|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
#include <assert.h> |
|
|
|
#include <immintrin.h> |
|
|
|
#include <xnnpack/gavgpool.h> |
|
#include <xnnpack/intrinsics-polyfill.h> |
|
|
|
|
|
void xnn_f16_gavgpool_minmax_ukernel_7x__f16c_c24( |
|
size_t rows, |
|
size_t channels, |
|
const void* input, |
|
size_t input_stride, |
|
const void* zero, |
|
void* output, |
|
const union xnn_f16_scaleminmax_params params[restrict XNN_MIN_ELEMENTS(1)]) XNN_OOB_READS |
|
{ |
|
assert(rows != 0); |
|
assert(rows <= 7); |
|
assert(channels != 0); |
|
|
|
const uint16_t* i0 = input; |
|
const uint16_t* i1 = (const uint16_t*) ((uintptr_t) i0 + input_stride); |
|
if XNN_UNPREDICTABLE(rows < 2) { |
|
i1 = (const uint16_t*) zero; |
|
} |
|
const uint16_t* i2 = (const uint16_t*) ((uintptr_t) i1 + input_stride); |
|
if XNN_UNPREDICTABLE(rows <= 2) { |
|
i2 = (const uint16_t*) zero; |
|
} |
|
const uint16_t* i3 = (const uint16_t*) ((uintptr_t) i2 + input_stride); |
|
if XNN_UNPREDICTABLE(rows < 4) { |
|
i3 = (const uint16_t*) zero; |
|
} |
|
const uint16_t* i4 = (const uint16_t*) ((uintptr_t) i3 + input_stride); |
|
if XNN_UNPREDICTABLE(rows <= 4) { |
|
i4 = (const uint16_t*) zero; |
|
} |
|
const uint16_t* i5 = (const uint16_t*) ((uintptr_t) i4 + input_stride); |
|
if XNN_UNPREDICTABLE(rows < 6) { |
|
i5 = (const uint16_t*) zero; |
|
} |
|
const uint16_t* i6 = (const uint16_t*) ((uintptr_t) i5 + input_stride); |
|
if XNN_UNPREDICTABLE(rows <= 6) { |
|
i6 = (const uint16_t*) zero; |
|
} |
|
uint16_t* o = (uint16_t*) output; |
|
|
|
const __m256 vscale = _mm256_load_ps(params->avx.scale); |
|
const __m256 vmin = _mm256_load_ps(params->avx.min); |
|
const __m256 vmax = _mm256_load_ps(params->avx.max); |
|
for (; channels >= 24; channels -= 24) { |
|
const __m256 vi0x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i0)); |
|
const __m256 vi0x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i0 + 8))); |
|
const __m256 vi0xGHIJKLMN = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i0 + 16))); |
|
i0 += 24; |
|
const __m256 vi1x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i1)); |
|
const __m256 vi1x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i1 + 8))); |
|
const __m256 vi1xGHIJKLMN = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i1 + 16))); |
|
i1 += 24; |
|
|
|
const __m256 vi2x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i2)); |
|
__m128i vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(vi0x01234567, vi1x01234567), _MM_FROUND_TO_NEAREST_INT); |
|
const __m256 vi2x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i2 + 8))); |
|
__m128i vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(vi0x89ABCDEF, vi1x89ABCDEF), _MM_FROUND_TO_NEAREST_INT); |
|
const __m256 vi2xGHIJKLMN = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i2 + 16))); |
|
__m128i vaccGHIJKLMN = _mm256_cvtps_ph(_mm256_add_ps(vi0xGHIJKLMN, vi1xGHIJKLMN), _MM_FROUND_TO_NEAREST_INT); |
|
i2 += 24; |
|
|
|
const __m256 vi3x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i3)); |
|
vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi2x01234567), _MM_FROUND_TO_NEAREST_INT); |
|
const __m256 vi3x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i3 + 8))); |
|
vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi2x89ABCDEF), _MM_FROUND_TO_NEAREST_INT); |
|
const __m256 vi3xGHIJKLMN = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i3 + 16))); |
|
i3 += 24; |
|
vaccGHIJKLMN = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vaccGHIJKLMN), vi2xGHIJKLMN), _MM_FROUND_TO_NEAREST_INT); |
|
const __m256 vi4x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i4)); |
|
vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi3x01234567), _MM_FROUND_TO_NEAREST_INT); |
|
const __m256 vi4x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i4 + 8))); |
|
vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi3x89ABCDEF), _MM_FROUND_TO_NEAREST_INT); |
|
const __m256 vi4xGHIJKLMN = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i4 + 16))); |
|
i4 += 24; |
|
vaccGHIJKLMN = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vaccGHIJKLMN), vi3xGHIJKLMN), _MM_FROUND_TO_NEAREST_INT); |
|
const __m256 vi5x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i5)); |
|
vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi4x01234567), _MM_FROUND_TO_NEAREST_INT); |
|
const __m256 vi5x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i5 + 8))); |
|
vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi4x89ABCDEF), _MM_FROUND_TO_NEAREST_INT); |
|
