Lines Matching refs:W512
81 &movq ("mm4",&QWP($Foff,$W512)); # load f
82 &movq ("mm5",&QWP($Goff,$W512)); # load g
83 &movq ("mm6",&QWP($Hoff,$W512)); # load h
100 &movq (&QWP($Foff,$W512),$E); # f = e
101 &movq (&QWP($Goff,$W512),"mm4"); # g = f
102 &movq (&QWP($Hoff,$W512),"mm5"); # h = g
109 &movq ("mm2",&QWP($Boff,$W512)); # load b
110 &movq ("mm3",&QWP($Coff,$W512)); # load c
111 &movq ($E,&QWP($Doff,$W512)); # e = d
115 &paddq ("mm7",&QWP(0,$W512,$widx,8)); # T1+=W512[i]
133 &movq (&QWP($Boff,$W512),$A); # b = a
134 &movq (&QWP($Coff,$W512),"mm2"); # c = b
135 &movq (&QWP($Doff,$W512),"mm3"); # d = c
172 $W512 = "esp"; # start using %esp as W512
173 &sub ($W512,$W512_SZ);
174 &and ($W512,-16); # ensure 128-bit alignment
186 &movdqa (&QWP($Aoff,$W512),"xmm0"); # a,b
187 &movdqa (&QWP($Coff,$W512),"xmm1"); # c,d
188 &movdqa (&QWP($Eoff,$W512),"xmm2"); # e,f
189 &movdqa (&QWP($Goff,$W512),"xmm3"); # g,h
210 &mov (&DWP(0,$W512,$Widx,8),"ebx"); # W512[i]
211 &mov (&DWP(4,$W512,$Widx,8),"eax");
212 &mov (&DWP(128+0,$W512,$Widx,8),"ebx"); # copy of W512[i]
213 &mov (&DWP(128+4,$W512,$Widx,8),"eax");
222 &mov (&DWP(0+8,$W512,$Widx,8),"ebx"); # W512[i]
223 &mov (&DWP(4+8,$W512,$Widx,8),"eax");
224 &mov (&DWP(128+0+8,$W512,$Widx,8),"ebx"); # copy of W512[i]
225 &mov (&DWP(128+4+8,$W512,$Widx,8),"eax");
245 &movdqu ("xmm0",&QWP(8*1,$W512,$Widx,8)); # s0=W512[i+1]
259 &movdqa ("xmm1",&QWP(8*14,$W512,$Widx,8)); # s1=W512[i+14]
275 &movdqu ("xmm6",&QWP(8*9,$W512,$Widx,8));
278 &paddq ("xmm0",&QWP(0,$W512,$Widx,8)); # s0 += W512[i]
280 &movdqa (&QWP(0,$W512,$Widx,8),"xmm0"); # W512[i] = s0
281 &movdqa (&QWP(16*8,$W512,$Widx,8),"xmm0"); # copy of W512[i]
292 &movq (&QWP($Aoff,$W512),$A); # write out a
293 &movq (&QWP($Eoff,$W512),$E); # write out e
298 &paddq ("xmm0",&QWP($Aoff,$W512)); # 128-bit additions...
299 &paddq ("xmm1",&QWP($Coff,$W512));
300 &paddq ("xmm2",&QWP($Eoff,$W512));
301 &paddq ("xmm3",&QWP($Goff,$W512));