From: Shen Peiting <shenpeit...@eswincomputing.com> Use RVB instruction clz to calculate the number of leading zeros of MSB instead of av_log2.
Benchmarks on Spike(cycles): ac3_extract_exponents_c: 8226 ac3_extract_exponents_rvb: 1167 Co-Authored by: Yang Xiaojun <yangxiao...@eswincomputing.com> Co-Authored by: Huang Xing <huangxi...@eswincomputing.com> Co-Authored by: Zeng Fanchen <zengfanc...@eswincomputing.com> Signed-off-by: Shen Peiting <shenpeit...@eswincomputing.com> --- libavcodec/riscv/Makefile | 3 ++- libavcodec/riscv/ac3dsp_init.c | 3 +++ libavcodec/riscv/ac3dsp_rvb.S | 42 ++++++++++++++++++++++++++++++++++ 3 files changed, 47 insertions(+), 1 deletion(-) create mode 100644 libavcodec/riscv/ac3dsp_rvb.S diff --git a/libavcodec/riscv/Makefile b/libavcodec/riscv/Makefile index a627924cac..3d0c196cb9 100644 --- a/libavcodec/riscv/Makefile +++ b/libavcodec/riscv/Makefile @@ -1,7 +1,8 @@ OBJS-$(CONFIG_AAC_DECODER) += riscv/aacpsdsp_init.o RVV-OBJS-$(CONFIG_AAC_DECODER) += riscv/aacpsdsp_rvv.o OBJS-$(CONFIG_AC3DSP) += riscv/ac3dsp_init.o -RVV-OBJS-$(CONFIG_AC3DSP) += riscv/ac3dsp_rvv.o +RVV-OBJS-$(CONFIG_AC3DSP) += riscv/ac3dsp_rvv.o \ + riscv/ac3dsp_rvb.o OBJS-$(CONFIG_ALAC_DECODER) += riscv/alacdsp_init.o RVV-OBJS-$(CONFIG_ALAC_DECODER) += riscv/alacdsp_rvv.o OBJS-$(CONFIG_AUDIODSP) += riscv/audiodsp_init.o \ diff --git a/libavcodec/riscv/ac3dsp_init.c b/libavcodec/riscv/ac3dsp_init.c index 4769213ebc..75cd3c7e11 100644 --- a/libavcodec/riscv/ac3dsp_init.c +++ b/libavcodec/riscv/ac3dsp_init.c @@ -26,6 +26,7 @@ void ff_ac3_exponent_min_rvv(uint8_t *exp, int num_reuse_blocks, int nb_coefs); void ff_float_to_fixed24_rvv(int32_t *dst, const float *src, unsigned int len); +void ff_ac3_extract_exponents_rvb(uint8_t *exp, int32_t *coef, int nb_coefs); void ff_ac3_sum_square_butterfly_int32_rvv(int64_t sum[4], const int32_t *coef0, const int32_t *coef1, @@ -40,6 +41,8 @@ void ff_ac3_compute_mantissa_size_rvv(uint16_t mant_cnt[6][16]); av_cold void ff_ac3dsp_init_riscv(AC3DSPContext *c) { int flags = av_get_cpu_flags(); + if (flags & AV_CPU_FLAG_RVB_BASIC) + c->extract_exponents = ff_ac3_extract_exponents_rvb; #if HAVE_RVV if (flags & AV_CPU_FLAG_RVV_I32) { c->ac3_exponent_min = ff_ac3_exponent_min_rvv; diff --git a/libavcodec/riscv/ac3dsp_rvb.S b/libavcodec/riscv/ac3dsp_rvb.S new file mode 100644 index 0000000000..3bf24c7392 --- /dev/null +++ b/libavcodec/riscv/ac3dsp_rvb.S @@ -0,0 +1,42 @@ +/* + * Copyright 2023 Beijing ESWIN Computing Technology Co., Ltd. + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "config.h" +#include "libavutil/riscv/asm.S" + +func ff_ac3_extract_exponents_rvb, zbb + li t1, __riscv_xlen - 24 +1: + lw t0, (a1) + bgez t0, 2f + neg t0, t0 + +2: + clz t4, t0 + sub t4, t4, t1 + sb t4,(a0) + addi a2, a2, -1 + addi a1, a1, 4 + addi a0, a0, 1 + + bgtz a2, 1b + + ret +endfunc \ No newline at end of file -- 2.17.1 _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".