Lines Matching refs:wq
58 movsxd wq, wd
91 shl wq, 1 ; this allows *16 (i.e. now *8) in lea instructions for the 8-tap filter
96 lea filterq, [filterq+wq*8]
98 lea dstq, [dstq+wq*(2>>wshr)]
100 lea dstq, [dstq+wq*(4>>wshr)]
102 lea fltposq, [fltposq+wq*(4>>wshr)]
103 neg wq
108 mov32 pos0q, dword [fltposq+wq*4+ 0] ; filterPos[0]
109 mov32 pos1q, dword [fltposq+wq*4+ 4] ; filterPos[1]
119 mov32 pos0q, dword [fltposq+wq*4+ 8] ; filterPos[2]
120 mov32 pos1q, dword [fltposq+wq*4+12] ; filterPos[3]
141 pmaddwd m0, [filterq+wq*8+mmsize*0] ; *= filter[{0,1,..,6,7}]
142 pmaddwd m1, [filterq+wq*8+mmsize*1] ; *= filter[{8,9,..,14,15}]
158 mov32 pos0q, dword [fltposq+wq*2+0] ; filterPos[0]
159 mov32 pos1q, dword [fltposq+wq*2+4] ; filterPos[1]
167 mov32 pos0q, dword [fltposq+wq*2+8] ; filterPos[2]
168 mov32 pos1q, dword [fltposq+wq*2+12] ; filterPos[3]
187 pmaddwd m0, [filterq+wq*8+mmsize*0] ; *= filter[{0,1,..,6,7}]
188 pmaddwd m1, [filterq+wq*8+mmsize*1] ; *= filter[{8,9,..,14,15}]
189 pmaddwd m4, [filterq+wq*8+mmsize*2] ; *= filter[{16,17,..,22,23}]
190 pmaddwd m5, [filterq+wq*8+mmsize*3] ; *= filter[{24,25,..,30,31}]
245 lea fltposq, [fltposq+wq*4]
247 lea dstq, [dstq+wq*2]
249 lea dstq, [dstq+wq*4]
252 neg wq
255 mov32 pos0q, dword [fltposq+wq*4+0] ; filterPos[0]
256 mov32 pos1q, dword [fltposq+wq*4+4] ; filterPos[1]
287 mov32 pos1q, dword [fltposq+wq*4+4] ; filterPos[1]
341 movh [dstq+wq*(2>>wshr)], m0
343 movd [dstq+wq*2], m0
348 mova [dstq+wq*(4>>wshr)], m0
350 movq [dstq+wq*4], m0
354 add wq, (mmsize<<wshr)/4 ; both 8tap and 4tap really only do 4 pixels
355 ; per iteration. see "shl wq,1" above as for why we do this
357 add wq, 2