I’m trying to build tensorflow lite from source for my tx2 board. I following the instructions on tensorflow-lite page(https://www.tensorflow.org/lite/guide/build_arm64), but the error occured.
error: cannot convert ‘const int8x8_t {aka const __vector(8) signed char}’ to ‘uint32x2_t {aka __vector(2) unsigned int}’ for argument ‘2’ to ‘void vst1_lane_u32(uint32_t*, uint32x2_t, int)’
After solving the problem as suggested by this answer(https://stackoverflow.com/questions/56055359/tensorflow-lite-arm64-error-cannot-convert-const-int8x8-t), new problem occured.
"add x29, %[output_block_data], x10\n"
x29 cannot be used in asm here
It seems to be invalid register usage in asm.
Does anyone succeeded in building tensorflow-lite from source?