From patchwork Mon Sep 25 12:38:35 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: kaustubh.raste@imgtec.com X-Patchwork-Id: 5261 Delivered-To: ffmpegpatchwork@gmail.com Received: by 10.2.36.26 with SMTP id f26csp2578438jaa; Mon, 25 Sep 2017 05:37:50 -0700 (PDT) X-Google-Smtp-Source: AOwi7QBLx5X2ubO8dDo2WMbiQMR9StW5m7uYzaGcmaXkZ8gb11ZzPL4T93L0v3xsJ7BTxBDu0klN X-Received: by 10.223.171.21 with SMTP id q21mr5971796wrc.125.1506343070440; Mon, 25 Sep 2017 05:37:50 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1506343070; cv=none; d=google.com; s=arc-20160816; b=Dvc21q7G6jxZm31+Ev65LagGE7vfIcDTKrXca22qTpfxs/0cr1ajJUmfw6q/1dosLV DytH8VsoHqp3dbILXTYV9NZmfpFvZC2vcge8lVjbd9qe1IGWYVPyZgs+JhOAkC1l2MqP 9y6WYEc+/3l1FJMpi5V0K7SArYU1FF1jyy98mlsfdHUnVgy/SqK6HpuG7YeTRJOhI1Yq 6eP4FPfjaqpaMyHZ+X7FyR/0QJO9RjQB9AgR/y3KJ8Owa8Zz3dEPriDi60qpsPytnqfW x9vhFIIXmDFxevZyhyF+g26ro4DqowP0JUttucbeHT852iSssdhW/0wmKMQjwoFeIcRY k9Lg== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=sender:errors-to:content-transfer-encoding:cc: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:arc-authentication-results; bh=IjKAsFjiGEoErjj2EmIhJ45j2P0L1CmaPjxQFre44sM=; b=krkhgQDD7FiMeW+iUHbnzc9Lj/me25NuQmHQ+YQPwbSKaCmPsKHLdAB0ftMKzKRhmT IDt+NOXwrCj1KKIz9oIyjaKxMRBnCdTXxfveAGP18MWJ1PurUx1LHNFKNUBkMgrpFHp/ FULBiPDfj5xiM+mx3pJ6zEYm/teyDWQa00TRyp6kF9I0RC8gM5s//MBQ0eeqB8MLq1jR ntRU1RFWCHrJgnw9TRc+NRimIZ4o3mnb3dIVg+mFJIdXtCidRyvcG1yRJIN2o8kWoZZs wP13d8SWV0n95N+srT4dq+6PRAEaSIX5xwt3VlGckF3FIN6W8UZZVzqGmQHPLhDCfpN2 VqRw== 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 b34si5202921wra.5.2017.09.25.05.37.49; Mon, 25 Sep 2017 05:37:50 -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 6B24E688325; Mon, 25 Sep 2017 15:37:37 +0300 (EEST) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from mailapp01.imgtec.com (mailapp01.imgtec.com [195.59.15.196]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 264EB680353 for ; Mon, 25 Sep 2017 15:37:31 +0300 (EEST) Received: from hhmail02.hh.imgtec.org (unknown [10.100.10.20]) by Forcepoint Email with ESMTPS id 6A621D7F9CE58 for ; Mon, 25 Sep 2017 13:37:37 +0100 (IST) Received: from pudesk204.pu.imgtec.org (192.168.91.13) by hhmail02.hh.imgtec.org (10.100.10.20) with Microsoft SMTP Server (TLS) id 14.3.361.1; Mon, 25 Sep 2017 13:37:39 +0100 From: To: Date: Mon, 25 Sep 2017 18:08:35 +0530 Message-ID: <1506343115-32667-1-git-send-email-kaustubh.raste@imgtec.com> X-Mailer: git-send-email 1.7.9.5 MIME-Version: 1.0 X-Originating-IP: [192.168.91.13] Subject: [FFmpeg-devel] [PATCH] avcodec/mips: preload data in hevc sao edge 45 degree filter msa functions X-BeenThere: ffmpeg-devel@ffmpeg.org X-Mailman-Version: 2.1.20 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 Cc: Kaustubh Raste Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" From: Kaustubh Raste Signed-off-by: Kaustubh Raste --- libavcodec/mips/hevc_lpf_sao_msa.c | 197 ++++++++++++++++++++++++------------ 1 