In this test the same 128 bit key consisting of all 0s was used for data encryption.

Input data consisted of all combinations of 4 bit 1 and 124 bit 0 out of 128 bit of the data block, therefore giving 10,668,000 combinations or 170,688,000 byte of input data for encryption and the following noise purity analysis.

Values separated by / refer to the results of encryption performed WITH and WITHOUT pre- and post- data whitening such as bit permutation or XOR or ADD or any other more complex operation if it's present. If algorithm does not include any pre or post- whitening, only one result is displayed and used.

Processor cycles per round (CPR) refer to the fastest known implementations for Intel® Pentium® II processor. Extra possible minor speed increases due to instruction pairing between encryption rounds themselves and between data whitening and the first and the last rounds of encryption were not taken into consideration.
 

* Safer+: Starting with 6 and up to the maximum 8 rounds of encryption, returned test results look very suspicious, possible minor flaw in the cipher.

* Crypton1: Only key schedule generation was modified in Crypton1, so it performed identically to Crypton in this test and therefore it's not included in this table.

We expected Bruce Schneier's Twofish to look good, but we didn't expect it to show such an outstanding performance and such a remarkable stability of the results… We also expected RC6 to perform better than it did, but… So far Twofish indubitably leads the list of the world's best 128 bit block ciphers.

Details on the results from the table above are under construction. Please also note that just passing these tests alone is not a proof of the cipher's security, but failing them by a single bit is what must be taken seriously.

We do not and never will provide a theoretical basis for or an explanation of the results of our tests. This is not a cryptoanalytic work. All tests are meant purely for demonstration of the practical difference in behavior and performance of different encryption algorithms hopefully finding and bringing to attention possible flaws in them.

Well organized AES Algorithms Efficiency page by Brian Gladman (e-mail ), contains next to perfect C sources of all AES algorithms, updated regularly.

One of the best AES Algorithm Performance Comparison documents by the TWOFISH team, not updated.

The one and only AES Block Cypher Lounge by Lars Knudsen (e-mail) contains names of the papers and links to the best known attacks on AES algorithms, updated often.

The largest Cryptography People List by David Wagner (e-mail), some links are outdated though.

A nice AES Ciphers Speed site maintained by Helger Lipmaa (e-mail), updated regularly.

And a very informative Candidate AES Analysis and Reviews site maintained by Franois Kouene (e-mail), UCL Crypto Group, updated often.

Revision History: