otherwise we can delete stuff from the next outer loop, which in 2D would be the one OpenMP-parallelized

Fixes #25

Vectorization improvements

See merge request !228

Fix installation from git

Closes #14

See merge request !227

Fix field size

Closes #32, #31, #3, and #7

See merge request !226

Fix benchmark generation

See merge request !223

ARM Neon: Fix makeVec and add Philox

See merge request !222

Ship c-file with pypi

See merge request !221

Improve ARM64 support

See merge request !220

Make RNG vectorization more robust

See merge request !219

Michael Kuron authored 4 years ago and Markus Holzer committed 4 years ago

The vectorized RNG can switch between scalar and vector types due to loop cutting.

Update setup.py

See merge request !218

Add kerncraft templates

See merge request !217

some fixes for lbmpy vectorization

See merge request !216

fix aligned_alloc on windows

See merge request !215

default to C++17 on macOS and Windows

Closes #24

See merge request !214

vectorization: improve treatment of unary minus

See merge request !212

Fix clear cache on startup

See merge request !213

Due two weird recursive dependencies `clear_cache` will not be
available at this point. In-lining this function will make
the config key `cache.clear_cache_on_start: true` work again.

Fix logic error in RNG test

See merge request !211

RNG SIMD

Closes #12

See merge request !48