One laptop, matched: FP16/FP32 26.48 → 52.97 TF/s and TF32 13.21 → 26.42 TF/s, exactly 2.00× on both.
Measured throughput
20 SMs · driver 595.99.02
Mode
Stock
Unlocked
Gain
f16/f32
26.48 TF/s
52.97 TF/s
2.00×
tf32
13.21 TF/s
26.42 TF/s
2.00×
Same card, same binary, stock versus unlocked. Modes that did not change are reported too — an unlock that does not move a path is a fact worth having.