weird load latency on Apple M2

[ocxtal]

does it have a speculative load mechanism?

on 24e76639beddd3b3f3421da9642dd7a5d4d74357

# AArch64 latency / throughput benchmark report
# Generated by https://github.com/ocxtal/insn_bench_aarch64 (commit: unknown).
# CPU frequency estimation
# measuring CPU frequency, assuming latency of 64bit addition is 1 cycle(s):
# - 3479.15 MHz
# - 3480.94 MHz
# - 3479.35 MHz
# Scalar load
# instruction   latency throughput
ldr (imm; ofs = 0)      1.17    2.83
ldr (imm; ofs = 16)     1.17    2.83
ldr (imm; pre, ofs = 0) 0.83    2.83
ldr (imm; pre, ofs = 16)        0.81    -
ldr (imm; post, ofs = 0)        0.85    2.83
ldr (imm; post, ofs = 16)       0.78    -
ldr (ptr fwd.; imm; pre, ofs = 16)      1.01    -
ldr (ptr fwd.; imm; post, ofs = 16)     1.01    -