file changed, 135 insertions(+), 62 deletions(-) diff --git a/libavcodec/mips/hevc_lpf_sao_msa.c b/libavcodec/mips/hevc_lpf_sao_msa.c index 39c647e..c192265 100644 --- a/libavcodec/mips/hevc_lpf_sao_msa.c +++ b/libavcodec/mips/hevc_lpf_sao_msa.c @@ -1878,23 +1878,25 @@ static void hevc_sao_edge_filter_45degree_4width_msa(uint8_t *dst, int32_t height) { uint8_t *src_orig; - int32_t h_cnt; uint32_t dst_val0, dst_val1; - v8i16 edge_idx = { 1, 2, 0, 3, 4, 0, 0, 0 }; + v16i8 edge_idx = { 1, 2, 0, 3, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; v16u8 const1 = (v16u8) __msa_ldi_b(1); - v16i8 zero = { 0 }; + v16i8 offset, sao_offset = LD_SB(sao_offset_val); v16u8 cmp_minus10, diff_minus10, src_minus10, cmp_minus11, diff_minus11; v16u8 src_minus11, src10, src11; v16i8 src_plus0, src_zero0, src_plus1, src_zero1, dst0; - v8i16 sao_offset, src00, src01, offset_mask0, offset_mask1; + v8i16 offset_mask0, offset_mask1; - sao_offset = LD_SH(sao_offset_val); + sao_offset = __msa_pckev_b(sao_offset, sao_offset); src_orig = src - 1; + + /* load in advance */ LD_UB2(src_orig - src_stride, src_stride, src_minus10, src_minus11); + LD_UB2(src_orig + src_stride, src_stride, src10, src11); - for (h_cnt = (height >> 1); h_cnt--;) { - LD_UB2(src_orig + src_stride, src_stride, src10, src11); + for (height -= 2; height; height -= 2) { + src_orig += (src_stride << 1); SLDI_B2_0_SB(src_minus11, src10, src_zero0, src_zero1, 1); SLDI_B2_0_SB(src10, src11, src_plus0, src_plus1, 2); @@ -1917,20 +1919,22 @@ static void hevc_sao_edge_filter_45degree_4width_msa(uint8_t *dst, offset_mask0 = (v8i16) (__msa_hadd_u_h(diff_minus10, diff_minus10) + 2); offset_mask1 = (v8i16) (__msa_hadd_u_h(diff_minus11, diff_minus11) + 2); - VSHF_H2_SH(edge_idx, edge_idx, sao_offset, sao_offset, offset_mask0, - offset_mask0, offset_mask0, offset_mask0); - VSHF_H2_SH(edge_idx, edge_idx, sao_offset, sao_offset, offset_mask1, - offset_mask1, offset_mask1, offset_mask1); - ILVEV_B2_SH(src_zero0, zero, src_zero1, zero, src00, src01); - ADD2(offset_mask0, src00, offset_mask1, src01, offset_mask0, - offset_mask1); - CLIP_SH2_0_255(offset_mask0, offset_mask1); + offset = __msa_pckev_b((v16i8) offset_mask1, (v16i8) offset_mask0); + dst0 = __msa_pckev_b((v16i8) src_zero1, (v16i8) src_zero0); - dst0 = __msa_pckev_b((v16i8) offset_mask1, (v16i8) offset_mask0); + VSHF_B2_SB(edge_idx, edge_idx, sao_offset, sao_offset, offset, + offset, offset, offset); + + dst0 = (v16i8) __msa_xori_b((v16u8) dst0, 128); + dst0 = __msa_adds_s_b(dst0, offset); + dst0 = (v16i8) __msa_xori_b((v16u8) dst0, 128); src_minus10 = src10; src_minus11 = src11; + /* load in advance */ + LD_UB2(src_orig + src_stride, src_stride, src10, src11); + dst_val0 = __msa_copy_u_w((v4i32) dst0, 0); dst_val1 = __msa_copy_u_w((v4i32) dst0, 2); SW(dst_val0, dst); @@ -1938,8 +1942,44 @@ static void hevc_sao_edge_filter_45degree_4width_msa(uint8_t *dst, SW(dst_val1, dst); dst += dst_stride; - src_orig += (src_stride << 1); } + + SLDI_B2_0_SB(src_minus11, src10, src_zero0, src_zero1, 1); + SLDI_B2_0_SB(src10, src11, src_plus0, src_plus1, 2); + + ILVR_B2_UB(src_plus0, src_minus10, src_plus1, src_minus11, src_minus10, + src_minus11); + ILVR_B2_SB(src_zero0, src_zero0, src_zero1, src_zero1, src_zero0, + src_zero1); + + cmp_minus10 = ((v16u8) src_zero0 == src_minus10); + diff_minus10 = __msa_nor_v(cmp_minus10, cmp_minus10); + cmp_minus10 = (src_minus10 < (v16u8) src_zero0); + diff_minus10 = __msa_bmnz_v(diff_minus10, const1, cmp_minus10); + + cmp_minus11 = ((v16u8) src_zero1 == src_minus11); + diff_minus11 = __msa_nor_v(cmp_minus11, cmp_minus11); + cmp_minus11 = (src_minus11 < (v16u8) src_zero1); + diff_minus11 = __msa_bmnz_v(diff_minus11, const1, cmp_minus11); + + offset_mask0 = (v8i16) (__msa_hadd_u_h(diff_minus10, diff_minus10) + 2); + offset_mask1 = (v8i16) (__msa_hadd_u_h(diff_minus11, diff_minus11) + 2); + + offset = __msa_pckev_b((v16i8) offset_mask1, (v16i8) offset_mask0); + dst0 = __msa_pckev_b((v16i8) src_zero1, (v16i8) src_zero0); + + VSHF_B2_SB(edge_idx, edge_idx, sao_offset, sao_offset, offset, offset, + offset, offset); + + dst0 = (v16i8) __msa_xori_b((v16u8) dst0, 128); + dst0 = __msa_adds_s_b(dst0, offset); + dst0 = (v16i8) __msa_xori_b((v16u8) dst0, 128); + + dst_val0 = __msa_copy_u_w((v4i32) dst0, 0); + dst_val1 = __msa_copy_u_w((v4i32) dst0, 2); + SW(dst_val0, dst); + dst += dst_stride; + SW(dst_val1, dst); } static void hevc_sao_edge_filter_45degree_8width_msa(uint8_t *dst, @@ -1950,23 +1990,24 @@ static void hevc_sao_edge_filter_45degree_8width_msa(uint8_t *dst, int32_t height) { uint8_t *src_orig; - int32_t h_cnt; uint64_t dst_val0, dst_val1; - v8i16 edge_idx = { 1, 2, 0, 3, 4, 0, 0, 0 }; + v16i8 edge_idx = { 1, 2, 0, 3, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; v16u8 const1 = (v16u8) __msa_ldi_b(1); - v16i8 zero = { 0 }; + v16i8 offset, sao_offset = LD_SB(sao_offset_val); v16u8 cmp_minus10, diff_minus10, cmp_minus11, diff_minus11; v16u8 src_minus10, src10, src_minus11, src11; - v16i8 src_zero0, src_plus10, src_zero1, src_plus11, dst0, dst1; - v8i16 sao_offset, src00, offset_mask0, src01, offset_mask1; + v16i8 src_zero0, src_plus10, src_zero1, src_plus11, dst0; + v8i16 offset_mask0, offset_mask1; - sao_offset = LD_SH(sao_offset_val); + sao_offset = __msa_pckev_b(sao_offset, sao_offset); src_orig = src - 1; + /* load in advance */ LD_UB2(src_orig - src_stride, src_stride, src_minus10, src_minus11); + LD_UB2(src_orig + src_stride, src_stride, src10, src11); - for (h_cnt = (height >> 1); h_cnt--;) { - LD_UB2(src_orig + src_stride, src_stride, src10, src11); + for (height -= 2; height; height -= 2) { + src_orig += (src_stride << 1); SLDI_B2_0_SB(src_minus11, src10, src_zero0, src_zero1, 1); SLDI_B2_0_SB(src10, src11, src_plus10, src_plus11, 2); @@ -1989,29 +2030,71 @@ static void hevc_sao_edge_filter_45degree_8width_msa(uint8_t *dst, offset_mask0 = (v8i16) (__msa_hadd_u_h(diff_minus10, diff_minus10) + 2); offset_mask1 = (v8i16) (__msa_hadd_u_h(diff_minus11, diff_minus11) + 2); - VSHF_H2_SH(edge_idx, edge_idx, sao_offset, sao_offset, offset_mask0, - offset_mask0, offset_mask0, offset_mask0); - VSHF_H2_SH(edge_idx, edge_idx, sao_offset, sao_offset, offset_mask1, - offset_mask1, offset_mask1, offset_mask1); - ILVEV_B2_SH(src_zero0, zero, src_zero1, zero, src00, src01); - ADD2(offset_mask0, src00, offset_mask1, src01, offset_mask0, - offset_mask1); - CLIP_SH2_0_255(offset_mask0, offset_mask1); - PCKEV_B2_SB(offset_mask0, offset_mask0, offset_mask1, offset_mask1, - dst0, dst1); + offset = __msa_pckev_b((v16i8) offset_mask1, (v16i8) offset_mask0); + dst0 = __msa_pckev_b((v16i8) src_zero1, (v16i8) src_zero0); + + VSHF_B2_SB(edge_idx, edge_idx, sao_offset, sao_offset, offset, offset, + offset, offset); + + dst0 = (v16i8) __msa_xori_b((v16u8) dst0, 128); + dst0 = __msa_adds_s_b(dst0, offset); + dst0 = (v16i8) __msa_xori_b((v16u8) dst0, 128); src_minus10 = src10; src_minus11 = src11; + /* load in advance */ + LD_UB2(src_orig + src_stride, src_stride, src10, src11); + dst_val0 = __msa_copy_u_d((v2i64) dst0, 0); - dst_val1 = __msa_copy_u_d((v2i64) dst1, 0); + dst_val1 = __msa_copy_u_d((v2i64) dst0, 1); SD(dst_val0, dst); dst += dst_stride; SD(dst_val1, dst); - dst += dst_stride; - src_orig += (src_stride << 1); } + + SLDI_B2_0_SB(src_minus11, src10, src_zero0, src_zero1, 1); + SLDI_B2_0_SB(src10, src11, src_plus10, src_plus11, 2); + ILVR_B2_UB(src_plus10, src_minus10, src_plus11, src_minus11, src_minus10, + src_minus11); + ILVR_B2_SB(src_zero0, src_zero0, src_zero1, src_zero1, src_zero0, + src_zero1); + + cmp_minus10 = ((v16u8) src_zero0 == src_minus10); + diff_minus10 = __msa_nor_v(cmp_minus10, cmp_minus10); + cmp_minus10 = (src_minus10 < (v16u8) src_zero0); + diff_minus10 = __msa_bmnz_v(diff_minus10, const1, cmp_minus10); + + cmp_minus11 = ((v16u8) src_zero1 == src_minus11); + diff_minus11 = __msa_nor_v(cmp_minus11, cmp_minus11); + cmp_minus11 = (src_minus11 < (v16u8) src_zero1); + diff_minus11 = __msa_bmnz_v(diff_minus11, const1, cmp_minus11); + + offset_mask0 = (v8i16) (__msa_hadd_u_h(diff_minus10, diff_minus10) + 2); + offset_mask1 = (v8i16) (__msa_hadd_u_h(diff_minus11, diff_minus11) + 2); + + offset = __msa_pckev_b((v16i8) offset_mask1, (v16i8) offset_mask0); + dst0 = __msa_pckev_b((v16i8) src_zero1, (v16i8) src_zero0); + + VSHF_B2_SB(edge_idx, edge_idx, sao_offset, sao_offset, offset, offset, + offset, offset); + + dst0 = (v16i8) __msa_xori_b((v16u8) dst0, 128); + dst0 = __msa_adds_s_b(dst0, offset); + dst0 = (v16i8) __msa_xori_b((v16u8) dst0, 128); + + src_minus10 = src10; + src_minus11 = src11; + + /* load in advance */ + LD_UB2(src_orig + src_stride, src_stride, src10, src11); + + dst_val0 = __msa_copy_u_d((v2i64) dst0, 0); + dst_val1 = __msa_copy_u_d((v2i64) dst0, 1); + SD(dst_val0, dst); + dst += dst_stride; + SD(dst_val1, dst); } static