Design of efficient of block ciphers needs consideration of many factors such as register size. i found that some ciphers can be SIMD implemented such as TWINE , SM4 , chacha20 ,SPIX ,and others.
What components in block (Feistel and SPN) and stream ciphers are tricky to be implemented in vectorizations?
Does the Sbox size (4 bit or 8 bit) or the F-function structure (such as simon-like Sbox in SPIX or SP/SPS such as in Clefia or piccolo) matter ?
Does the branch number in Feistel Type (2,4,6,8 ..) have related in number of blocks to be packed for vectorization?