Timeline for Gimli, make it even shorter?
Current License: CC BY-SA 3.0
9 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Sep 9, 2017 at 13:57 | comment | added | orlp | @Biv And this technique, to use a bitshift as a small lookup table is actually a side-channel resistant form of doing lookup tables, so potentially useful in implementing crypto primitives :) Obviously in the way it's used here the access pattern is always consistent and public information, so no reason to worry about cache effects, but the technique is interesting regardless. | |
| Sep 9, 2017 at 12:04 | comment | added | Biv | @orlp I really like your column traversal too. :) | |
| Sep 9, 2017 at 11:56 | comment | added | orlp |
@PeterTaylor Even though my code is similar, I'm not really competing against OP's code. I'm working under the rules of PPCG, where it must work with at least an implementation on a platform, and it does with gcc on a 32-bit or 64-bit Intel CPU (and probably many more).
|
|
| Sep 9, 2017 at 9:20 | comment | added | Peter Taylor |
This is still definitely an improvement, but it's somewhat cheating to use unsigned instead of uint32_t (and OP's code was somewhat cheating to use long) because the idea behind the cipher is that it's highly portable. (In fact, fundamentally this saves a mere 8 bytes).
|
|
| Sep 9, 2017 at 1:43 | comment | added | hyperneutrino♦ | Ah ok :P makes sense :P :P | |
| Sep 9, 2017 at 1:42 | comment | added | orlp | @HyperNeutrino Gained, making me a loser :) | |
| Sep 9, 2017 at 1:41 | history | edited | orlp | CC BY-SA 3.0 |
added 2 characters in body
|
| Sep 9, 2017 at 1:41 | comment | added | hyperneutrino♦ | lost or gained? | |
| Sep 9, 2017 at 1:37 | history | answered | orlp | CC BY-SA 3.0 |