Use VPERMB _mm512_permutexvar_epi8 for DES and/or Lotus

This is another recent instruction introduced in Intel Cannon Lake (9th gen, but not all) and above (consistently since Ice Lake, 10th gen) through the VBMI extension on top of AVX-512. It appears to perform a mapping that's just right for one DES S-box, 64 times in parallel. So a non-bitslice DES implementation using this instruction may outperform bitslice.

https://www.intel.com/content/www/us/en/docs/intrinsics-guide/index.html#text=_mm512_permutexvar_epi8&ig_expand=5071

__m512i _mm512_permutexvar_epi8 (__m512i idx, __m512i a)
#include <immintrin.h>
Instruction: vpermb zmm, zmm, zmm
CPUID Flags: AVX512_VBMI
Description
Shuffle 8-bit integers in a across lanes using the corresponding index in idx, and store the results in dst.
Operation
FOR j := 0 to 63
	i := j*8
	id := idx[i+5:i]*8
	dst[i+7:i] := a[id+7:id]
ENDFOR
dst[MAX:512] := 0

Start by locating the DES and Lotus implementations and their existing SIMD or bitslice paths. Read the linked Intel documentation for _mm512_permutexvar_epi8 and check the AVX512_VBMI requirements; the work is done when the proposed approach is evaluated for those cipher paths, including whether it improves performance.