Implements trunc using vcvtq vs vrndq

neon-trunc.cpp

#include <arm_neon.h>
#include <iostream>

int main(void) {
  const float inp[4] = { 3.1, 2.7, -2.9, -1.0000000200408773e+20 };
  std::cout << "inp= " << inp[0] << " " << inp[1] << " " << inp[2] << " " << inp[3] << std::endl;
  float32x4_t f = vld1q_f32(inp);
  float32x4_t old_trunc = vcvtq_f32_s32(vcvtq_s32_f32(f));
  float32x4_t new_trunc = vrndq_f32(f);
  float out[4];
  vst1q_f32(out, old_trunc);
  std::cout << "old_trunc= " << out[0] << " " << out[1] << " " << out[2] << " " << out[3] << std::endl;
  vst1q_f32(out, new_trunc);
  std::cout << "new_trunc= " << out[0] << " " << out[1] << " " << out[2] << " " << out[3] << std::endl;
  return 0;
}