Rust native benchmark comparing SIMD (aarch64/neon
intrinsics) vs scalar implementations of element-wise float32 vector
addition.
Vector length: 1024000, Loop: 200
| Implementation | Avg time / run | Speedup vs Scalar | Sample result (first 4) |
|---|---|---|---|
| Scalar | 679.049 µs | 1.00x | [0.0, 3.0, 6.0, 9.0] |
| SIMD | 117.784 µs | 5.77x | [0.0, 3.0, 6.0, 9.0] |
Run cargo run --release in this directory to regenerate this
report on your own machine.