Closed philipc closed 2 years ago
Also some other small fixes.
Performance difference compared to the C++ implementation (but I'm not sure how relevant the data used for the benchmark is):
align_i32x8 time: [32.800 ms 32.834 ms 32.870 ms] change: [-15.726% -15.382% -15.037%] (p = 0.00 < 0.05) Performance has improved. align_i32x16 time: [34.820 ms 34.859 ms 34.898 ms] change: [-9.8908% -9.2732% -8.8869%] (p = 0.00 < 0.05) Performance has improved.
Also some other small fixes.
Performance difference compared to the C++ implementation (but I'm not sure how relevant the data used for the benchmark is):