nb_kernel312_ia32_sse.s

来自「最著名最快的分子模拟软件」· S 代码 · 共 2,034 行 · 第 1/5 页

S
2,034
字号
    cvtpi2ps %mm7,%xmm2        movlhps  %xmm2,%xmm3        subps    %xmm3,%xmm1    ## xmm1=eps     movaps %xmm1,%xmm2    mulps  %xmm2,%xmm2      ## xmm2=eps2         pslld   $2,%mm6        pslld   $2,%mm7    movd %mm6,%eax    psrlq $32,%mm6    movd %mm7,%ecx    psrlq $32,%mm7    movd %mm6,%ebx    movd %mm7,%edx    movlps (%esi,%eax,4),%xmm5    movlps (%esi,%ecx,4),%xmm7    movhps (%esi,%ebx,4),%xmm5    movhps (%esi,%edx,4),%xmm7 ## got half coulomb table     movaps %xmm5,%xmm4    shufps $136,%xmm7,%xmm4 ## constant 10001000    shufps $221,%xmm7,%xmm5 ## constant 11011101    movlps 8(%esi,%eax,4),%xmm7    movlps 8(%esi,%ecx,4),%xmm3    movhps 8(%esi,%ebx,4),%xmm7    movhps 8(%esi,%edx,4),%xmm3    ## other half of coulomb table      movaps %xmm7,%xmm6    shufps $136,%xmm3,%xmm6 ## constant 10001000    shufps $221,%xmm3,%xmm7 ## constant 11011101    ## coulomb table ready, in xmm4-xmm7      mulps  %xmm1,%xmm6      ## xmm6=Geps     mulps  %xmm2,%xmm7      ## xmm7=Heps2     addps  %xmm6,%xmm5    addps  %xmm7,%xmm5      ## xmm5=Fp     mulps  nb312_two(%esp),%xmm7         ## two*Heps2     movaps nb312_qqOH(%esp),%xmm3    addps  %xmm6,%xmm7    addps  %xmm5,%xmm7 ## xmm7=FF     mulps  %xmm1,%xmm5 ## xmm5=eps*Fp     addps  %xmm4,%xmm5 ## xmm5=VV     mulps  %xmm3,%xmm5 ## vcoul=qq*VV      mulps  %xmm7,%xmm3 ## fijC=FF*qq     ## at this point mm5 contains vcoul and mm3 fijC     addps  nb312_vctot(%esp),%xmm5    movaps %xmm5,nb312_vctot(%esp)        xorps  %xmm1,%xmm1        mulps  nb312_tsc(%esp),%xmm3        mulps  %xmm0,%xmm3        subps  %xmm3,%xmm1        movaps %xmm1,%xmm0        movaps %xmm1,%xmm2        xorps %xmm3,%xmm3        movaps %xmm3,%xmm4        movaps %xmm3,%xmm5        mulps nb312_dxOH1(%esp),%xmm0        mulps nb312_dyOH1(%esp),%xmm1        mulps nb312_dzOH1(%esp),%xmm2        subps %xmm0,%xmm3        subps %xmm1,%xmm4        subps %xmm2,%xmm5        addps nb312_fixO(%esp),%xmm0        addps nb312_fiyO(%esp),%xmm1        addps nb312_fizO(%esp),%xmm2        movaps %xmm3,nb312_fjxH1(%esp)        movaps %xmm4,nb312_fjyH1(%esp)        movaps %xmm5,nb312_fjzH1(%esp)        movaps %xmm0,nb312_fixO(%esp)        movaps %xmm1,nb312_fiyO(%esp)        movaps %xmm2,nb312_fizO(%esp)        ## O-H2 interaction          movaps nb312_rinvOH2(%esp),%xmm0        movaps %xmm0,%xmm1        mulps  nb312_rsqOH2(%esp),%xmm1   ## xmm1=r         mulps  nb312_tsc(%esp),%xmm1        movhlps %xmm1,%xmm2    cvttps2pi %xmm1,%mm6    cvttps2pi %xmm2,%mm7    ## mm6/mm7 contain lu indices     cvtpi2ps %mm6,%xmm3    cvtpi2ps %mm7,%xmm2        movlhps  %xmm2,%xmm3        subps    %xmm3,%xmm1    ## xmm1=eps     movaps %xmm1,%xmm2    mulps  %xmm2,%xmm2      ## xmm2=eps2         pslld   $2,%mm6        pslld   $2,%mm7    movd %mm6,%eax    psrlq $32,%mm6    movd %mm7,%ecx    psrlq $32,%mm7    movd %mm6,%ebx    movd %mm7,%edx    movlps (%esi,%eax,4),%xmm5    movlps (%esi,%ecx,4),%xmm7    movhps (%esi,%ebx,4),%xmm5    movhps (%esi,%edx,4),%xmm7 ## got half coulomb table     movaps %xmm5,%xmm4    shufps $136,%xmm7,%xmm4 ## constant 10001000    shufps $221,%xmm7,%xmm5 ## constant 11011101    movlps 8(%esi,%eax,4),%xmm7    movlps 8(%esi,%ecx,4),%xmm3    movhps 8(%esi,%ebx,4),%xmm7    movhps 8(%esi,%edx,4),%xmm3    ## other half of coulomb table      movaps %xmm7,%xmm6    shufps $136,%xmm3,%xmm6 ## constant 10001000    shufps $221,%xmm3,%xmm7 ## constant 11011101    ## coulomb table ready, in xmm4-xmm7      mulps  %xmm1,%xmm6      ## xmm6=Geps     mulps  %xmm2,%xmm7      ## xmm7=Heps2     addps  %xmm6,%xmm5    addps  %xmm7,%xmm5      ## xmm5=Fp     mulps  nb312_two(%esp),%xmm7         ## two*Heps2     movaps nb312_qqOH(%esp),%xmm3    addps  %xmm6,%xmm7    addps  %xmm5,%xmm7 ## xmm7=FF     mulps  %xmm1,%xmm5 ## xmm5=eps*Fp     addps  %xmm4,%xmm5 ## xmm5=VV     mulps  %xmm3,%xmm5 ## vcoul=qq*VV      mulps  %xmm7,%xmm3 ## fijC=FF*qq     ## at this point mm5 contains vcoul and mm3 fijC     addps  nb312_vctot(%esp),%xmm5    movaps %xmm5,nb312_vctot(%esp)        xorps  %xmm1,%xmm1        mulps  nb312_tsc(%esp),%xmm3        mulps  %xmm0,%xmm3        subps  %xmm3,%xmm1        movaps %xmm1,%xmm0        movaps %xmm1,%xmm2        xorps %xmm3,%xmm3        movaps %xmm3,%xmm4        movaps %xmm3,%xmm5        mulps nb312_dxOH2(%esp),%xmm0        mulps nb312_dyOH2(%esp),%xmm1        mulps nb312_dzOH2(%esp),%xmm2        subps %xmm0,%xmm3        subps %xmm1,%xmm4        subps %xmm2,%xmm5        addps nb312_fixO(%esp),%xmm0        addps nb312_fiyO(%esp),%xmm1        addps nb312_fizO(%esp),%xmm2        movaps %xmm3,nb312_fjxH2(%esp)        movaps %xmm4,nb312_fjyH2(%esp)        movaps %xmm5,nb312_fjzH2(%esp)        movaps %xmm0,nb312_fixO(%esp)        movaps %xmm1,nb312_fiyO(%esp)        movaps %xmm2,nb312_fizO(%esp)        ## H1-O interaction         movaps nb312_rinvH1O(%esp),%xmm0        movaps %xmm0,%xmm1        mulps  nb312_rsqH1O(%esp),%xmm1   ## xmm1=r         mulps  nb312_tsc(%esp),%xmm1        movhlps %xmm1,%xmm2    cvttps2pi %xmm1,%mm6    cvttps2pi %xmm2,%mm7    ## mm6/mm7 contain lu indices     cvtpi2ps %mm6,%xmm3    cvtpi2ps %mm7,%xmm2        movlhps  %xmm2,%xmm3        subps    %xmm3,%xmm1    ## xmm1=eps     movaps %xmm1,%xmm2    mulps  %xmm2,%xmm2      ## xmm2=eps2         pslld   $2,%mm6        pslld   $2,%mm7    movd %mm6,%eax    psrlq $32,%mm6    movd %mm7,%ecx    psrlq $32,%mm7    movd %mm6,%ebx    movd %mm7,%edx    movlps (%esi,%eax,4),%xmm5    movlps (%esi,%ecx,4),%xmm7    movhps (%esi,%ebx,4),%xmm5    movhps (%esi,%edx,4),%xmm7 ## got half coulomb table     movaps %xmm5,%xmm4    shufps $136,%xmm7,%xmm4 ## constant 10001000    shufps $221,%xmm7,%xmm5 ## constant 11011101    movlps 8(%esi,%eax,4),%xmm7    movlps 8(%esi,%ecx,4),%xmm3    movhps 8(%esi,%ebx,4),%xmm7    movhps 8(%esi,%edx,4),%xmm3    ## other half of coulomb table      movaps %xmm7,%xmm6    shufps $136,%xmm3,%xmm6 ## constant 10001000    shufps $221,%xmm3,%xmm7 ## constant 11011101    ## coulomb table ready, in xmm4-xmm7      mulps  %xmm1,%xmm6      ## xmm6=Geps     mulps  %xmm2,%xmm7      ## xmm7=Heps2     addps  %xmm6,%xmm5    addps  %xmm7,%xmm5      ## xmm5=Fp     mulps  nb312_two(%esp),%xmm7         ## two*Heps2     movaps nb312_qqOH(%esp),%xmm3    addps  %xmm6,%xmm7    addps  %xmm5,%xmm7 ## xmm7=FF     mulps  %xmm1,%xmm5 ## xmm5=eps*Fp     addps  %xmm4,%xmm5 ## xmm5=VV     mulps  %xmm3,%xmm5 ## vcoul=qq*VV      mulps  %xmm7,%xmm3 ## fijC=FF*qq     ## at this point mm5 contains vcoul and mm3 fijC     addps  nb312_vctot(%esp),%xmm5    movaps %xmm5,nb312_vctot(%esp)        xorps  %xmm1,%xmm1        mulps  nb312_tsc(%esp),%xmm3        mulps  %xmm0,%xmm3        subps  %xmm3,%xmm1        movaps %xmm1,%xmm0        movaps %xmm1,%xmm2        movaps nb312_fjxO(%esp),%xmm3        movaps nb312_fjyO(%esp),%xmm4        movaps nb312_fjzO(%esp),%xmm5        mulps nb312_dxH1O(%esp),%xmm0        mulps nb312_dyH1O(%esp),%xmm1        mulps nb312_dzH1O(%esp),%xmm2        subps %xmm0,%xmm3        subps %xmm1,%xmm4        subps %xmm2,%xmm5        addps nb312_fixH1(%esp),%xmm0        addps nb312_fiyH1(%esp),%xmm1        addps nb312_fizH1(%esp),%xmm2        movaps %xmm3,nb312_fjxO(%esp)        movaps %xmm4,nb312_fjyO(%esp)        movaps %xmm5,nb312_fjzO(%esp)        movaps %xmm0,nb312_fixH1(%esp)        movaps %xmm1,nb312_fiyH1(%esp)        movaps %xmm2,nb312_fizH1(%esp)        ## H1-H1 interaction         movaps nb312_rinvH1H1(%esp),%xmm0        movaps %xmm0,%xmm1        mulps  nb312_rsqH1H1(%esp),%xmm1   ## xmm1=r         mulps  nb312_tsc(%esp),%xmm1        movhlps %xmm1,%xmm2    cvttps2pi %xmm1,%mm6    cvttps2pi %xmm2,%mm7    ## mm6/mm7 contain lu indices     cvtpi2ps %mm6,%xmm3    cvtpi2ps %mm7,%xmm2        movlhps  %xmm2,%xmm3        subps    %xmm3,%xmm1    ## xmm1=eps     movaps %xmm1,%xmm2    mulps  %xmm2,%xmm2      ## xmm2=eps2         pslld   $2,%mm6        pslld   $2,%mm7    movd %mm6,%eax    psrlq $32,%mm6    movd %mm7,%ecx    psrlq $32,%mm7    movd %mm6,%ebx    movd %mm7,%edx    movlps (%esi,%eax,4),%xmm5    movlps (%esi,%ecx,4),%xmm7    movhps (%esi,%ebx,4),%xmm5    movhps (%esi,%edx,4),%xmm7 ## got half coulomb table     movaps %xmm5,%xmm4    shufps $136,%xmm7,%xmm4 ## constant 10001000    shufps $221,%xmm7,%xmm5 ## constant 11011101    movlps 8(%esi,%eax,4),%xmm7    movlps 8(%esi,%ecx,4),%xmm3    movhps 8(%esi,%ebx,4),%xmm7    movhps 8(%esi,%edx,4),%xmm3    ## other half of coulomb table      movaps %xmm7,%xmm6    shufps $136,%xmm3,%xmm6 ## constant 10001000    shufps $221,%xmm3,%xmm7 ## constant 11011101    ## coulomb table ready, in xmm4-xmm7      mulps  %xmm1,%xmm6      ## xmm6=Geps     mulps  %xmm2,%xmm7      ## xmm7=Heps2     addps  %xmm6,%xmm5    addps  %xmm7,%xmm5      ## xmm5=Fp     mulps  nb312_two(%esp),%xmm7         ## two*Heps2     movaps nb312_qqHH(%esp),%xmm3    addps  %xmm6,%xmm7    addps  %xmm5,%xmm7 ## xmm7=FF     mulps  %xmm1,%xmm5 ## xmm5=eps*Fp     addps  %xmm4,%xmm5 ## xmm5=VV     mulps  %xmm3,%xmm5 ## vcoul=qq*VV      mulps  %xmm7,%xmm3 ## fijC=FF*qq     ## at this point mm5 contains vcoul and mm3 fijC     addps  nb312_vctot(%esp),%xmm5    