> +INIT_XMM sse2 > +cglobal conv_s16p_to_flt_6ch, 2,8,8, dst, src, src1, src2, src3, src4, src5, > len > +%if ARCH_X86_64 > + mov lend, r2d > +%else > + %define lend dword r2m > +%endif > + mov src1q, [srcq+1*gprsize] > + mov src2q, [srcq+2*gprsize] > + mov src3q, [srcq+3*gprsize] > + mov src4q, [srcq+4*gprsize] > + mov src5q, [srcq+5*gprsize] > + mov srcq, [srcq] > + sub src1q, srcq > + sub src2q, srcq > + sub src3q, srcq > + sub src4q, srcq > + sub src5q, srcq > + mova m7, [pf_s32_inv_scale] > + ALIGN 16 > +.loop: > + movlpd m0, [srcq ] > + movlpd m1, [srcq+src1q] > + movlpd m2, [srcq+src2q] > + movhpd m0, [srcq+src3q] ; m0 = 0, 6, 12, 18, 3, 9, 15, 21 > + movhpd m1, [srcq+src4q] ; m1 = 1, 7, 13, 19, 4, 10, 16, 22 > + movhpd m2, [srcq+src5q] ; m2 = 2, 8, 14, 20, 5, 11, 17, 23 > + ; unpack words:
Like before, use movq and movhps. > + movhlps m3, m0 ; m3 = 3, 9, 15, 21, x, x, x, x movhlps has a dependency on the top half of m3, which is uninitialized. Maybe do this? + movq m3, [srcq+src3q] + movhps m0, m3 > + punpcklwd m0, m1 ; m0 = 0, 1, 6, 7, 12, 13, 18, 19 > + punpckhwd m1, m2 ; m1 = 4, 5, 10, 11, 16, 17, 22, 23 > + punpcklwd m2, m3 ; m2 = 2, 3, 8, 9, 14, 15, 20, 21 > + ; unpack dwords: > + punpckhdq m3, m0, m2 ; m3 = 12, 13, 14, 15, 18, 19, 20, 21 > + punpckldq m2, m1 ; m2 = 2, 3, 4, 5, 8, 9, 10, 11 > + ; shuffle qwords: > + movhlps m4, m2 ; m4 = 8, 9, 10, 11, x, x, x, x > + movlhps m4, m3 ; m4 = 8, 9, 10, 11, 12, 13, 14, 15 (final) > + movlhps m2, m0 ; m2 = 2, 3, 4, 5, 0, 1, 6, 7 > + movhlps m3, m1 ; m3 = 16, 17, 22, 23, 18, 19, 20, 21 > + ; shuffle dwords: > + pshufd m2, m2, q3102 ; m2 = 0, 1, 2, 3, 4, 5, 6, 7 (final) > + pshufd m3, m3, q1320 ; m3 = 16, 17, 18, 19, 20, 21, 22, 23 (final) Same comment on shufps here as in previous emails. > + pxor m0, m0 ; convert s16 in m2 to s32 in m0 and m1 > + pxor m1, m1 > + punpcklwd m0, m2 > + punpckhwd m1, m2 > + pxor m2, m2 ; convert s16 in m4 to s32 in m2 and m5 > + pxor m5, m5 > + punpcklwd m2, m4 > + punpckhwd m5, m4 > + pxor m4, m4 ; convert s16 in m3 to s32 in m4 and m6 > + pxor m6, m6 > + punpcklwd m4, m3 > + punpckhwd m6, m3 An AVX version can avoid all these xors with 3-operand. > + cvtdq2ps m0, m0 ; convert s32 to float > + cvtdq2ps m1, m1 > + cvtdq2ps m2, m2 > + cvtdq2ps m5, m5 > + cvtdq2ps m4, m4 > + cvtdq2ps m6, m6 > + mulps m0, m7 ; scale float from s32 range to [-1.0,1.0] > + mulps m1, m7 > + mulps m2, m7 > + mulps m5, m7 > + mulps m4, m7 > + mulps m6, m7 > + mova [dstq ], m0 > + mova [dstq+ mmsize], m1 > + mova [dstq+2*mmsize], m2 > + mova [dstq+3*mmsize], m5 > + mova [dstq+4*mmsize], m4 > + mova [dstq+5*mmsize], m6 > + add srcq, mmsize/2 > + add dstq, mmsize*6 > + sub lend, mmsize/4 > + jg .loop > + REP_RET Otherwise looks okay. Jason _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
