nb_kernel202_ia32_sse2.intel_syntax.s
来自「最著名最快的分子模拟软件」· S 代码 · 共 2,125 行 · 第 1/5 页
S
2,125 行
;# H1-H2 interaction movapd xmm0, [esp + nb202_rinvH1H2] movapd xmm7, xmm0 ;# xmm7=rinv movapd xmm5, [esp + nb202_krf] movapd xmm1, xmm0 mulsd xmm5, [esp + nb202_rsqH1H2] ;# xmm5=krsq movapd xmm4, xmm5 addsd xmm4, xmm7 ;# xmm4=r inv+ krsq mulsd xmm0, xmm0 subsd xmm4, [esp + nb202_crf] mulsd xmm4, [esp + nb202_qqHH] ;# xmm4=voul=qq*(rinv+ krsq) mulsd xmm5, [esp + nb202_two] subsd xmm7, xmm5 ;# xmm7=rinv-2*krsq mulsd xmm7, [esp + nb202_qqHH] ;# xmm7 = coul part of fscal addsd xmm6, xmm4 ;# add to local vctot mulsd xmm0, xmm7 ;# fsOH2 movapd xmm1, xmm0 movapd xmm2, xmm0 movapd xmm3, [esp + nb202_fjxH2] movapd xmm4, [esp + nb202_fjyH2] movapd xmm5, [esp + nb202_fjzH2] mulsd xmm0, [esp + nb202_dxH1H2] mulsd xmm1, [esp + nb202_dyH1H2] mulsd xmm2, [esp + nb202_dzH1H2] subsd xmm3, xmm0 subsd xmm4, xmm1 subsd xmm5, xmm2 addsd xmm0, [esp + nb202_fixH1] addsd xmm1, [esp + nb202_fiyH1] addsd xmm2, [esp + nb202_fizH1] movlpd [esp + nb202_fjxH2], xmm3 movlpd [esp + nb202_fjyH2], xmm4 movlpd [esp + nb202_fjzH2], xmm5 movlpd [esp + nb202_fixH1], xmm0 movlpd [esp + nb202_fiyH1], xmm1 movlpd [esp + nb202_fizH1], xmm2 ;# H2-O interaction movapd xmm0, [esp + nb202_rinvH2O] movapd xmm7, xmm0 ;# xmm7=rinv movapd xmm5, [esp + nb202_krf] movapd xmm1, xmm0 mulsd xmm5, [esp + nb202_rsqH2O] ;# xmm5=krsq movapd xmm4, xmm5 addsd xmm4, xmm7 ;# xmm4=r inv+ krsq subsd xmm4, [esp + nb202_crf] mulsd xmm0, xmm0 mulsd xmm4, [esp + nb202_qqOH] ;# xmm4=voul=qq*(rinv+ krsq) mulsd xmm5, [esp + nb202_two] subsd xmm7, xmm5 ;# xmm7=rinv-2*krsq mulsd xmm7, [esp + nb202_qqOH] ;# xmm7 = coul part of fscal addsd xmm6, xmm4 ;# add to local vctot mulsd xmm0, xmm7 ;# fsOH2 movapd xmm1, xmm0 movapd xmm2, xmm0 movapd xmm3, [esp + nb202_fjxO] movapd xmm4, [esp + nb202_fjyO] movapd xmm5, [esp + nb202_fjzO] mulsd xmm0, [esp + nb202_dxH2O] mulsd xmm1, [esp + nb202_dyH2O] mulsd xmm2, [esp + nb202_dzH2O] subsd xmm3, xmm0 subsd xmm4, xmm1 subsd xmm5, xmm2 addsd xmm0, [esp + nb202_fixH2] addsd xmm1, [esp + nb202_fiyH2] addsd xmm2, [esp + nb202_fizH2] movlpd [esp + nb202_fjxO], xmm3 movlpd [esp + nb202_fjyO], xmm4 movlpd [esp + nb202_fjzO], xmm5 movlpd [esp + nb202_fixH2], xmm0 movlpd [esp + nb202_fiyH2], xmm1 movlpd [esp + nb202_fizH2], xmm2 ;# H2-H1 interaction movapd xmm0, [esp + nb202_rinvH2H1] movapd xmm7, xmm0 ;# xmm7=rinv movapd xmm5, [esp + nb202_krf] movapd xmm1, xmm0 mulsd xmm5, [esp + nb202_rsqH2H1] ;# xmm5=krsq movapd xmm4, xmm5 addsd xmm4, xmm7 ;# xmm4=r inv+ krsq subsd xmm4, [esp + nb202_crf] mulsd xmm0, xmm0 mulsd xmm4, [esp + nb202_qqHH] ;# xmm4=voul=qq*(rinv+ krsq) mulsd xmm5, [esp + nb202_two] subsd xmm7, xmm5 ;# xmm7=rinv-2*krsq mulsd xmm7, [esp + nb202_qqHH] ;# xmm7 = coul part of fscal addsd xmm6, xmm4 ;# add to local vctot mulsd xmm0, xmm7 ;# fsOH2 movapd xmm1, xmm0 movapd xmm2, xmm0 movapd xmm3, [esp + nb202_fjxH1] movapd xmm4, [esp + nb202_fjyH1] movapd xmm5, [esp + nb202_fjzH1] mulsd xmm0, [esp + nb202_dxH2H1] mulsd xmm1, [esp + nb202_dyH2H1] mulsd xmm2, [esp + nb202_dzH2H1] subsd xmm3, xmm0 subsd xmm4, xmm1 subsd xmm5, xmm2 addsd xmm0, [esp + nb202_fixH2] addsd xmm1, [esp + nb202_fiyH2] addsd xmm2, [esp + nb202_fizH2] movlpd [esp + nb202_fjxH1], xmm3 movlpd [esp + nb202_fjyH1], xmm4 movlpd [esp + nb202_fjzH1], xmm5 movlpd [esp + nb202_fixH2], xmm0 movlpd [esp + nb202_fiyH2], xmm1 movlpd [esp + nb202_fizH2], xmm2 ;# H2-H2 interaction movapd xmm0, [esp + nb202_rinvH2H2] movapd xmm7, xmm0 ;# xmm7=rinv movapd xmm5, [esp + nb202_krf] movapd xmm1, xmm0 mulsd xmm5, [esp + nb202_rsqH2H2] ;# xmm5=krsq movapd xmm4, xmm5 addsd xmm4, xmm7 ;# xmm4=r inv+ krsq subsd xmm4, [esp + nb202_crf] mulsd xmm0, xmm0 mulsd xmm4, [esp + nb202_qqHH] ;# xmm4=voul=qq*(rinv+ krsq) mulsd xmm5, [esp + nb202_two] subsd xmm7, xmm5 ;# xmm7=rinv-2*krsq mulsd xmm7, [esp + nb202_qqHH] ;# xmm7 = coul part of fscal addsd xmm6, xmm4 ;# add to local vctot mulsd xmm0, xmm7 ;# fsOH2 movapd xmm1, xmm0 movapd xmm2, xmm0 movapd xmm1, xmm0 movlpd [esp + nb202_vctot], xmm6 movapd xmm2, xmm0 movapd xmm3, [esp + nb202_fjxH2] movapd xmm4, [esp + nb202_fjyH2] movapd xmm5, [esp + nb202_fjzH2] mulsd xmm0, [esp + nb202_dxH2H2] mulsd xmm1, [esp + nb202_dyH2H2] mulsd xmm2, [esp + nb202_dzH2H2] subsd xmm3, xmm0 subsd xmm4, xmm1 subsd xmm5, xmm2 addsd xmm0, [esp + nb202_fixH2] addsd xmm1, [esp + nb202_fiyH2] addsd xmm2, [esp + nb202_fizH2] movlpd [esp + nb202_fjxH2], xmm3 movlpd [esp + nb202_fjyH2], xmm4 movlpd [esp + nb202_fjzH2], xmm5 movlpd [esp + nb202_fixH2], xmm0 movlpd [esp + nb202_fiyH2], xmm1 movlpd [esp + nb202_fizH2], xmm2 mov edi, [ebp + nb202_faction] ;# Did all interactions - now update j forces movlpd xmm0, [edi + eax*8] movlpd xmm1, [edi + eax*8 + 8] movlpd xmm2, [edi + eax*8 + 16] movlpd xmm3, [edi + eax*8 + 24] movlpd xmm4, [edi + eax*8 + 32] movlpd xmm5, [edi + eax*8 + 40] movlpd xmm6, [edi + eax*8 + 48] movlpd xmm7, [edi + eax*8 + 56] addsd xmm0, [esp + nb202_fjxO] addsd xmm1, [esp + nb202_fjyO] addsd xmm2, [esp + nb202_fjzO] addsd xmm3, [esp + nb202_fjxH1] addsd xmm4, [esp + nb202_fjyH1] addsd xmm5, [esp + nb202_fjzH1] addsd xmm6, [esp + nb202_fjxH2] addsd xmm7, [esp + nb202_fjyH2] movlpd [edi + eax*8], xmm0 movlpd [edi + eax*8 + 8], xmm1 movlpd [edi + eax*8 + 16], xmm2 movlpd [edi + eax*8 + 24], xmm3 movlpd [edi + eax*8 + 32], xmm4 movlpd [edi + eax*8 + 40], xmm5 movlpd [edi + eax*8 + 48], xmm6 