Searched refs:vsB (Results 1 – 8 of 8) sorted by relevance
/aosp_15_r20/external/XNNPACK/src/f16-raddstoreexpminusmax/gen/ |
H A D | neonfp16arith-rr2-p2-x96-acc2.c | 92 const float16x8_t vsB = vreinterpretq_f16_s16(vshlq_n_s16(vreinterpretq_s16_f16(vnB), 10)); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x96_acc2() local 157 vtB = vmulq_f16(vtB, vsB); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x96_acc2() 181 float16x8_t vfB = vfmaq_f16(vsB, vpB, vtB); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x96_acc2()
|
H A D | avx2-rr1-p2-x96-acc6.c | 96 const __m256 vsB = _mm256_castsi256_ps(_mm256_slli_epi32(_mm256_castps_si256(vnB), 23)); in xnn_f16_raddstoreexpminusmax_ukernel__avx2_rr1_p2_x96_acc6() local 148 vtB = _mm256_mul_ps(vtB, vsB); in xnn_f16_raddstoreexpminusmax_ukernel__avx2_rr1_p2_x96_acc6() 161 __m256 vfB = _mm256_fmadd_ps(vtB, vpB, vsB); in xnn_f16_raddstoreexpminusmax_ukernel__avx2_rr1_p2_x96_acc6()
|
H A D | neonfp16arith-rr2-p2-x96-acc3.c | 93 const float16x8_t vsB = vreinterpretq_f16_s16(vshlq_n_s16(vreinterpretq_s16_f16(vnB), 10)); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x96_acc3() local 158 vtB = vmulq_f16(vtB, vsB); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x96_acc3() 182 float16x8_t vfB = vfmaq_f16(vsB, vpB, vtB); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x96_acc3()
|
H A D | avx2-rr1-p2-x96.c | 91 const __m256 vsB = _mm256_castsi256_ps(_mm256_slli_epi32(_mm256_castps_si256(vnB), 23)); in xnn_f16_raddstoreexpminusmax_ukernel__avx2_rr1_p2_x96() local 143 vtB = _mm256_mul_ps(vtB, vsB); in xnn_f16_raddstoreexpminusmax_ukernel__avx2_rr1_p2_x96() 156 __m256 vfB = _mm256_fmadd_ps(vtB, vpB, vsB); in xnn_f16_raddstoreexpminusmax_ukernel__avx2_rr1_p2_x96()
|
H A D | neonfp16arith-rr2-p2-x96.c | 91 const float16x8_t vsB = vreinterpretq_f16_s16(vshlq_n_s16(vreinterpretq_s16_f16(vnB), 10)); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x96() local 156 vtB = vmulq_f16(vtB, vsB); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x96() 180 float16x8_t vfB = vfmaq_f16(vsB, vpB, vtB); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x96()
|
H A D | avx2-rr1-p2-x96-acc3.c | 93 const __m256 vsB = _mm256_castsi256_ps(_mm256_slli_epi32(_mm256_castps_si256(vnB), 23)); in xnn_f16_raddstoreexpminusmax_ukernel__avx2_rr1_p2_x96_acc3() local 145 vtB = _mm256_mul_ps(vtB, vsB); in xnn_f16_raddstoreexpminusmax_ukernel__avx2_rr1_p2_x96_acc3() 158 __m256 vfB = _mm256_fmadd_ps(vtB, vpB, vsB); in xnn_f16_raddstoreexpminusmax_ukernel__avx2_rr1_p2_x96_acc3()
|
H A D | avx2-rr1-p2-x96-acc2.c | 92 const __m256 vsB = _mm256_castsi256_ps(_mm256_slli_epi32(_mm256_castps_si256(vnB), 23)); in xnn_f16_raddstoreexpminusmax_ukernel__avx2_rr1_p2_x96_acc2() local 144 vtB = _mm256_mul_ps(vtB, vsB); in xnn_f16_raddstoreexpminusmax_ukernel__avx2_rr1_p2_x96_acc2() 157 __m256 vfB = _mm256_fmadd_ps(vtB, vpB, vsB); in xnn_f16_raddstoreexpminusmax_ukernel__avx2_rr1_p2_x96_acc2()
|
H A D | neonfp16arith-rr2-p2-x96-acc6.c | 96 const float16x8_t vsB = vreinterpretq_f16_s16(vshlq_n_s16(vreinterpretq_s16_f16(vnB), 10)); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x96_acc6() local 161 vtB = vmulq_f16(vtB, vsB); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x96_acc6() 185 float16x8_t vfB = vfmaq_f16(vsB, vpB, vtB); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x96_acc6()
|