Efficient Parallel Implementation of Matrix Multiplication for Lattice-Based Cryptography on Modern ARM Processor

Table 3

Matrix multiplication performance (unit: ms).

N	M	L	C version [3]	Proposed (NEON)
			(Auto-Vectorization)
536	1024	256	148.8991	93.91285
663	1024	256	171.0976	159.2069
816	1024	384	334.7499	224.5633
952	1024	384	391.7564	361.7326