提交 01916248 编写于 作者: V Vitaly Tuzov

Disabled universal intrinsic based implementation for bit-exact resize of 3-channel images

上级 1bc1f3d3
......@@ -443,104 +443,6 @@ void hlineResizeCn<uint8_t, ufixedpoint16, 2, true, 2>(uint8_t* src, int, int *o
}
}
template <>
void hlineResizeCn<uint8_t, ufixedpoint16, 2, true, 3>(uint8_t* src, int, int *ofst, ufixedpoint16* m, ufixedpoint16* dst, int dst_min, int dst_max, int dst_width)
{
int i = 0;
ufixedpoint16 srccn[3] = { src[0], src[1], src[2] };
v_uint16x8 v_srccn0 = v_setall_u16(((uint16_t*)srccn)[0]);
v_uint16x8 v_srccn1 = v_setall_u16(((uint16_t*)srccn)[1]);
v_uint16x8 v_srccn2 = v_setall_u16(((uint16_t*)srccn)[2]);
for (; i < dst_min - 7; i += 8, m += 16, dst += 24) // Points that fall left from src image so became equal to leftmost src point
{
v_store_interleave((uint16_t*)dst, v_srccn0, v_srccn1, v_srccn2);
}
for (; i < dst_min; i++, m += 2)
{
*(dst++) = srccn[0];
*(dst++) = srccn[1];
*(dst++) = srccn[2];
}
for (; i < dst_max - 7 && ofst[i + 7] + 15 <= ofst[dst_width - 1]; i += 8, m += 16, dst += 24)
{
v_uint8x16 v_src0_c1, v_src0_c2, v_src0_c3;
v_load_deinterleave(src + 3 * ofst[i ], v_src0_c1, v_src0_c2, v_src0_c3);
v_uint8x16 v_src1_c1, v_src1_c2, v_src1_c3;
v_load_deinterleave(src + 3 * ofst[i + 1], v_src1_c1, v_src1_c2, v_src1_c3);
v_uint8x16 v_src2_c1, v_src2_c2, v_src2_c3;
v_load_deinterleave(src + 3 * ofst[i + 2], v_src2_c1, v_src2_c2, v_src2_c3);
v_uint8x16 v_src3_c1, v_src3_c2, v_src3_c3;
v_load_deinterleave(src + 3 * ofst[i + 3], v_src3_c1, v_src3_c2, v_src3_c3);
v_uint16x8 v_tmp0, v_tmp1, v_tmp2;
v_uint16x8 v_src0123_c1, v_src0123_c2, v_src0123_c3;
v_zip(v_reinterpret_as_u16(v_src0_c1), v_reinterpret_as_u16(v_src2_c1), v_tmp0, v_tmp1);
v_zip(v_reinterpret_as_u16(v_src1_c1), v_reinterpret_as_u16(v_src3_c1), v_tmp1, v_tmp2);
v_zip(v_tmp0, v_tmp1, v_src0123_c1, v_tmp2);
v_zip(v_reinterpret_as_u16(v_src0_c2), v_reinterpret_as_u16(v_src2_c2), v_tmp0, v_tmp1);
v_zip(v_reinterpret_as_u16(v_src1_c2), v_reinterpret_as_u16(v_src3_c2), v_tmp1, v_tmp2);
v_zip(v_tmp0, v_tmp1, v_src0123_c2, v_tmp2);
v_zip(v_reinterpret_as_u16(v_src0_c3), v_reinterpret_as_u16(v_src2_c3), v_tmp0, v_tmp1);
v_zip(v_reinterpret_as_u16(v_src1_c3), v_reinterpret_as_u16(v_src3_c3), v_tmp1, v_tmp2);
v_zip(v_tmp0, v_tmp1, v_src0123_c3, v_tmp2);
v_load_deinterleave(src + 3 * ofst[i + 4], v_src0_c1, v_src0_c2, v_src0_c3);
v_load_deinterleave(src + 3 * ofst[i + 5], v_src1_c1, v_src1_c2, v_src1_c3);
v_load_deinterleave(src + 3 * ofst[i + 6], v_src2_c1, v_src2_c2, v_src2_c3);
v_load_deinterleave(src + 3 * ofst[i + 7], v_src3_c1, v_src3_c2, v_src3_c3);
v_uint16x8 v_src4567_c1, v_src4567_c2, v_src4567_c3;
v_zip(v_reinterpret_as_u16(v_src0_c1), v_reinterpret_as_u16(v_src2_c1), v_tmp0, v_tmp1);
v_zip(v_reinterpret_as_u16(v_src1_c1), v_reinterpret_as_u16(v_src3_c1), v_tmp1, v_tmp2);
v_zip(v_tmp0, v_tmp1, v_src4567_c1, v_tmp2);
v_zip(v_reinterpret_as_u16(v_src0_c2), v_reinterpret_as_u16(v_src2_c2), v_tmp0, v_tmp1);
v_zip(v_reinterpret_as_u16(v_src1_c2), v_reinterpret_as_u16(v_src3_c2), v_tmp1, v_tmp2);
v_zip(v_tmp0, v_tmp1, v_src4567_c2, v_tmp2);
v_zip(v_reinterpret_as_u16(v_src0_c3), v_reinterpret_as_u16(v_src2_c3), v_tmp0, v_tmp1);
v_zip(v_reinterpret_as_u16(v_src1_c3), v_reinterpret_as_u16(v_src3_c3), v_tmp1, v_tmp2);
v_zip(v_tmp0, v_tmp1, v_src4567_c3, v_tmp2);
v_expand(v_reinterpret_as_u8(v_combine_low(v_src0123_c1, v_src4567_c1)),
v_src0123_c1, v_src4567_c1
);
v_expand(v_reinterpret_as_u8(v_combine_low(v_src0123_c2, v_src4567_c2)),
v_src0123_c2, v_src4567_c2
);
v_expand(v_reinterpret_as_u8(v_combine_low(v_src0123_c3, v_src4567_c3)),
v_src0123_c3, v_src4567_c3
);
v_int16x8 v_mul0123 = v_load((int16_t*)m);
v_int16x8 v_mul4567 = v_load((int16_t*)m+8);
v_uint32x4 v_res0123_c1 = v_reinterpret_as_u32(v_dotprod(v_reinterpret_as_s16(v_src0123_c1), v_mul0123));
v_uint32x4 v_res0123_c2 = v_reinterpret_as_u32(v_dotprod(v_reinterpret_as_s16(v_src0123_c2), v_mul0123));
v_uint32x4 v_res0123_c3 = v_reinterpret_as_u32(v_dotprod(v_reinterpret_as_s16(v_src0123_c3), v_mul0123));
v_uint32x4 v_res4567_c1 = v_reinterpret_as_u32(v_dotprod(v_reinterpret_as_s16(v_src4567_c1), v_mul4567));
v_uint32x4 v_res4567_c2 = v_reinterpret_as_u32(v_dotprod(v_reinterpret_as_s16(v_src4567_c2), v_mul4567));
v_uint32x4 v_res4567_c3 = v_reinterpret_as_u32(v_dotprod(v_reinterpret_as_s16(v_src4567_c3), v_mul4567));
v_store_interleave((uint16_t*)dst , v_pack(v_res0123_c1, v_res4567_c1), v_pack(v_res0123_c2, v_res4567_c2), v_pack(v_res0123_c3, v_res4567_c3));
}
for (; i < dst_max; i += 1, m += 2)
{
uint8_t* px = src + 3 * ofst[i];
*(dst++) = m[0] * px[0] + m[1] * px[3];
*(dst++) = m[0] * px[1] + m[1] * px[4];
*(dst++) = m[0] * px[2] + m[1] * px[5];
}
srccn[0] = (src + 3 * ofst[dst_width - 1])[0]; v_srccn0 = v_setall_u16(((uint16_t*)srccn)[0]);
srccn[1] = (src + 3 * ofst[dst_width - 1])[1]; v_srccn1 = v_setall_u16(((uint16_t*)srccn)[1]);
srccn[2] = (src + 3 * ofst[dst_width - 1])[2]; v_srccn2 = v_setall_u16(((uint16_t*)srccn)[2]);
for (; i < dst_width - 7; i += 8, dst += 24) // Points that fall left from src image so became equal to leftmost src point
{
v_store_interleave((uint16_t*)dst, v_srccn0, v_srccn1, v_srccn2);
}
for (; i < dst_width; i++)
{
*(dst++) = srccn[0];
*(dst++) = srccn[1];
*(dst++) = srccn[2];
}
}
template <>
void hlineResizeCn<uint8_t, ufixedpoint16, 2, true, 4>(uint8_t* src, int, int *ofst, ufixedpoint16* m, ufixedpoint16* dst, int dst_min, int dst_max, int dst_width)
{
int i = 0;
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册