Keeping alignment when slicing is easy since it matches the size
of the xmm registers: one has to partition the array by blocks of
2 doubles, 4 floats, etc. For AVX, the ideal alignment is on
32-byte boundaries but the really bad performance hit happens
only when an unaligned access crosses a cacheline boundary. With
SSE2, this concerns every single access.
- Alignment of dynamic arrays Luc Bourhis via Digitalmars-d
- Re: Alignment of dynamic arra... bearophile via Digitalmars-d
- Re: Alignment of dynamic ... Luc Bourhis via Digitalmars-d
- Re: Alignment of dynamic ... Robert burner Schadek via Digitalmars-d
- Re: Alignment of dyna... bearophile via Digitalmars-d
- Re: Alignment of ... Robert burner Schadek via Digitalmars-d
- Re: Alignmen... Steven Schveighoffer via Digitalmars-d
- Re: Alig... Steven Schveighoffer via Digitalmars-d
- Re: ... Luc Bourhis via Digitalmars-d
- Re: Alignment of dyna... Steven Schveighoffer via Digitalmars-d
- Re: Alignment of ... via Digitalmars-d
- Re: Alignment of ... Robert burner Schadek via Digitalmars-d