Search code examples
tensorflowarm64tensorflow-lite

Error building Tensorflow Lite on AARCH64

Hi I am new to Tensorflow and I am trying to build Tensorflow Lite for a Pine64 A64+ board. I followed the instructions on tensorflow lite page and got a lot of errors such as this one:

./tensorflow/lite/kernels/internal/optimized/depthwiseconv_uint8_3x3_filter.h:55:78: error: cannot convert ‘int32x4_t {aka __vector(4) int}’ to ‘int8x16_t {aka __vector(16) signed char}’ in assignment
 #define vld1q_dup_s8x4(src) vld1q_dup_s32(reinterpret_cast<const int32*>(src))
                                                                              ^
./tensorflow/lite/kernels/internal/optimized/depthwiseconv_uint8_3x3_filter.h:5994:30: note: in expansion of macro ‘vld1q_dup_s8x4’
           input_bank_c_reg = vld1q_dup_s8x4(

What am I doing wrong?

Solution

  • I think this should be fixed already, can you sync to head and try again?