nb_kernel312_ia32_sse2.intel_syntax.s
来自「最著名最快的分子模拟软件」· S 代码 · 共 2,210 行 · 第 1/5 页
S
2,210 行
cvtsd2ss xmm1, xmm0 cvtsd2ss xmm5, xmm4 rsqrtss xmm1, xmm1 rsqrtss xmm5, xmm5 cvtss2sd xmm1, xmm1 cvtss2sd xmm5, xmm5 movapd xmm2, xmm1 ;# copy of luA movapd xmm6, xmm5 ;# copy of luB mulsd xmm1, xmm1 ;# luA*luA mulsd xmm5, xmm5 ;# luB*luB movapd xmm3, [esp + nb312_three] mulsd xmm1, xmm0 ;# rsqA*luA*luA mulsd xmm5, xmm4 ;# rsqB*luB*luB movapd xmm7, xmm3 subsd xmm3, xmm1 ;# 3-rsqA*luA*luA subsd xmm7, xmm5 ;# 3-rsqB*luB*luB mulsd xmm3, xmm2 ;# luA*(3-rsqA*luA*luA) mulsd xmm7, xmm6 ;# luB*(3-rsqB*luB*luB) mulsd xmm3, [esp + nb312_half] ;# iter1 of mulsd xmm7, [esp + nb312_half] ;# iter1 of movapd xmm2, xmm3 ;# copy of luA movapd xmm6, xmm7 ;# copy of luB mulsd xmm3, xmm3 ;# luA*luA mulsd xmm7, xmm7 ;# luB*luB movapd xmm1, [esp + nb312_three] mulsd xmm3, xmm0 ;# rsqA*luA*luA mulsd xmm7, xmm4 ;# rsqB*luB*luB movapd xmm5, xmm1 subsd xmm1, xmm3 ;# 3-rsqA*luA*luA subsd xmm5, xmm7 ;# 3-rsqB*luB*luB mulsd xmm1, xmm2 ;# luA*(3-rsqA*luA*luA) mulsd xmm5, xmm6 ;# luB*(3-rsqB*luB*luB) mulsd xmm1, [esp + nb312_half] ;# rinv mulsd xmm5, [esp + nb312_half] ;# rinv movapd [esp + nb312_rinvOO], xmm1 movapd [esp + nb312_rinvOH1], xmm5 movapd xmm0, [esp + nb312_rsqOH2] movapd xmm4, [esp + nb312_rsqH1O] cvtsd2ss xmm1, xmm0 cvtsd2ss xmm5, xmm4 rsqrtss xmm1, xmm1 rsqrtss xmm5, xmm5 cvtss2sd xmm1, xmm1 cvtss2sd xmm5, xmm5 movapd xmm2, xmm1 ;# copy of luA movapd xmm6, xmm5 ;# copy of luB mulsd xmm1, xmm1 ;# luA*luA mulsd xmm5, xmm5 ;# luB*luB movapd xmm3, [esp + nb312_three] mulsd xmm1, xmm0 ;# rsqA*luA*luA mulsd xmm5, xmm4 ;# rsqB*luB*luB movapd xmm7, xmm3 subsd xmm3, xmm1 ;# 3-rsqA*luA*luA subsd xmm7, xmm5 ;# 3-rsqB*luB*luB mulsd xmm3, xmm2 ;# luA*(3-rsqA*luA*luA) mulsd xmm7, xmm6 ;# luB*(3-rsqB*luB*luB) mulsd xmm3, [esp + nb312_half] ;# iter1 mulsd xmm7, [esp + nb312_half] ;# iter1 movapd xmm2, xmm3 ;# copy of luA movapd xmm6, xmm7 ;# copy of luB mulsd xmm3, xmm3 ;# luA*luA mulsd xmm7, xmm7 ;# luB*luB movapd xmm1, [esp + nb312_three] mulsd xmm3, xmm0 ;# rsqA*luA*luA mulsd xmm7, xmm4 ;# rsqB*luB*luB movapd xmm5, xmm1 subsd xmm1, xmm3 ;# 3-rsqA*luA*luA subsd xmm5, xmm7 ;# 3-rsqB*luB*luB mulsd xmm1, xmm2 ;# luA*(3-rsqA*luA*luA) mulsd xmm5, xmm6 ;# luB*(3-rsqB*luB*luB) mulsd xmm1, [esp + nb312_half] ;# rinv mulsd xmm5, [esp + nb312_half] ;# rinv movapd [esp + nb312_rinvOH2], xmm1 movapd [esp + nb312_rinvH1O], xmm5 movapd xmm0, [esp + nb312_rsqH1H1] movapd xmm4, [esp + nb312_rsqH1H2] cvtsd2ss xmm1, xmm0 cvtsd2ss xmm5, xmm4 rsqrtss xmm1, xmm1 rsqrtss xmm5, xmm5 cvtss2sd xmm1, xmm1 cvtss2sd xmm5, xmm5 movapd xmm2, xmm1 ;# copy of luA movapd xmm6, xmm5 ;# copy of luB mulsd xmm1, xmm1 ;# luA*luA mulsd xmm5, xmm5 ;# luB*luB movapd xmm3, [esp + nb312_three] mulsd xmm1, xmm0 ;# rsqA*luA*luA mulsd xmm5, xmm4 ;# rsqB*luB*luB movapd xmm7, xmm3 subsd xmm3, xmm1 ;# 3-rsqA*luA*luA subsd xmm7, xmm5 ;# 3-rsqB*luB*luB mulsd xmm3, xmm2 ;# luA*(3-rsqA*luA*luA) mulsd xmm7, xmm6 ;# luB*(3-rsqB*luB*luB) mulsd xmm3, [esp + nb312_half] ;# iter1a mulsd xmm7, [esp + nb312_half] ;# iter1b movapd xmm2, xmm3 ;# copy of luA movapd xmm6, xmm7 ;# copy of luB mulsd xmm3, xmm3 ;# luA*luA mulsd xmm7, xmm7 ;# luB*luB movapd xmm1, [esp + nb312_three] mulsd xmm3, xmm0 ;# rsqA*luA*luA mulsd xmm7, xmm4 ;# rsqB*luB*luB movapd xmm5, xmm1 subsd xmm1, xmm3 ;# 3-rsqA*luA*luA subsd xmm5, xmm7 ;# 3-rsqB*luB*luB mulsd xmm1, xmm2 ;# luA*(3-rsqA*luA*luA) mulsd xmm5, xmm6 ;# luB*(3-rsqB*luB*luB) mulsd xmm1, [esp + nb312_half] ;# rinv mulsd xmm5, [esp + nb312_half] ;# rinv movapd [esp + nb312_rinvH1H1], xmm1 movapd [esp + nb312_rinvH1H2], xmm5 movapd xmm0, [esp + nb312_rsqH2O] cvtsd2ss xmm1, xmm0 rsqrtss xmm1, xmm1 cvtss2sd xmm1, xmm1 movapd xmm2, xmm1 ;# copy of luA mulsd xmm1, xmm1 ;# luA*luA movapd xmm3, [esp + nb312_three] mulsd xmm1, xmm0 ;# rsqA*luA*luA subsd xmm3, xmm1 ;# 3-rsqA*luA*luA mulsd xmm3, xmm2 ;# luA*(3-rsqA*luA*luA) mulsd xmm3, [esp + nb312_half] ;# iter1 movapd xmm2, xmm3 ;# copy of luA mulsd xmm3, xmm3 ;# luA*luA movapd xmm1, [esp + nb312_three] mulsd xmm3, xmm0 ;# rsqA*luA*luA subsd xmm1, xmm3 ;# 3-rsqA*luA*luA mulsd xmm1, xmm2 ;# luA*(3-rsqA*luA*luA) mulsd xmm1, [esp + nb312_half] ;# rinv movapd [esp + nb312_rinvH2O], xmm1 ;# start with OO interaction movapd xmm0, [esp + nb312_rinvOO] movapd xmm1, xmm0 mulsd xmm1, [esp + nb312_rsqOO] ;# xmm1=r mulsd xmm1, [esp + nb312_tsc] movd mm0, eax cvttsd2si eax, xmm1 ;# mm6 = lu idx cvtsi2sd xmm6, eax subsd xmm1, xmm6 ;# xmm1=eps movapd xmm2, xmm1 mulsd xmm2, xmm2 ;# xmm2=eps2 shl eax, 2 ;# idx *= 4 mov esi, [ebp + nb312_VFtab] movsd xmm4, [esi + eax*8] ;# Y1 movsd xmm5, [esi + eax*8 + 8] ;# F1 movsd xmm6, [esi + eax*8 + 16] ;# G1 movsd xmm7, [esi + eax*8 + 24] ;# H1 ;# coulomb table ready, in xmm4-xmm7 mulsd xmm6, xmm1 ;# xmm6=Geps mulsd xmm7, xmm2 ;# xmm7=Heps2 addsd xmm5, xmm6 addsd xmm5, xmm7 ;# xmm5=Fp