void hevc_sao_edge_filter_45degree_16multiple_msa(uint8_t *dst, @@ -2025,7 +2108,7 @@ static void hevc_sao_edge_filter_45degree_16multiple_msa(uint8_t *dst, { uint8_t *src_orig = src; uint8_t *dst_orig = dst; - int32_t h_cnt, v_cnt; + int32_t v_cnt; v16i8 edge_idx = { 1, 2, 0, 3, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; v16u8 const1 = (v16u8) __msa_ldi_b(1); v16u8 cmp_minus10, cmp_plus10, diff_minus10, diff_plus10, cmp_minus11; @@ -2037,22 +2120,20 @@ static void hevc_sao_edge_filter_45degree_16multiple_msa(uint8_t *dst, v16u8 src12, src_minus12, dst2, src13, src_minus13, dst3; v16i8 src_zero0, src_plus10, src_zero1, src_plus11, src_zero2, src_plus12; v16i8 src_zero3, sao_offset; - v8i16 src0, src1, src2, src3, src4, src5, src6, src7; - v8i16 temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7; sao_offset = LD_SB(sao_offset_val); sao_offset = __msa_pckev_b(sao_offset, sao_offset); - for (h_cnt = (height >> 2); h_cnt--;) { + for (; height; height -= 4) { src_orig = src - 1; dst_orig = dst; - LD_UB4(src_orig, src_stride, - src_minus11, src_minus12, src_minus13, src_minus14); + LD_UB4(src_orig, src_stride, src_minus11, src_minus12, src_minus13, + src_minus14); - for (v_cnt = 0; v_cnt < (width >> 4); v_cnt++) { + for (v_cnt = 0; v_cnt < width; v_cnt += 16) { src_minus10 = LD_UB(src_orig - src_stride); LD_UB4(src_orig + 16, src_stride, src10, src11, src12, src13); - src_plus13 = LD_UB(src + 1 + (v_cnt << 4) + (src_stride << 2)); + src_plus13 = LD_UB(src + 1 + v_cnt + (src_stride << 2)); src_orig += 16; SLDI_B2_SB(src10, src11, src_minus11, src_minus12, src_zero0, @@ -2114,22 +2195,14 @@ static void hevc_sao_edge_filter_45degree_16multiple_msa(uint8_t *dst, VSHF_B2_SB(edge_idx, edge_idx, sao_offset, sao_offset, offset_mask3, offset_mask3, offset_mask3, offset_mask3); - UNPCK_UB_SH(src_zero0, src0, src1); - UNPCK_SB_SH(offset_mask0, temp0, temp1); - UNPCK_UB_SH(src_zero1, src2, src3); - UNPCK_SB_SH(offset_mask1, temp2, temp3); - UNPCK_UB_SH(src_zero2, src4, src5); - UNPCK_SB_SH(offset_mask2, temp4, temp5); - UNPCK_UB_SH(src_zero3, src6, src7); - UNPCK_SB_SH(offset_mask3, temp6, temp7); - ADD4(temp0, src0, temp1, src1, temp2, src2, temp3, src3, temp0, - temp1, temp2, temp3); - ADD4(temp4, src4, temp5, src5, temp6, src6, temp7, src7, temp4, - temp5, temp6, temp7); - CLIP_SH4_0_255(temp0, temp1, temp2, temp3); - CLIP_SH4_0_255(temp4, temp5, temp6, temp7); - PCKEV_B4_UB(temp1, temp0, temp3, temp2, temp5, temp4, - temp7, temp6, dst0, dst1, dst2, dst3); + XORI_B4_128_SB(src_zero0, src_zero1, src_zero2, src_zero3); + + dst0 = (v16u8) __msa_adds_s_b((v16i8) src_zero0, offset_mask0); + dst1 = (v16u8) __msa_adds_s_b((v16i8) src_zero1, offset_mask1); + dst2 = (v16u8) __msa_adds_s_b((v16i8) src_zero2, offset_mask2); + dst3 = (v16u8) __msa_adds_s_b((v16i8) src_zero3, offset_mask3); + + XORI_B4_128_UB(dst0, dst1, dst2, dst3); src_minus11 = src10; src_minus12 = src11;