Searched refs:vneg_mask_lo (Results 1 – 5 of 5) sorted by relevance
/external/XNNPACK/src/qu8-avgpool/ |
D | 9x-minmax-neon-c8.c | 141 const int32x4_t vneg_mask_lo = vreinterpretq_s32_u32(vcltq_s32(vacc_lo, vmovq_n_s32(0))); in xnn_qu8_avgpool_minmax_ukernel_9x__neon_c8() local 150 const int64x2_t vadjusted_product01 = vaddw_s32(vproduct01, vget_low_s32(vneg_mask_lo)); in xnn_qu8_avgpool_minmax_ukernel_9x__neon_c8() 151 const int64x2_t vadjusted_product23 = vaddw_high_s32(vproduct23, vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9x__neon_c8() 160 const int64x2_t vadjusted_product01 = vaddw_s32(vproduct01, vget_low_s32(vneg_mask_lo)); in xnn_qu8_avgpool_minmax_ukernel_9x__neon_c8() 161 const int64x2_t vadjusted_product23 = vaddw_s32(vproduct23, vget_high_s32(vneg_mask_lo)); in xnn_qu8_avgpool_minmax_ukernel_9x__neon_c8() 214 const int32x4_t vneg_mask_lo = vreinterpretq_s32_u32(vcltq_s32(vacc_lo, vmovq_n_s32(0))); in xnn_qu8_avgpool_minmax_ukernel_9x__neon_c8() local 223 const int64x2_t vadjusted_product01 = vaddw_s32(vproduct01, vget_low_s32(vneg_mask_lo)); in xnn_qu8_avgpool_minmax_ukernel_9x__neon_c8() 224 const int64x2_t vadjusted_product23 = vaddw_high_s32(vproduct23, vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9x__neon_c8() 233 const int64x2_t vadjusted_product01 = vaddw_s32(vproduct01, vget_low_s32(vneg_mask_lo)); in xnn_qu8_avgpool_minmax_ukernel_9x__neon_c8() 234 const int64x2_t vadjusted_product23 = vaddw_s32(vproduct23, vget_high_s32(vneg_mask_lo)); in xnn_qu8_avgpool_minmax_ukernel_9x__neon_c8()
|
D | 9x-minmax-sse2-c8.c | 146 const __m128i vneg_mask_lo = _mm_cmpgt_epi32(_mm_setzero_si128(), vacc_lo); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8() local 149 const __m128i vabs_lo0123 = _mm_sub_epi32(_mm_xor_si128(vacc_lo, vneg_mask_lo), vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8() 174 …onst __m128i vscaled_lo = _mm_sub_epi32(_mm_xor_si128(vabs_scaled_lo, vneg_mask_lo), vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8() 221 const __m128i vneg_mask_lo = _mm_cmpgt_epi32(_mm_setzero_si128(), vacc_lo); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8() local 224 const __m128i vabs_lo0123 = _mm_sub_epi32(_mm_xor_si128(vacc_lo, vneg_mask_lo), vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8() 249 …onst __m128i vscaled_lo = _mm_sub_epi32(_mm_xor_si128(vabs_scaled_lo, vneg_mask_lo), vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8()
|
D | 9p8x-minmax-neon-c8.c | 285 const int32x4_t vneg_mask_lo = vreinterpretq_s32_u32(vcltq_s32(vacc_lo, vmovq_n_s32(0))); in xnn_qu8_avgpool_minmax_ukernel_9p8x__neon_c8() local 294 const int64x2_t vadjusted_product01 = vaddw_s32(vproduct01, vget_low_s32(vneg_mask_lo)); in xnn_qu8_avgpool_minmax_ukernel_9p8x__neon_c8() 295 const int64x2_t vadjusted_product23 = vaddw_high_s32(vproduct23, vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9p8x__neon_c8() 304 const int64x2_t vadjusted_product01 = vaddw_s32(vproduct01, vget_low_s32(vneg_mask_lo)); in xnn_qu8_avgpool_minmax_ukernel_9p8x__neon_c8() 305 const int64x2_t vadjusted_product23 = vaddw_s32(vproduct23, vget_high_s32(vneg_mask_lo)); in