Johan Mabille
Johan Mabille
The lgamma test with negative inputs have been disabled for architecture supporting AVX but not AVX2, because they lead to small differences for a few inputs: ```bash idx = 1533...
The azure-pipeline build pins cmake 3.14 to avoid failure when buiding with clang-cl. It seems that some options are not set correctly by cmake, this should be investigated so that...
This issue tracks the progress on implementing shuffle functions in xsimd. The aproach is to implement different operations in dedicated functions (similar to what is done in [libscimdpp](https://p12tic.github.io/libsimdpp/v2.2-dev/libsimdpp/w/shuffle.html). If you...
The build on linux aarch64 is currently disabled in the CI (because it fails) since #409 has been merged. It should be fixed and enabled again.
A lot of mathematical functions from the STL convert their argument for integer to double and then rely on the overload for double ([remainder](https://en.cppreference.com/w/cpp/numeric/math/remainder), [nearbyint](https://en.cppreference.com/w/cpp/numeric/math/nearbyint), etc...). `xsimd` should follow the...
The current CI configuration allow to test AVX512BW instruction set. It would be nice to add a configuration for AVX512VL or AVX512F to improve test coverage. Once it is done,...
As reported in #289, neon instruction set is not detected when building on armv7hl. A workaround is `#define XSIMD_FORCE_ARM_INSTR_SET = 70000000` before including any header of xsimd.
`pow` of `complex` has some accuracy issues with AVX512.
Some bitwise casts functions (from and to unsigned integers) are missing.