Re: Share Your Experience with 3DNow, SSE, SSE2 etc.

I forgot prefetchN = 8.

On Aug 4, 11:42 am, aruzinsky <aruzin...@xxxxxxxxxxxxxxxxxxxx> wrote:
On Aug 4, 12:15 am, Hendrik van der Heijden <h...@xxxxxx> wrote:

aruzinsky schrieb:

Thank you for your input.  I would feel more confident if you tested
my code, though.

Then post a pointer to your code, sources preferred.

Hendrik

Simple enough to cut and paste here, but this is an untested excerpt
that assumes the arrays are a multiple of 16 floats aligned on 16 byte
boundaries.

void equalSSE2(int mn, float *aax, float *bbx, int prefetchN)
{
        int NN = 16, MN = mn/NN, MN4 = NN*MN, N = mn-MN4;
        __asm
        {
                push eax
                push ecx
                push edx
                push esi
                mov ecx, prefetchN
                imul ecx, 64
                mov     eax, aax
                mov     edx, bbx
                mov     esi, MN
                test esi, esi
                jle $L1
                align 16;
$L2:
                movaps  xmm0, [edx]
                movaps  xmm1, [edx+16]
                movaps  xmm2, [edx+32]
                movaps  xmm3, [edx+48]

                PREFETCHNTA [edx+ecx]

                movntpd  [eax], xmm0
                movntpd  [eax+16], xmm1
                movntpd  [eax+32], xmm2
                movntpd  [eax+48], xmm3

                add     edx, 64
                add     eax, 64

                dec esi
                jnz     $L2
$L1:
                pop     esi
                pop     edx
                pop     ecx
                pop     eax
        }
        //code for array elements MN4 to mn-1 would go here



}- Hide quoted text -

- Show quoted text -

.