npm run bench.
All benchmarks measure computation time from JS and Python, respectively. To learn more, check out benchmark methodology.
Benchmark Summary
- Average speedup: 1.11x vs NumPy
- Best case: 2422.82x
- Worst case: 0.09x
- Total benchmarks: 7159
- Machine: Apple M4 Max (16 cores, 128 GB, arm64)
- numpy-ts version: 1.4.0
Performance by Category
| Category | Avg Speedup | Count | Faster | Slower |
|---|---|---|---|---|
| creation | 1.76x | 639 | 476 | 163 |
| arithmetic | 1.05x | 885 | 400 | 485 |
| math | 0.62x | 375 | 82 | 293 |
| trig | 0.61x | 648 | 151 | 497 |
| gradient | 4.48x | 66 | 66 | 0 |
| linalg | 1.50x | 807 | 523 | 284 |
| reductions | 0.96x | 1239 | 613 | 626 |
| manipulation | 1.16x | 693 | 329 | 364 |
| io | 2.71x | 187 | 162 | 25 |
| indexing | 0.66x | 345 | 127 | 218 |
| bitwise | 0.65x | 30 | 7 | 23 |
| sorting | 0.82x | 225 | 58 | 167 |
| logic | 1.69x | 426 | 242 | 184 |
| statistics | 1.54x | 78 | 48 | 30 |
| sets | 3.33x | 99 | 77 | 22 |
| random | 0.91x | 138 | 49 | 89 |
| polynomials | 2.06x | 81 | 63 | 18 |
| fft | 0.91x | 198 | 88 | 110 |
Performance by DType
| DType | Avg Speedup | Median Speedup | Count |
|---|---|---|---|
| float64 | 1.02x | 0.95x | 863 |
| float32 | 1.05x | 0.98x | 713 |
| float16 | 1.36x | 1.36x | 632 |
| int64 | 1.03x | 0.94x | 587 |
| uint64 | 1.00x | 0.93x | 563 |
| int32 | 1.16x | 1.05x | 611 |
| uint32 | 1.19x | 1.07x | 566 |
| int16 | 1.11x | 1.01x | 554 |
| uint16 | 1.12x | 1.02x | 551 |
| int8 | 1.27x | 1.02x | 554 |
| uint8 | 1.27x | 1.05x | 557 |
| complex128 | 0.85x | 0.80x | 204 |
| complex64 | 0.80x | 0.62x | 204 |