This is now firmly slower than C. SiFive-U74 (cycles): audiodsp.vector_clipf_c: 31.2 audiodsp.vector_clipf_rvf: 39.5
Unified · Split
| @@ -9,7 +9,6 @@ RVVB-OBJS-$(CONFIG_AC3DSP) += riscv/ac3dsp_rvvb.o · 8 unmodified lines | |||
| 9 | OBJS-$(CONFIG_ALAC_DECODER) += riscv/alacdsp_init.o | 9 | OBJS-$(CONFIG_ALAC_DECODER) += riscv/alacdsp_init.o |
| 10 | RVV-OBJS-$(CONFIG_ALAC_DECODER) += riscv/alacdsp_rvv.o | 10 | RVV-OBJS-$(CONFIG_ALAC_DECODER) += riscv/alacdsp_rvv.o |
| 11 | OBJS-$(CONFIG_AUDIODSP) += riscv/audiodsp_init.o | 11 | OBJS-$(CONFIG_AUDIODSP) += riscv/audiodsp_init.o |
| 12 | RV-OBJS-$(CONFIG_AUDIODSP) += riscv/audiodsp_rvf.o | ||
| 13 | RVV-OBJS-$(CONFIG_AUDIODSP) += riscv/audiodsp_rvv.o | 12 | RVV-OBJS-$(CONFIG_AUDIODSP) += riscv/audiodsp_rvv.o |
| 14 | OBJS-$(CONFIG_BLOCKDSP) += riscv/blockdsp_init.o | 13 | OBJS-$(CONFIG_BLOCKDSP) += riscv/blockdsp_init.o |
| 15 | RVV-OBJS-$(CONFIG_BLOCKDSP) += riscv/blockdsp_rvv.o | 14 | RVV-OBJS-$(CONFIG_BLOCKDSP) += riscv/blockdsp_rvv.o |
| @@ -24,8 +24,6 @@ · 23 unmodified lines | |||
| 24 | #include "libavutil/cpu.h" | 24 | #include "libavutil/cpu.h" |
| 25 | #include "libavcodec/audiodsp.h" | 25 | #include "libavcodec/audiodsp.h" |
| 26 | 26 | ||
| 27 | void ff_vector_clipf_rvf(float *dst, const float *src, int len, float min, float max); | ||
| 28 | |||
| 29 | int32_t ff_scalarproduct_int16_rvv(const int16_t *v1, const int16_t *v2, int len); | 27 | int32_t ff_scalarproduct_int16_rvv(const int16_t *v1, const int16_t *v2, int len); |
| 30 | void ff_vector_clip_int32_rvv(int32_t *dst, const int32_t *src, int32_t min, | 28 | void ff_vector_clip_int32_rvv(int32_t *dst, const int32_t *src, int32_t min, |
| 31 | int32_t max, unsigned int len); | 29 | int32_t max, unsigned int len); |
| @@ -33,12 +31,9 @@ void ff_vector_clipf_rvv(float *dst, const float *src, int len, float min, float · 1 unmodified line | |||
| 33 | 31 | ||
| 34 | av_cold void ff_audiodsp_init_riscv(AudioDSPContext *c) | 32 | av_cold void ff_audiodsp_init_riscv(AudioDSPContext *c) |
| 35 | { | 33 | { |
| 36 | #if HAVE_RV | 34 | #if HAVE_RVV |
| 37 | int flags = av_get_cpu_flags(); | 35 | int flags = av_get_cpu_flags(); |
| 38 | 36 | ||
| 39 | if (flags & AV_CPU_FLAG_RVF) | ||
| 40 | c->vector_clipf = ff_vector_clipf_rvf; | ||
| 41 | #if HAVE_RVV | ||
| 42 | if (flags & AV_CPU_FLAG_RVB_ADDR) { | 37 | if (flags & AV_CPU_FLAG_RVB_ADDR) { |
| 43 | if (flags & AV_CPU_FLAG_RVV_I32) { | 38 | if (flags & AV_CPU_FLAG_RVV_I32) { |
| 44 | c->scalarproduct_int16 = ff_scalarproduct_int16_rvv; | 39 | c->scalarproduct_int16 = ff_scalarproduct_int16_rvv; |
| @@ -48,5 +43,4 @@ av_cold void ff_audiodsp_init_riscv(AudioDSPContext *c) · 3 unmodified lines | |||
| 48 | c->vector_clipf = ff_vector_clipf_rvv; | 43 | c->vector_clipf = ff_vector_clipf_rvv; |
| 49 | } | 44 | } |
| 50 | #endif | 45 | #endif |
| 51 | #endif | ||
| 52 | } | 46 | } |
| @@ -1,50 +0,0 @@ | |||
| 1 | /* | ||
| 2 | * Copyright © 2022 Rémi Denis-Courmont. | ||
| 3 | * | ||
| 4 | * This file is part of FFmpeg. | ||
| 5 | * | ||
| 6 | * FFmpeg is free software; you can redistribute it and/or | ||
| 7 | * modify it under the terms of the GNU Lesser General Public | ||
| 8 | * License as published by the Free Software Foundation; either | ||
| 9 | * version 2.1 of the License, or (at your option) any later version. | ||
| 10 | * | ||
| 11 | * FFmpeg is distributed in the hope that it will be useful, | ||
| 12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | ||
| 13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | ||
| 14 | * Lesser General Public License for more details. | ||
| 15 | * | ||
| 16 | * You should have received a copy of the GNU Lesser General Public | ||
| 17 | * License along with FFmpeg; if not, write to the Free Software | ||
| 18 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA | ||
| 19 | */ | ||
| 20 | |||
| 21 | #include "libavutil/riscv/asm.S" | ||
| 22 | |||
| 23 | func ff_vector_clipf_rvf, f | ||
| 24 | lpad 0 | ||
| 25 | NOHWF fmv.w.x fa0, a3 | ||
| 26 | NOHWF fmv.w.x fa1, a4 | ||
| 27 | 1: | ||
| 28 | flw ft0, (a1) | ||
| 29 | flw ft1, 4(a1) | ||
| 30 | fmax.s ft0, ft0, fa0 | ||
| 31 | flw ft2, 8(a1) | ||
| 32 | fmax.s ft1, ft1, fa0 | ||
| 33 | flw ft3, 12(a1) | ||
| 34 | fmax.s ft2, ft2, fa0 | ||
| 35 | addi a2, a2, -4 | ||
| 36 | fmax.s ft3, ft3, fa0 | ||
| 37 | addi a1, a1, 16 | ||
| 38 | fmin.s ft0, ft0, fa1 | ||
| 39 | fmin.s ft1, ft1, fa1 | ||
| 40 | fsw ft0, (a0) | ||
| 41 | fmin.s ft2, ft2, fa1 | ||
| 42 | fsw ft1, 4(a0) | ||
| 43 | fmin.s ft3, ft3, fa1 | ||
| 44 | fsw ft2, 8(a0) | ||
| 45 | fsw ft3, 12(a0) | ||
| 46 | addi a0, a0, 16 | ||
| 47 | bnez a2, 1b | ||
| 48 | |||
| 49 | ret | ||
| 50 | endfunc | ||