Other Considerations

• Comparing benchmark results for processors with different data word sizes can be misleading
  • e.g., 24-bit data word provides better accuracy than 16-bit data word

• Comparing fixed-point results to floating-point results can be misleading
  • Floating-point provides better precision...
  • ... but AD and DA converters use fixed-point
  • Meeting bit-exact standards may require extra work on floating-point processors

Previous slide

Next slide

Back to first slide

View graphic version