movaps %xmm5,nb312_vctot(%esp)        xorps  %xmm1,%xmm1        mulps  nb312_tsc(%esp),%xmm3        mulps  %xmm0,%xmm3        subps  %xmm3,%xmm1        movaps %xmm1,%xmm0        movaps %xmm1,%xmm2        movaps nb312_fjxH1(%esp),%xmm3        movaps nb312_fjyH1(%esp),%xmm4        movaps nb312_fjzH1(%esp),%xmm5        mulps nb312_dxH1H1(%esp),%xmm0        mulps nb312_dyH1H1(%esp),%xmm1        mulps nb312_dzH1H1(%esp),%xmm2        subps %xmm0,%xmm3        subps %xmm1,%xmm4        subps %xmm2,%xmm5        addps nb312_fixH1(%esp),%xmm0        addps nb312_fiyH1(%esp),%xmm1        addps nb312_fizH1(%esp),%xmm2        movaps %xmm3,nb312_fjxH1(%esp)        movaps %xmm4,nb312_fjyH1(%esp)        movaps %xmm5,nb312_fjzH1(%esp)        movaps %xmm0,nb312_fixH1(%esp)        movaps %xmm1,nb312_fiyH1(%esp)        movaps %xmm2,nb312_fizH1(%esp)        ## H1-H2 interaction         movaps nb312_rinvH1H2(%esp),%xmm0        movaps %xmm0,%xmm1        mulps  nb312_rsqH1H2(%esp),%xmm1   ## xmm1=r         mulps  nb312_tsc(%esp),%xmm1        movhlps %xmm1,%xmm2    cvttps2pi %xmm1,%mm6    cvttps2pi %xmm2,%mm7    ## mm6/mm7 contain lu indices     cvtpi2ps %mm6,%xmm3    cvtpi2ps %mm7,%xmm2        movlhps  %xmm2,%xmm3        subps    %xmm3,%xmm1    ## xmm1=eps     movaps %xmm1,%xmm2    mulps  %xmm2,%xmm2      ## xmm2=eps2         pslld   $2,%mm6        pslld   $2,%mm7    movd %mm6,%eax    psrlq $32,%mm6    movd %mm7,%ecx    psrlq $32,%mm7    movd %mm6,%ebx    movd %mm7,%edx    movlps (%esi,%eax,4),%xmm5    movlps (%esi,%ecx,4),%xmm7    movhps (%esi,%ebx,4),%xmm5    movhps (%esi,%edx,4),%xmm7 ## got half coulomb table     movaps %xmm5,%xmm4    shufps $136,%xmm7,%xmm4 ## constant 10001000    shufps $221,%xmm7,%xmm5 ## constant 11011101    movlps 8(%esi,%eax,4),%xmm7    movlps 8(%esi,%ecx,4),%xmm3    movhps 8(%esi,%ebx,4),%xmm7    movhps 8(%esi,%edx,4),%xmm3    ## other half of coulomb table      movaps %xmm7,%xmm6    shufps $136,%xmm3,%xmm6 ## constant 10001000    shufps $221,%xmm3,%xmm7 ## constant 11011101    ## coulomb table ready, in xmm4-xmm7      mulps  %xmm1,%xmm6      ## xmm6=Geps     mulps  %xmm2,%xmm7      ## xmm7=Heps2     addps  %xmm6,%xmm5    addps  %xmm7,%xmm5      ## xmm5=Fp     mulps  nb312_two(%esp),%xmm7         ## two*Heps2     movaps nb312_qqHH(%esp),%xmm3    addps  %xmm6,%xmm7    addps  %xmm5,%xmm7 ## xmm7=FF     mulps  %xmm1,%xmm5 ## xmm5=eps*Fp     addps  %xmm4,%xmm5 ## xmm5=VV     mulps  %xmm3,%xmm5 ## vcoul=qq*VV      mulps  %xmm7,%xmm3 ## fijC=FF*qq     ## at this point mm5 contains vcoul and mm3 fijC     addps  nb312_vctot(%esp),%xmm5    movaps %xmm5,nb312_vctot(%esp)        xorps  %xmm1,%xmm1        mulps  nb312_tsc(%esp),%xmm3        mulps  %xmm0,%xmm3        subps  %xmm3,%xmm1        movaps %xmm1,%xmm0        movaps %xmm1,%xmm2        movaps nb312_fjxH2(%esp),%xmm3        movaps nb312_fjyH2(%esp),%xmm4        movaps nb312_fjzH2(%esp),%xmm5        mulps nb312_dxH1H2(%esp),%xmm0        mulps nb312_dyH1H2(%esp),%xmm1

⌨️ 快捷键说明

复制代码Ctrl + C
搜索代码Ctrl + F
全屏模式F11
增大字号Ctrl + =
减小字号Ctrl + -
显示快捷键?