On Wed, May 26, 2010 at 12:11 AM, Thomas DuBuisson
* For Crypto-Algs pure Haskell code may be preferred but should not be used when costing over ~20% performance degradation when compared to other available implementations.
20% is quite a large performance hit, but within the bounds of reason. Something to keep in mind, however: The newer Intel processors have some hardware support for AES. As I should be acquiring one of those in the near future, I intend to rewrite the AES package to take advantage; of course, preferably by exploiting existing working code. I don't believe any software implementation will get within 20% of the resulting performance. In other words, you will need to do the same. Though I wouldn't mind if you use mine.. -- Svein Ove Aas