movlpd [edi + eax*8 + 56], xmm7 movlpd xmm0, [edi + eax*8 + 64] addsd xmm0, [esp + nb202_fjzH2] movlpd [edi + eax*8 + 64], xmm0 .nb202_updateouterdata: mov ecx, [esp + nb202_ii3] mov edi, [ebp + nb202_faction] mov esi, [ebp + nb202_fshift] mov edx, [esp + nb202_is3] ;# accumulate Oi forces in xmm0, xmm1, xmm2 movapd xmm0, [esp + nb202_fixO] movapd xmm1, [esp + nb202_fiyO] movapd xmm2, [esp + nb202_fizO] movhlps xmm3, xmm0 movhlps xmm4, xmm1 movhlps xmm5, xmm2 addsd xmm0, xmm3 addsd xmm1, xmm4 addsd xmm2, xmm5 ;# sum is in low xmm0-xmm2 movapd xmm3, xmm0 movapd xmm4, xmm1 movapd xmm5, xmm2 ;# increment i force movsd xmm3, [edi + ecx*8] movsd xmm4, [edi + ecx*8 + 8] movsd xmm5, [edi + ecx*8 + 16] addsd xmm3, xmm0 addsd xmm4, xmm1 addsd xmm5, xmm2 movsd [edi + ecx*8], xmm3 movsd [edi + ecx*8 + 8], xmm4 movsd [edi + ecx*8 + 16], xmm5 ;# accumulate force in xmm6/xmm7 for fshift movapd xmm6, xmm0 movsd xmm7, xmm2 unpcklpd xmm6, xmm1 ;# accumulate H1i forces in xmm0, xmm1, xmm2 movapd xmm0, [esp + nb202_fixH1] movapd xmm1, [esp + nb202_fiyH1] movapd xmm2, [esp + nb202_fizH1] movhlps xmm3, xmm0 movhlps xmm4, xmm1 movhlps xmm5, xmm2 addsd xmm0, xmm3 addsd xmm1, xmm4 addsd xmm2, xmm5 ;# sum is in low xmm0-xmm2 ;# increment i force movsd xmm3, [edi + ecx*8 + 24] movsd xmm4, [edi + ecx*8 + 32] movsd xmm5, [edi + ecx*8 + 40] addsd xmm3, xmm0 addsd xmm4, xmm1 addsd xmm5, xmm2 movsd [edi + ecx*8 + 24], xmm3 movsd [edi + ecx*8 + 32], xmm4 movsd [edi + ecx*8 + 40], xmm5 ;# accumulate force in xmm6/xmm7 for fshift addsd xmm7, xmm2 unpcklpd xmm0, xmm1 addpd xmm6, xmm0 ;# accumulate H2i forces in xmm0, xmm1, xmm2 movapd xmm0, [esp + nb202_fixH2] movapd xmm1, [esp + nb202_fiyH2] movapd xmm2, [esp + nb202_fizH2] movhlps xmm3, xmm0 movhlps xmm4, xmm1 movhlps xmm5, xmm2 addsd xmm0, xmm3 addsd xmm1, xmm4 addsd xmm2, xmm5 ;# sum is in low xmm0-xmm2 movapd xmm3, xmm0 movapd xmm4, xmm1 movapd xmm5, xmm2 ;# increment i force movsd xmm3, [edi + ecx*8 + 48] movsd xmm4, [edi + ecx*8 + 56] movsd xmm5, [edi + ecx*8 + 64] addsd xmm3, xmm0 addsd xmm4, xmm1 addsd xmm5, xmm2 movsd [edi + ecx*8 + 48], xmm3 movsd [edi + ecx*8 + 56], xmm4 movsd [edi + ecx*8 + 64], xmm5 ;# accumulate force in xmm6/xmm7 for fshift addsd xmm7, xmm2 unpcklpd xmm0, xmm1 addpd xmm6, xmm0 ;# increment fshift force movlpd xmm3, [esi + edx*8] movhpd xmm3, [esi + edx*8 + 8] movsd xmm4, [esi + edx*8 + 16] addpd xmm3, xmm6 addsd xmm4, xmm7 movlpd [esi + edx*8], xmm3 movhpd [esi + edx*8 + 8], xmm3 movsd [esi + edx*8 + 16], xmm4 ;# get n from stack mov esi, [esp + nb202_n] ;# get group index for i particle mov edx, [ebp + nb202_gid] ;# base of gid[] mov edx, [edx + esi*4] ;# ggid=gid[n] ;# accumulate total potential energy and