cayoenrique
Senior Member
- Messages
- 478
@C0der
Listen I do not forget you are out there. And you do not need the CSA stuff. BC only or Colibri csa.cu kernel can be done. Like I said ,speed should come up the almost the same. OpenCl is like Cuda. Now changing the way to compute Keyschedule or the way we read form SBOX & Permute. That makes difference in speed. Like they say trial and error, Instead "trial and speed test". But that is more an advance math project. Problem is that if I try that I will have even less time for the learning stuff.
Dale once posted that an encrypted pack transponder like in Abertis will produce a very special situation. Where instead of the normal require 1 stream + 1 block cypher per TS, we can instead find solution with 2 block cypher but of two different ts. No stream. Or 2 stream no block. Let me see if I can remeber, try to refresh those formulas and I may try to explain the case later.
Listen guys. I have not much time. But I give my best to speed up. I will post something tomorrow.
Listen I do not forget you are out there. And you do not need the CSA stuff. BC only or Colibri csa.cu kernel can be done. Like I said ,speed should come up the almost the same. OpenCl is like Cuda. Now changing the way to compute Keyschedule or the way we read form SBOX & Permute. That makes difference in speed. Like they say trial and error, Instead "trial and speed test". But that is more an advance math project. Problem is that if I try that I will have even less time for the learning stuff.
Dale once posted that an encrypted pack transponder like in Abertis will produce a very special situation. Where instead of the normal require 1 stream + 1 block cypher per TS, we can instead find solution with 2 block cypher but of two different ts. No stream. Or 2 stream no block. Let me see if I can remeber, try to refresh those formulas and I may try to explain the case later.
Listen guys. I have not much time. But I give my best to speed up. I will post something tomorrow.