const __m256 vi5xGHIJKLMN = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i5 + 16))); |
|
i5 += 24; |
|
vaccGHIJKLMN = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vaccGHIJKLMN), vi4xGHIJKLMN), _MM_FROUND_TO_NEAREST_INT); |
|
const __m256 vi6x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i6)); |
|
vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi5x01234567), _MM_FROUND_TO_NEAREST_INT); |
|
const __m256 vi6x89ABCDEF = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i6 + 8))); |
|
vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi5x89ABCDEF), _MM_FROUND_TO_NEAREST_INT); |
|
const __m256 vi6xGHIJKLMN = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) (i6 + 16))); |
|
i6 += 24; |
|
vaccGHIJKLMN = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vaccGHIJKLMN), vi5xGHIJKLMN), _MM_FROUND_TO_NEAREST_INT); |
|
vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi6x01234567), _MM_FROUND_TO_NEAREST_INT); |
|
vacc89ABCDEF = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc89ABCDEF), vi6x89ABCDEF), _MM_FROUND_TO_NEAREST_INT); |
|
vaccGHIJKLMN = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vaccGHIJKLMN), vi6xGHIJKLMN), _MM_FROUND_TO_NEAREST_INT); |
|
|
|
vacc01234567 = _mm256_cvtps_ph(_mm256_mul_ps(_mm256_cvtph_ps(vacc01234567), vscale), _MM_FROUND_TO_NEAREST_INT); |
|
vacc89ABCDEF = _mm256_cvtps_ph(_mm256_mul_ps(_mm256_cvtph_ps(vacc89ABCDEF), vscale), _MM_FROUND_TO_NEAREST_INT); |
|
vaccGHIJKLMN = _mm256_cvtps_ph(_mm256_mul_ps(_mm256_cvtph_ps(vaccGHIJKLMN), vscale), _MM_FROUND_TO_NEAREST_INT); |
|
|
|
__m256 vout01234567 = _mm256_max_ps(_mm256_cvtph_ps(vacc01234567), vmin); |
|
__m256 vout89ABCDEF = _mm256_max_ps(_mm256_cvtph_ps(vacc89ABCDEF), vmin); |
|
__m256 voutGHIJKLMN = _mm256_max_ps(_mm256_cvtph_ps(vaccGHIJKLMN), vmin); |
|
|
|
vout01234567 = _mm256_min_ps(vout01234567, vmax); |
|
vout89ABCDEF = _mm256_min_ps(vout89ABCDEF, vmax); |
|
voutGHIJKLMN = _mm256_min_ps(voutGHIJKLMN, vmax); |
|
|
|
_mm_storeu_si128((__m128i*) o, _mm256_cvtps_ph(vout01234567, _MM_FROUND_TO_NEAREST_INT)); |
|
_mm_storeu_si128((__m128i*) (o + 8), _mm256_cvtps_ph(vout89ABCDEF, _MM_FROUND_TO_NEAREST_INT)); |
|
_mm_storeu_si128((__m128i*) (o + 16), _mm256_cvtps_ph(voutGHIJKLMN, _MM_FROUND_TO_NEAREST_INT)); |
|
o += 24; |
|
} |
|
if XNN_UNLIKELY(channels != 0) { |
|
do { |
|
const __m256 vi0x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i0)); |
|
i0 += 8; |
|
const __m256 vi1x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i1)); |
|
i1 += 8; |
|
|
|
const __m256 vi2x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i2)); |
|
__m128i vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(vi0x01234567, vi1x01234567), _MM_FROUND_TO_NEAREST_INT); |
|
i2 += 8; |
|
|
|
const __m256 vi3x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i3)); |
|
i3 += 8; |
|
vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi2x01234567), _MM_FROUND_TO_NEAREST_INT); |
|
const __m256 vi4x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i4)); |
|
i4 += 8; |
|
vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi3x01234567), _MM_FROUND_TO_NEAREST_INT); |
|
const __m256 vi5x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i5)); |
|
i5 += 8; |
|
vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi4x01234567), _MM_FROUND_TO_NEAREST_INT); |
|
const __m256 vi6x01234567 = _mm256_cvtph_ps(_mm_loadu_si128((const __m128i*) i6)); |
|
i6 += 8; |
|
vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi5x01234567), _MM_FROUND_TO_NEAREST_INT); |
|
vacc01234567 = _mm256_cvtps_ph(_mm256_add_ps(_mm256_cvtph_ps(vacc01234567), vi6x01234567), _MM_FROUND_TO_NEAREST_INT); |
|
|
|
vacc01234567 = _mm256_cvtps_ph(_mm256_mul_ps(_mm256_cvtph_ps(vacc01234567), vscale), _MM_FROUND_TO_NEAREST_INT); |
|
__m256 vout01234567 = _mm256_max_ps(_mm256_cvtph_ps(vacc01234567), vmin); |
|
vout01234567 = _mm256_min_ps(vout01234567, vmax); |
|
|
|
if XNN_LIKELY(channels >= 8) { |
|
_mm_storeu_si128((__m128i*) o, _mm256_cvtps_ph(vout01234567, _MM_FROUND_TO_NEAREST_INT)); |
|
o += 8; |
|
channels -= 8; |
|
} else { |
|
__m128i vh01234567 = _mm256_cvtps_ph(vout01234567, _MM_FROUND_TO_NEAREST_INT); |
|
if (channels & 4) { |
|
_mm_storel_epi64((__m128i*) o, vh01234567); |
|
o += 4; |
|
vh01234567 = _mm_unpackhi_epi64(vh01234567, vh01234567); |
|
} |
|
if (channels & 2) { |
|
_mm_storeu_si32(o, vh01234567); |
|
o += 2; |
|
vh01234567 = _mm_srli_epi64(vh01234567, 32); |
|
} |
|
if (channels & 1) { |
|
*o = (uint16_t) _mm_extract_epi16(vh01234567, 0); |
|
} |
|
channels = 0; |
|
} |
|
} while (channels != 0); |
|
} |
|
} |
|
|