T-Head C908: ac3_exponent_min_reuse0_c: 7.5 ac3_exponent_min_reuse0_rvv_i32: 7.5 ac3_exponent_min_reuse1_c: 1820.7 ac3_exponent_min_reuse1_rvv_i32: 102.5 ac3_exponent_min_reuse2_c: 3088.5 ac3_exponent_min_reuse2_rvv_i32: 138.7 ac3_exponent_min_reuse3_c: 5073.7 ac3_exponent_min_reuse3_rvv_i32: 174.7 ac3_exponent_min_reuse4_c: 4624.2 ac3_exponent_min_reuse4_rvv_i32: 204.2 ac3_exponent_min_reuse5_c: 5138.7 ac3_exponent_min_reuse5_rvv_i32: 238.0 --- libavcodec/riscv/ac3dsp_init.c | 4 ++++ libavcodec/riscv/ac3dsp_rvv.S | 22 ++++++++++++++++++++++ 2 files changed, 26 insertions(+)
diff --git a/libavcodec/riscv/ac3dsp_init.c b/libavcodec/riscv/ac3dsp_init.c index e120aa2dce..c7c375273d 100644 --- a/libavcodec/riscv/ac3dsp_init.c +++ b/libavcodec/riscv/ac3dsp_init.c @@ -26,6 +26,7 @@ #include "libavutil/cpu.h" #include "libavcodec/ac3dsp.h" +void ff_ac3_exponent_min_rvv(uint8_t *exp, int, int); void ff_extract_exponents_rvb(uint8_t *exp, int32_t *coef, int nb_coefs); void ff_float_to_fixed24_rvv(int32_t *dst, const float *src, size_t len); void ff_sum_square_butterfly_int32_rvv(int64_t *, const int32_t *, @@ -38,6 +39,9 @@ av_cold void ff_ac3dsp_init_riscv(AC3DSPContext *c) #if HAVE_RV int flags = av_get_cpu_flags(); + if (flags & AV_CPU_FLAG_RVV_I32) + c->ac3_exponent_min = ff_ac3_exponent_min_rvv; + if (flags & AV_CPU_FLAG_RVB_ADDR) { if (flags & AV_CPU_FLAG_RVB_BASIC) c->extract_exponents = ff_extract_exponents_rvb; diff --git a/libavcodec/riscv/ac3dsp_rvv.S b/libavcodec/riscv/ac3dsp_rvv.S index 397e000ab0..1b5f67a9ec 100644 --- a/libavcodec/riscv/ac3dsp_rvv.S +++ b/libavcodec/riscv/ac3dsp_rvv.S @@ -21,6 +21,28 @@ #include "config.h" #include "libavutil/riscv/asm.S" +func ff_ac3_exponent_min_rvv, zve32x + beqz a1, 3f +1: + vsetvli t2, a2, e8, m8, ta, ma + vle8.v v8, (a0) + addi t0, a0, 256 + sub a2, a2, t2 + mv t1, a1 +2: + vle8.v v16, (t0) + addi t1, t1, -1 + vminu.vv v8, v8, v16 + addi t0, t0, 256 + bnez t1, 2b + + vse8.v v8, (a0) + add a0, a0, t2 + bnez a2, 1b +3: + ret +endfunc + func ff_float_to_fixed24_rvv, zve32f li t1, 1 << 24 fcvt.s.w f0, t1 -- 2.43.0 _______________________________________________ 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".