hashirama/uvg266
1
0
Fork 0
mirror of https://github.com/ultravideo/uvg266.git synced 2024-11-28 03:34:06 +00:00
Code Issues Actions Projects Releases Wiki Activity

Tidy 32x32 fast DCT's helper functions

Browse source
This commit is contained in:
Pauli Oikkonen 2019-10-10 18:26:59 +03:00
parent e382339182
commit e9da2d851b
1 changed files with 11 additions and 6 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

17
src/strategies/avx2/dct-avx2.c
View file

@ -855,7 +855,10 @@ static void matrix_dct_16x16_avx2(int8_t bitdepth, const int16_t *input, int16_t
matmul_16x16_a_bt (dct, tmpres, output, shift_2nd); matmul_16x16_a_bt (dct, tmpres, output, shift_2nd);
} }
static __m256i get_overflows(const __m256i a, const __m256i b, const __m256i res, const __m256i of_adjust_mask) static __m256i get_overflows(const __m256i a,
const __m256i b,
const __m256i res,
const __m256i of_adjust_mask)
{ {
const __m256i ones = _mm256_set1_epi16(1); const __m256i ones = _mm256_set1_epi16(1);
@ -880,7 +883,8 @@ static __m256i get_overflows(const __m256i a, const __m256i b, const __m256i res
* of_possible_mask is either all zero bits for subtraction, or all ones for * of_possible_mask is either all zero bits for subtraction, or all ones for
* addition * addition
*/ */
static void sub_16_16_hilo(const __m256i a, const __m256i b, __m256i *lo, __m256i *hi) static void sub_16_16_hilo(const __m256i a, const __m256i b,
__m256i *lo, __m256i *hi)
{ {
const __m256i zero = _mm256_setzero_si256(); const __m256i zero = _mm256_setzero_si256();
@ -888,7 +892,8 @@ static void sub_16_16_hilo(const __m256i a, const __m256i b, __m256i *lo, __m256
*hi = get_overflows(a, b, *lo, zero); *hi = get_overflows(a, b, *lo, zero);
} }
static void add_16_16_hilo(const __m256i a, const __m256i b, __m256i *lo, __m256i *hi) static void add_16_16_hilo(const __m256i a, const __m256i b,
__m256i *lo, __m256i *hi)
{ {
const __m256i ff = _mm256_set1_epi8(-1); const __m256i ff = _mm256_set1_epi8(-1);
@ -896,7 +901,7 @@ static void add_16_16_hilo(const __m256i a, const __m256i b, __m256i *lo, __m256
*hi = get_overflows(a, b, *lo, ff); *hi = get_overflows(a, b, *lo, ff);
} }
static __m256i reverse_16x16b_in_lanes(const __m256i v) static INLINE __m256i reverse_16x16b_in_lanes(const __m256i v)
{ {
const __m256i lanerev = _mm256_setr_epi16(0x0f0e, 0x0d0c, 0x0b0a, 0x0908, const __m256i lanerev = _mm256_setr_epi16(0x0f0e, 0x0d0c, 0x0b0a, 0x0908,
0x0706, 0x0504, 0x0302, 0x0100, 0x0706, 0x0504, 0x0302, 0x0100,
@ -905,13 +910,13 @@ static __m256i reverse_16x16b_in_lanes(const __m256i v)
return _mm256_shuffle_epi8(v, lanerev); return _mm256_shuffle_epi8(v, lanerev);
} }
static __m256i reverse_16x16b(const __m256i v) static INLINE __m256i reverse_16x16b(const __m256i v)
{ {
__m256i tmp = reverse_16x16b_in_lanes(v); __m256i tmp = reverse_16x16b_in_lanes(v);
return _mm256_permute4x64_epi64(tmp, _MM_SHUFFLE(1, 0, 3, 2)); return _mm256_permute4x64_epi64(tmp, _MM_SHUFFLE(1, 0, 3, 2));
} }
static __m256i m256_from_2xm128(const __m128i lo, const __m128i hi) static INLINE __m256i m256_from_2xm128(const __m128i lo, const __m128i hi)
{ {
__m256i result = _mm256_castsi128_si256 (lo); __m256i result = _mm256_castsi128_si256 (lo);
return _mm256_inserti128_si256(result, hi, 1); return _mm256_inserti128_si256(result, hi, 1);