NUM_FPR
uint64_t fpr_in[NUM_FPR * 2];
uint64_t fpr_out[NUM_FPR * 2];
uint64_t fpr_zero[NUM_FPR * 2];
__uint128_t v_in[NUM_FPR];
__uint128_t v_expected[NUM_FPR];
__uint128_t v_out[NUM_FPR];