Performance

• Easy to implement algorithm

  • Uses only common, 32-bit ALU operations
  • Can eliminate 128 bit rotates by compressing quad ops into 4 sequential stages

• Parallelization

  • First 6 quad-rounds are not parallelizable
  • Last 6 have better parallel qualities

• Operations are simple but key schedule takes MANY operations

Previous slide

Next slide

Back to first slide

View graphic version