Home
last modified time | relevance | path

Searched refs:prod_16x8 (Results 1 – 1 of 1) sorted by relevance

/external/tensorflow/tensorflow/lite/kernels/internal/optimized/
Dneon_tensor_utils.cc806 int16x8_t prod_16x8 = in NeonMatrixBatchVectorMultiplyImpl() local
814 prod_16x8 = in NeonMatrixBatchVectorMultiplyImpl()
815 vmlal_s8(prod_16x8, vget_high_s8(s1_8x16), vget_high_s8(s2_8x16)); in NeonMatrixBatchVectorMultiplyImpl()
817 dotprod_32x4 = vpadalq_s16(dotprod_32x4, prod_16x8); in NeonMatrixBatchVectorMultiplyImpl()
829 const int16x8_t prod_16x8 = vmull_s8(s1_8x8, s2_8x8); in NeonMatrixBatchVectorMultiplyImpl() local
830 dotprod_32x4 = vpadalq_s16(dotprod_32x4, prod_16x8); in NeonMatrixBatchVectorMultiplyImpl()
1130 int16x8_t prod_16x8 = in NeonMatrixBatchVectorMultiplyAccumulate() local
1138 prod_16x8 = in NeonMatrixBatchVectorMultiplyAccumulate()
1139 vmlal_s8(prod_16x8, vget_high_s8(s1_8x16), vget_high_s8(s2_8x16)); in NeonMatrixBatchVectorMultiplyAccumulate()
1141 dotprod_32x4 = vpadalq_s16(dotprod_32x4, prod_16x8); in NeonMatrixBatchVectorMultiplyAccumulate()
[all …]