=6i6XKuYozSF0Ny9kI93Fa6lfTa9KRWOvSHvMus0elXQ=; fh=hQcp50obTJ8bXC1it5NuEN23RGKfx0/zZ3s2gmreL+A=; b=SgtmP6+BjYQDxxgd8u9GklfqZHooo4jkjLe+F9CEauwZtoz+gJzT4HA0HQnLCjbD6C MFhxEUfuMSw0JNoAuLEoABnHxzjLH/cpqAY5oUO8Nt3ppoRvjeVdOFG+YWG6ASerM9QD Od2DFSqknX/62315ZW2ObA94gxXzQEYIWwiJ/4PENiH6lB8EBKwH3gWTZ76OCq33u9Y6 gyCsfYKBeDlsvLBoBJRUMchXt6r2AKq2zXq3jq24enfobWyskwt9ReSHcdEyauKKtlu/ d4luvrDRsW8XZQ21n41nyisOp18SCNbURg6td8DjhClYL8BSb6YyGx2yM6fRLUVACGvL po3g== ARC-Authentication-Results: i=1; mx.google.com; spf=pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) smtp.mailfrom=ffmpeg-devel-bounces@ffmpeg.org Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org. [79.124.17.100]) by mx.google.com with ESMTP id g13-20020a50ee0d000000b0051e0252e3e9si2504555eds.362.2023.07.16.08.14.10; Sun, 16 Jul 2023 08:14:10 -0700 (PDT) Received-SPF: pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) client-ip=79.124.17.100; Authentication-Results: mx.google.com; spf=pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) smtp.mailfrom=ffmpeg-devel-bounces@ffmpeg.org Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id BD6F568C69A; Sun, 16 Jul 2023 18:14:00 +0300 (EEST) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from ursule.remlab.net (vps-a2bccee9.vps.ovh.net [51.75.19.47]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 4CDBE68C4C1 for ; Sun, 16 Jul 2023 18:13:53 +0300 (EEST) Received: from basile.remlab.net (localhost [IPv6:::1]) by ursule.remlab.net (Postfix) with ESMTP id D734CC00C0 for ; Sun, 16 Jul 2023 18:13:52 +0300 (EEST) From: =?utf-8?q?R=C3=A9mi_Denis-Courmont?= To: ffmpeg-devel@ffmpeg.org Date: Sun, 16 Jul 2023 18:13:52 +0300 Message-Id: <20230716151352.121105-2-remi@remlab.net> X-Mailer: git-send-email 2.40.1 MIME-Version: 1.0 Subject: [FFmpeg-devel] [PATCH 2/2] lavc/bswapdsp: purge RISC-V V bswap32 X-BeenThere: ffmpeg-devel@ffmpeg.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: FFmpeg development discussions and patches List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Reply-To: FFmpeg development discussions and patches Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" X-TUID: 3FzwtMnfuI4u This cannot beat the Zbb implementation, and it is unlikely that a real meaningful CPU design would support V and not Zbb. The best loop rewrite that I could come up with (4 shifts, 2 ands, 3 ors) is still ~40% slower than Zbb. A proper faster vector implementation should be feasible with the cryptographic vector extensions, but that is a story for another time. --- libavcodec/riscv/bswapdsp_init.c | 5 +---- libavcodec/riscv/bswapdsp_rvv.S | 23 ----------------------- 2 files changed, 1 insertion(+), 27 deletions(-) diff --git a/libavcodec/riscv/bswapdsp_init.c b/libavcodec/riscv/bswapdsp_init.c index 6ad63e3805..ed666c9b3a 100644 --- a/libavcodec/riscv/bswapdsp_init.c +++ b/libavcodec/riscv/bswapdsp_init.c @@ -26,7 +26,6 @@ #include "libavcodec/bswapdsp.h" void ff_bswap32_buf_rvb(uint32_t *dst, const uint32_t *src, int len); -void ff_bswap32_buf_rvv(uint32_t *dst, const uint32_t *src, int len); void ff_bswap16_buf_rvv(uint16_t *dst, const uint16_t *src, int len); av_cold void ff_bswapdsp_init_riscv(BswapDSPContext *c) @@ -39,10 +38,8 @@ av_cold void ff_bswapdsp_init_riscv(BswapDSPContext *c) c->bswap_buf = ff_bswap32_buf_rvb; #endif #if HAVE_RVV - if (flags & AV_CPU_FLAG_RVV_I32) { - c->bswap_buf = ff_bswap32_buf_rvv; + if (flags & AV_CPU_FLAG_RVV_I32) c->bswap16_buf = ff_bswap16_buf_rvv; - } #endif } } diff --git a/libavcodec/riscv/bswapdsp_rvv.S b/libavcodec/riscv/bswapdsp_rvv.S index 8b585ec5c9..b37fe26255 100644 --- a/libavcodec/riscv/bswapdsp_rvv.S +++ b/libavcodec/riscv/bswapdsp_rvv.S @@ -21,29 +21,6 @@ #include "config.h" #include "libavutil/riscv/asm.S" -func ff_bswap32_buf_rvv, zve32x - li t4, 4 - addi t1, a0, 1 - addi t2, a0, 2 - addi t3, a0, 3 -1: - vsetvli t0, a2, e8, m1, ta, ma - vlseg4e8.v v8, (a1) - sub a2, a2, t0 - sh2add a1, t0, a1 - vsse8.v v8, (t3), t4 - sh2add t3, t0, t3 - vsse8.v v9, (t2), t4 - sh2add t2, t0, t2 - vsse8.v v10, (t1), t4 - sh2add t1, t0, t1 - vsse8.v v11, (a0), t4 - sh2add a0, t0, a0 - bnez a2, 1b - - ret -endfunc - func ff_bswap16_buf_rvv, zve32x 1: vsetvli t0, a2, e16, m8, ta, ma