[FFmpeg-devel] libavcodec/exr : add SIMD for reorder pixels (SSE and AVX2) v3 (WIP)

[Henrik Gramner]

On Sun, Sep 10, 2017 at 5:17 PM, Martin Vignali
<martin.vignali at gmail.com> wrote:
> +    void (*reorder_pixels)(uint8_t *src, uint8_t *dst, int size);

size should be ptrdiff_t instead of int since it's used as a 64-bit
operand in the asm on x86-64 and the upper 32 bits are undefined
otherwise.

> +++ b/libavcodec/x86/exrdsp.asm

Couldn't this be simplified to something like:

%macro REORDER_PIXELS 0
cglobal reorder_pixels, 3,4,3, src1, dst, size, src2
    lea     src2q, [src1q+sizeq]
    add      dstq, sizeq
    shr     sizeq, 1
    add     src1q, sizeq
    neg     sizeq
.loop:
    mova       m0, [src1q+sizeq]
    movu       m1, [src2q+sizeq]
    SBUTTERFLY bw, 0, 1, 2
    mova [dstq+2*sizeq   ], xm0
    mova [dstq+2*sizeq+16], xm1
%if mmsize == 32
    vperm2i128 m0, m0, m1, 0x31
    mova [dstq+2*sizeq+32], m0
%endif
    add     sizeq, mmsize
    jl .loop
    RET
%endmacro

(completely untested)