xnn_qu8_avgpool_minmax_ukernel_9p8x__neon_c8() 359 const int32x4_t vneg_mask_lo = vreinterpretq_s32_u32(vcltq_s32(vacc_lo, vmovq_n_s32(0))); in xnn_qu8_avgpool_minmax_ukernel_9p8x__neon_c8() local 368 const int64x2_t vadjusted_product01 = vaddw_s32(vproduct01, vget_low_s32(vneg_mask_lo)); in xnn_qu8_avgpool_minmax_ukernel_9p8x__neon_c8() 369 const int64x2_t vadjusted_product23 = vaddw_high_s32(vproduct23, vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9p8x__neon_c8() 378 const int64x2_t vadjusted_product01 = vaddw_s32(vproduct01, vget_low_s32(vneg_mask_lo)); in xnn_qu8_avgpool_minmax_ukernel_9p8x__neon_c8() 379 const int64x2_t vadjusted_product23 = vaddw_s32(vproduct23, vget_high_s32(vneg_mask_lo)); in xnn_qu8_avgpool_minmax_ukernel_9p8x__neon_c8()
|
D | 9p8x-minmax-sse2-c8.c | 311 const __m128i vneg_mask_lo = _mm_cmpgt_epi32(_mm_setzero_si128(), vacc_lo); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8() local 314 … const __m128i vabs_lo0123 = _mm_sub_epi32(_mm_xor_si128(vacc_lo, vneg_mask_lo), vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8() 339 …onst __m128i vscaled_lo = _mm_sub_epi32(_mm_xor_si128(vabs_scaled_lo, vneg_mask_lo), vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8() 386 const __m128i vneg_mask_lo = _mm_cmpgt_epi32(_mm_setzero_si128(), vacc_lo); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8() local 389 … const __m128i vabs_lo0123 = _mm_sub_epi32(_mm_xor_si128(vacc_lo, vneg_mask_lo), vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8() 414 …onst __m128i vscaled_lo = _mm_sub_epi32(_mm_xor_si128(vabs_scaled_lo, vneg_mask_lo), vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8()
|
/external/XNNPACK/src/amalgam/ |
D | sse2.c | 8087 const __m128i vneg_mask_lo = _mm_cmpgt_epi32(_mm_setzero_si128(), vacc_lo); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8() local 8090 … const __m128i vabs_lo0123 = _mm_sub_epi32(_mm_xor_si128(vacc_lo, vneg_mask_lo), vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8() 8115 …onst __m128i vscaled_lo = _mm_sub_epi32(_mm_xor_si128(vabs_scaled_lo, vneg_mask_lo), vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8() 8162 const __m128i vneg_mask_lo = _mm_cmpgt_epi32(_mm_setzero_si128(), vacc_lo); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8() local 8165 … const __m128i vabs_lo0123 = _mm_sub_epi32(_mm_xor_si128(vacc_lo, vneg_mask_lo), vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8() 8190 …onst __m128i vscaled_lo = _mm_sub_epi32(_mm_xor_si128(vabs_scaled_lo, vneg_mask_lo), vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8() 8348 const __m128i vneg_mask_lo = _mm_cmpgt_epi32(_mm_setzero_si128(), vacc_lo); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8() local 8351 const __m128i vabs_lo0123 = _mm_sub_epi32(_mm_xor_si128(vacc_lo, vneg_mask_lo), vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8() 8376 …onst __m128i vscaled_lo = _mm_sub_epi32(_mm_xor_si128(vabs_scaled_lo, vneg_mask_lo), vneg_mask_lo); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8() 8423 const __m128i vneg_mask_lo = _mm_cmpgt_epi32(_mm_setzero_si128(), vacc_lo); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8() local [all …]
|