Simsimd

[3.6.4](https://github.com/ashvardanian/SimSIMD/compare/v3.6.3...v3.6.4) (2024-01-08)


Docs

* TypeScript declaration file (53) ([5f6a688](https://github.com/ashvardanian/SimSIMD/commit/5f6a688bfff2b75b73b669ee09cbe1635fc9813e)), closes [#53](https://github.com/ashvardanian/SimSIMD/issues/53)

Make

* Prebuild JavaScript bindings (56) ([1bd9001](https://github.com/ashvardanian/SimSIMD/commit/1bd90012daba4b08f178ded5c0a1243099e79901)), closes [#56](https://github.com/ashvardanian/SimSIMD/issues/56)

[3.6.3](https://github.com/ashvardanian/SimSIMD/compare/v3.6.2...v3.6.3) (2024-01-06)


Make

* Revert test location ([82c4dcb](https://github.com/ashvardanian/SimSIMD/commit/82c4dcbe027da65d040241614099cf739ad9b24c))

[3.6.2](https://github.com/ashvardanian/SimSIMD/compare/v3.6.1...v3.6.2) (2024-01-06)


Docs

* Describe usage in C ([555ce0c](https://github.com/ashvardanian/SimSIMD/commit/555ce0c1a67465aaa6b9f237b48952c9697587be))
* JS installation, grammar and counters (50) ([ba0e233](https://github.com/ashvardanian/SimSIMD/commit/ba0e23398ee4fc229b6ee1a67c24e440d418b4bd)), closes [#50](https://github.com/ashvardanian/SimSIMD/issues/50)
* typo in README.md (49) ([330c039](https://github.com/ashvardanian/SimSIMD/commit/330c0397968119af183d7a8476ac9be58ad1d49c)), closes [#49](https://github.com/ashvardanian/SimSIMD/issues/49)

Fix

* Type errors in JS benchmarks (51) ([57ced28](https://github.com/ashvardanian/SimSIMD/commit/57ced28ed6519655e6aee96daeb12b2fa65beadb)), closes [#51](https://github.com/ashvardanian/SimSIMD/issues/51)

[3.6.1](https://github.com/ashvardanian/SimSIMD/compare/v3.6.0...v3.6.1) (2023-12-19)


Docs

* New header ([248082d](https://github.com/ashvardanian/SimSIMD/commit/248082d8277444968e75982296e6405fc90095a6))

Fix

* SEGFAULT creating NumPy Array ([6cccca9](https://github.com/ashvardanian/SimSIMD/commit/6cccca973ea8792caebbd62383baee1208fd9bd4))

Improve

* Cleaner accumulator init ([af4a818](https://github.com/ashvardanian/SimSIMD/commit/af4a818cacff8f4cb3033f854e659026809cb762))
* Logging exceptions ([c5b4c0e](https://github.com/ashvardanian/SimSIMD/commit/c5b4c0ed7428ae827aacb72a9448c0a6f43dbb8d))

Make

* Update Python library `__version__` ([14559ed](https://github.com/ashvardanian/SimSIMD/commit/14559ed6d4734b8443237cef3a82fd441ab249de))

Test

* Increase error tolerance ([d216035](https://github.com/ashvardanian/SimSIMD/commit/d2160354a035bd7e4bdd9870a3a1620ec9d9e255))

As was discussed in the [SciPy integration thread](https://github.com/scipy/scipy/issues/19454), Python libraries use double-precision floating-point numbers by default. So in this release I've extended the spatial distance functions - `cosine`, `sqeuclidean`, `inner` with support for `double` arguments with specialized implementations on AVX-512-capable x86 CPUs and SVE-capable Arm CPUs.

Benchmarking SimSIMD vs. SciPy on Intel Sapphire Rapids CPU

- Vector dimensions: 1536
- Vectors count: 1000
- Hardware capabilities: `serial`, `x86_avx2`, `x86_avx512`, `x86_avx2fp16`, `x86_avx512fp16`, `x86_avx512vpopcntdq`, `x86_avx512vnni`
- NumPy BLAS dependency: `openblas64`
- NumPy LAPACK dependency: `dep140640983012528`

Between 2 Vectors, Batch Size: 1

| Datatype | Method | Ops/s | SimSIMD Ops/s | SimSIMD Improvement |
| :------- | :-------------------- | -------------------: | -------------------: | ------------------: |
| `f64` | `scipy.cosine` | 63,612 | 572,605 | 9.00 x |
| `f64` | `scipy.sqeuclidean` | 238,547 | 915,596 | 3.84 x |
| `f64` | `numpy.inner` | 449,499 | 986,522 | 2.19 x |

Between 2 Vectors, Batch Size: 1,000

| Datatype | Method | Ops/s | SimSIMD Ops/s | SimSIMD Improvement |
| :------- | :-------------------- | -------------------: | -------------------: | ------------------: |
| `f64` | `scipy.cosine` | 68,962 | 1,457,172 | 21.13 x |
| `f64` | `scipy.sqeuclidean` | 247,727 | 1,535,547 | 6.20 x |
| `f64` | `numpy.inner` | 463,509 | 1,512,004 | 3.26 x |


Benchmarking SimSIMD vs. SciPy on AWS Graviton 3

- Vector dimensions: 1536
- Vectors count: 1000
- Hardware capabilities: `serial`, `arm_neon`, `arm_sve`
- NumPy BLAS dependency: `openblas64`
- NumPy LAPACK dependency: `openblas64`

Between 2 Vectors, Batch Size: 1

| Datatype | Method | Ops/s | SimSIMD Ops/s | SimSIMD Improvement |
| :------- | :-------------------- | -------------------: | -------------------: | ------------------: |
| `f64` | `scipy.cosine` | 40,729 | 725,382 | 17.81 x |
| `f64` | `scipy.sqeuclidean` | 160,812 | 728,114 | 4.53 x |
| `f64` | `numpy.inner` | 473,443 | 767,374 | 1.62 x |
| `f64` | `scipy.jensenshannon` | 15,684 | 38,528 | 2.46 x |
| `f64` | `scipy.kl_div` | 49,983 | 61,811 | 1.24 x |

Between 2 Vectors, Batch Size: 1,000

| Datatype | Method | Ops/s | SimSIMD Ops/s | SimSIMD Improvement |
| :------- | :-------------------- | -------------------: | -------------------: | ------------------: |
| `f64` | `scipy.cosine` | 41,130 | 1,460,850 | 35.52 x |
| `f64` | `scipy.sqeuclidean` | 162,147 | 1,486,255 | 9.17 x |
| `f64` | `numpy.inner` | 473,856 | 1,580,136 | 3.33 x |

[3.5.5](https://github.com/ashvardanian/SimSIMD/compare/v3.5.4...v3.5.5) (2023-11-11)


Docs

* Reorder sections ([ac2ee97](https://github.com/ashvardanian/SimSIMD/commit/ac2ee97ea63134198b7c5909289338773c07666a))

Improve

* Detecting compile-time capabilities ([e02a24f](https://github.com/ashvardanian/SimSIMD/commit/e02a24f906ec67b42f66819b0160f4e6f90f7e6e))