|Kryptos Plato: Searching for Hill cipher solutions|
|Microsoft Windows running on an AMD x86_64 or Intel EM64T CPU||0.05 (vbox64_mt)||7 Jan 2021, 23:11:37 UTC||334 GigaFLOPS|
|Linux running on an AMD x86_64 or Intel EM64T CPU||0.05 (vbox64_mt)||7 Jan 2021, 23:11:37 UTC||287 GigaFLOPS|
|Intel 64-bit Mac OS 10.5 or later||0.05 (vbox64_mt)||7 Jan 2021, 23:11:36 UTC||14 GigaFLOPS|
|Kryptos Aristotle: Faster and more explorative search for Hill cipher solutions (beta test)|
|Linux running on an AMD x86_64 or Intel EM64T CPU||0.80 (beta test)||11 Apr 2021, 16:36:36 UTC||8 GigaFLOPS|
Total average computing: 642 GigaFLOPS
The Kryptos puzzle is unique in that it has resisted the test of time and has remained unsolved for more than thirty years! Furthermore, the creator has provided only one "copy" of each of the four parts that comprise the puzzle.
This naturally suggests that completely new and out-of-the box methods and approaches need to be devised and tested. Consequently, the calibration of any such method becomes a challenge, due to small sample size, by which any individual puzzle is characterized.
We may occasionally use solved or unsolved puzzles, other than Kryptos, for the purpose of validation and/or calibration of some of our algorithms.
For the purpose of full disclosure and transparency, the full list of such puzzles will be maintained on the current page.
Currently, we don't use any puzzles, besides Kryptos, for validation and/or calibration of our algorithms.
Our Kryptos Plato application is our first BOINC application, that has been deployed to and employed in our production environment.
We use genetic algorithms to search for Hill cipher solutions to K4.
Technically speaking, we are looking for invertible matrices which - when employed as Hill cipher keys - result in matches to the cryptographic cribs (which have been provided to the Kryptos community, in the form of K4 clues, by James Sanborn himself), and also provide some meaningful decryptions of the remainder of K4.