From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org [79.124.17.100]) by master.gitmailbox.com (Postfix) with ESMTP id AD5EE4C348 for ; Thu, 25 Jul 2024 18:15:34 +0000 (UTC) Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 4345C68D6A9; Thu, 25 Jul 2024 21:15:31 +0300 (EEST) Received: from ursule.remlab.net (vps-a2bccee9.vps.ovh.net [51.75.19.47]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 86D4668D097 for ; Thu, 25 Jul 2024 21:15:24 +0300 (EEST) Received: from basile.remlab.net (localhost [IPv6:::1]) by ursule.remlab.net (Postfix) with ESMTP id 2223CC013B for ; Thu, 25 Jul 2024 21:15:24 +0300 (EEST) From: =?UTF-8?q?R=C3=A9mi=20Denis-Courmont?= To: ffmpeg-devel@ffmpeg.org Date: Thu, 25 Jul 2024 21:15:23 +0300 Message-ID: <20240725181523.93924-1-remi@remlab.net> X-Mailer: git-send-email 2.45.2 MIME-Version: 1.0 Subject: [FFmpeg-devel] [PATCH] lavc/h264dsp: use saturing add/sub for R-V V 8-bit DC add 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 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" Archived-At: List-Archive: List-Post: T-Head C908 (cycles): h264_idct4_dc_add_8bpp_c: 109.2 h264_idct4_dc_add_8bpp_rvv_i32: 34.5 (before) h264_idct4_dc_add_8bpp_rvv_i32: 25.5 (after) h264_idct8_dc_add_8bpp_c: 418.7 h264_idct8_dc_add_8bpp_rvv_i64: 69.5 (before) h264_idct8_dc_add_8bpp_rvv_i64: 33.5 (after) --- libavcodec/riscv/h264dsp_init.c | 5 +++-- libavcodec/riscv/h264idct_rvv.S | 28 ++++++++++++++++++---------- 2 files changed, 21 insertions(+), 12 deletions(-) diff --git a/libavcodec/riscv/h264dsp_init.c b/libavcodec/riscv/h264dsp_init.c index 836c073559..671330d664 100644 --- a/libavcodec/riscv/h264dsp_init.c +++ b/libavcodec/riscv/h264dsp_init.c @@ -98,8 +98,8 @@ av_cold void ff_h264dsp_init_riscv(H264DSPContext *dsp, const int bit_depth, dsp->h264_idct_add = ff_h264_idct_add_8_rvv; dsp->h264_idct8_add = ff_h264_idct8_add_8_rvv; - dsp->h264_idct_dc_add = ff_h264_idct4_dc_add_8_rvv; if (flags & AV_CPU_FLAG_RVB) { + dsp->h264_idct_dc_add = ff_h264_idct4_dc_add_8_rvv; dsp->h264_idct_add16 = ff_h264_idct_add16_8_rvv; dsp->h264_idct_add16intra = ff_h264_idct_add16intra_8_rvv; # if __riscv_xlen == 64 @@ -108,7 +108,8 @@ av_cold void ff_h264dsp_init_riscv(H264DSPContext *dsp, const int bit_depth, } if (flags & AV_CPU_FLAG_RVV_I64) { dsp->h264_add_pixels8_clear = ff_h264_add_pixels8_8_rvv; - dsp->h264_idct8_dc_add = ff_h264_idct8_dc_add_8_rvv; + if (flags & AV_CPU_FLAG_RVB) + dsp->h264_idct8_dc_add = ff_h264_idct8_dc_add_8_rvv; } dsp->h264_add_pixels4_clear = ff_h264_add_pixels4_8_rvv; } diff --git a/libavcodec/riscv/h264idct_rvv.S b/libavcodec/riscv/h264idct_rvv.S index a49a32c47e..3c547f1eb0 100644 --- a/libavcodec/riscv/h264idct_rvv.S +++ b/libavcodec/riscv/h264idct_rvv.S @@ -416,33 +416,41 @@ endfunc .endr .macro idct_dc_add8 width -func ff_h264_idct\width\()_dc_add_8_rvv, zve64x +func ff_h264_idct\width\()_dc_add_8_rvv, zve64x, b .if \width == 8 vsetivli zero, \width, e8, mf2, ta, ma .else vsetivli zero, \width, e8, mf4, ta, ma .endif lh t0, 0(a1) + li t1, 255 addi t0, t0, 32 srai t0, t0, 6 sh zero, 0(a1) .if \width == 8 li a6, \width * \width vlse64.v v24, (a0), a2 - vsetvli zero, a6, e16, m8, ta, ma + vsetvli zero, a6, e8, m4, ta, ma .else vlse32.v v24, (a0), a2 - vsetivli zero, \width * \width, e16, m2, ta, ma + vsetivli zero, \width * \width, e8, m1, ta, ma .endif - vzext.vf2 v0, v24 - vadd.vx v0, v0, t0 - vmax.vx v0, v0, zero -.if \width == 8 - vsetvli zero, zero, e8, m4, ta, ma + bgez t0, 1f + + neg t0, t0 + minu t0, t0, t1 + vssubu.vx v24, v24, t0 + .if \width == 8 + vsetivli zero, \width, e8, mf2, ta, ma + vsse64.v v24, (a0), a2 .else - vsetvli zero, zero, e8, m1, ta, ma + vsetivli zero, \width, e8, mf4, ta, ma + vsse32.v v24, (a0), a2 .endif - vnclipu.wi v24, v0, 0 + ret +1: + minu t0, t0, t1 + vsaddu.vx v24, v24, t0 .if \width == 8 vsetivli zero, \width, e8, mf2, ta, ma vsse64.v v24, (a0), a2 -- 2.45.2 _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".