update it movapd xmm7, [esp + nb202_vctot] ;# accumulate movhlps xmm6, xmm7 addsd xmm7, xmm6 ;# low xmm7 has the sum now ;# add earlier value from mem mov eax, [ebp + nb202_Vc] addsd xmm7, [eax + edx*8] ;# move back to mem movsd [eax + edx*8], xmm7 ;# finish if last mov ecx, [esp + nb202_nn1] ;# esi already loaded with n inc esi sub ecx, esi jz .nb202_outerend ;# not last, iterate outer loop once more! mov [esp + nb202_n], esi jmp .nb202_outer.nb202_outerend: ;# check if more outer neighborlists remain mov ecx, [esp + nb202_nri] ;# esi already loaded with n above sub ecx, esi jz .nb202_end ;# non-zero, do one more workunit jmp .nb202_threadloop.nb202_end: emms mov eax, [esp + nb202_nouter] mov ebx, [esp + nb202_ninner] mov ecx, [ebp + nb202_outeriter] mov edx, [ebp + nb202_inneriter] mov [ecx], eax mov [edx], ebx mov eax, [esp + nb202_salign] add esp, eax add esp, 1480 pop edi pop esi pop edx pop ecx pop ebx pop eax leave ret .globl nb_kernel202nf_ia32_sse2.globl _nb_kernel202nf_ia32_sse2nb_kernel202nf_ia32_sse2: _nb_kernel202nf_ia32_sse2: .equiv nb202nf_p_nri, 8.equiv nb202nf_iinr, 12.equiv nb202nf_jindex, 16.equiv nb202nf_jjnr, 20.equiv nb202nf_shift, 24.equiv nb202nf_shiftvec, 28.equiv nb202nf_fshift, 32.equiv nb202nf_gid, 36.equiv nb202nf_pos, 40.equiv nb202nf_faction, 44.equiv nb202nf_charge, 48.equiv nb202nf_p_facel, 52.equiv nb202nf_argkrf, 56.equiv nb202nf_argcrf, 60.equiv nb202nf_Vc, 64.equiv nb202nf_type, 68.equiv nb202nf_p_ntype, 72.equiv nb202nf_vdwparam, 76.equiv nb202nf_Vvdw, 80.equiv nb202nf_p_tabscale, 84.equiv nb202nf_VFtab, 88.equiv nb202nf_invsqrta, 92.equiv nb202nf_dvda, 96.equiv nb202nf_p_gbtabscale, 100.equiv nb202nf_GBtab, 104.equiv nb202nf_p_nthreads, 108.equiv nb202nf_count, 112.equiv nb202nf_mtx, 116.equiv nb202nf_outeriter, 120.equiv nb202nf_inneriter, 124.equiv nb202nf_work, 128 ;# stack offsets for local variables ;# bottom of stack is cache-aligned for sse use .equiv nb202nf_ixO, 0.equiv nb202nf_iyO, 16.equiv nb202nf_izO, 32.equiv nb202nf_ixH1, 48.equiv nb202nf_iyH1, 64.equiv nb202nf_izH1, 80.equiv nb202nf_ixH2, 96.equiv nb202nf_iyH2, 112.equiv nb202nf_izH2, 128.equiv nb202nf_jxO, 144.equiv nb202nf_jyO, 160.equiv nb202nf_jzO, 176.equiv nb202nf_jxH1, 192.equiv nb202nf_jyH1, 208.equiv nb202nf_jzH1, 224.equiv nb202nf_jxH2, 240.equiv nb202nf_jyH2, 256.equiv nb202nf_jzH2, 272.equiv nb202nf_qqOO, 288.equiv nb202nf_qqOH, 304.equiv nb202nf_qqHH, 320.equiv nb202nf_vctot, 336.equiv nb202nf_half, 352.equiv nb202nf_three, 368.equiv nb202nf_rsqOO, 384.equiv nb202nf_rsqOH1, 400.equiv nb202nf_rsqOH2, 416.equiv nb202nf_rsqH1O, 432.equiv nb202nf_rsqH1H1, 448.equiv nb202nf_rsqH1H2, 464.equiv ⌨️ 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?