Przemysław Karpiński
Przemysław Karpiński
Current interface definition states, that the gather/scatter operations are using vectors of the same precision as the scalar type used. This is OK for 32b and 64b types, but might...
All reduction operations should also have a version with a scalar initializer. These operations are necessary to improve the programmability. For example to perform a hoirzontal reduction with a scalar...
When calculations are performed within an 'if-else' statement for scalar operations, only one part of the statement has to be executed ('if' block or 'else' block). When using masking, both...
Originally reported by: **Guilherme Amadio (Bitbucket: [amadio](https://bitbucket.org/amadio), GitHub: [amadio](https://github.com/amadio))** ---------------------------------------- It would be nice to have at least a simple (and thread-safe, if possible) random number generator in UME::SIMD. Even...
Originally reported by: **edanor (Bitbucket: [edanor](https://bitbucket.org/edanor), GitHub: [edanor](https://github.com/edanor))** ---------------------------------------- Compiling current test suite takes insane amounts of time. For development purposes only a minimal set of tests is required, usually...
Originally reported by: **edanor (Bitbucket: [edanor](https://bitbucket.org/edanor), GitHub: [edanor](https://github.com/edanor))** ---------------------------------------- Some floating point operations rely on approximation. The problem is to test such functions against standard version with automatically generated data...
Originally reported by: **edanor (Bitbucket: [edanor](https://bitbucket.org/edanor), GitHub: [edanor](https://github.com/edanor))** ---------------------------------------- As for other interfaces, masks should also implement MFI functions with masking operand. An example where it causes problems comes from...
Originally reported by: **edanor (Bitbucket: [edanor](https://bitbucket.org/edanor), GitHub: [edanor](https://github.com/edanor))** ---------------------------------------- Implementation of SIMD2 types for 64b scalars is not using intrinsic types. This should be fixed. ---------------------------------------- - Bitbucket: https://bitbucket.org/edanor/umesimd/issue/29
Originally reported by: **edanor (Bitbucket: [edanor](https://bitbucket.org/edanor), GitHub: [edanor](https://github.com/edanor))** ---------------------------------------- Except operators using scalar LHS operand, all overloaded operators are implemented as members in specialized classes. Because of that it is...
Originally reported by: **edanor (Bitbucket: [edanor](https://bitbucket.org/edanor), GitHub: [edanor](https://github.com/edanor))** ---------------------------------------- There are multiple unit test fails even when running with full scalar emulation. This can be as well caused by incorrect...