mulsd xmm7, [esp + nb312_two] ;# two*Heps2 movapd xmm3, [esp + nb312_qqOO] addsd xmm7, xmm6 addsd xmm7, xmm5 ;# xmm7=FF mulsd xmm5, xmm1 ;# xmm5=eps*Fp addsd xmm5, xmm4 ;# xmm5=VV mulsd xmm5, xmm3 ;# vcoul=qq*VV mulsd xmm3, xmm7 ;# fijC=FF*qq ;# at this point mm5 contains vcoul and xmm3 fijC ;# increment vcoul - then we can get rid of mm5 ;# update vctot addsd xmm5, [esp + nb312_vctot] xorpd xmm2, xmm2 movlpd [esp + nb312_vctot], xmm5 mulsd xmm3, [esp + nb312_tsc] ;# start doing lj movapd xmm2, xmm0 mulsd xmm2, xmm2 movapd xmm1, xmm2 mulsd xmm1, xmm2 mulsd xmm1, xmm2 ;# xmm1=rinvsix movapd xmm2, xmm1 mulsd xmm2, xmm2 ;# xmm2=rinvtwelve mulsd xmm1, [esp + nb312_c6] mulsd xmm2, [esp + nb312_c12] movapd xmm4, xmm2 subsd xmm4, xmm1 addsd xmm4, [esp + nb312_Vvdwtot] mulsd xmm1, [esp + nb312_six] mulsd xmm2, [esp + nb312_twelve] movlpd [esp + nb312_Vvdwtot], xmm4 subsd xmm2, xmm1 mulsd xmm2, xmm0 subsd xmm2, xmm3 mulsd xmm0, xmm2 movapd xmm1, xmm0 movapd xmm2, xmm0 xorpd xmm3, xmm3 movapd xmm4, xmm3 movapd xmm5, xmm3 mulpd xmm0, [esp + nb312_dxOO] mulpd xmm1, [esp + nb312_dyOO] mulpd xmm2, [esp + nb312_dzOO] subsd xmm3, xmm0 subsd xmm4, xmm1 subsd xmm5, xmm2 addsd xmm0, [esp + nb312_fixO] addsd xmm1, [esp + nb312_fiyO] addsd xmm2, [esp + nb312_fizO] movlpd [esp + nb312_fjxO], xmm3 movlpd [esp + nb312_fjyO], xmm4 movlpd [esp + nb312_fjzO], xmm5 movlpd [esp + nb312_fixO], xmm0 movlpd [esp + nb312_fiyO], xmm1 movlpd [esp + nb312_fizO], xmm2 ;# O-H1 interaction movapd xmm0, [esp + nb312_rinvOH1] movapd xmm1, xmm0 mulsd xmm1, [esp + nb312_rsqOH1] ;# xmm1=r mulsd xmm1, [esp + nb312_tsc] cvttsd2si eax, xmm1 ;# mm6 = lu idx cvtsi2sd xmm6, eax subsd xmm1, xmm6 ;# xmm1=eps movapd xmm2, xmm1 mulsd xmm2, xmm2 ;# xmm2=eps2 shl eax, 2 ;# idx *= 4 mov esi, [ebp + nb312_VFtab] movsd xmm4, [esi + eax*8] ;# Y1 movsd xmm5, [esi + eax*8 + 8] ;# F1 movsd xmm6, [esi + eax*8 + 16] ;# G1 movsd xmm7, [esi + eax*8 + 24] ;# H1 ;# coulomb table ready, in xmm4-xmm7 mulsd xmm6, xmm1 ;# xmm6=Geps mulsd xmm7, xmm2 ;# xmm7=Heps2 addsd xmm5, xmm6 addsd xmm5, xmm7 ;# xmm5=Fp mulsd xmm7, [esp + nb312_two] ;# two*Heps2 movapd xmm3, [esp + nb312_qqOH] addsd xmm7, xmm6 addsd xmm7, xmm5 ;# xmm7=FF mulsd xmm5, xmm1 ;# xmm5=eps*Fp addsd xmm5, xmm4 ;# xmm5=VV mulsd xmm5, xmm3 ;# vcoul=qq*VV mulsd xmm3, xmm7 ;# fijC=FF*qq ;# at this point mm5 contains vcoul and xmm3 fijC addsd xmm5, [esp + nb312_vctot] movlpd [esp + nb312_vctot], xmm5 xorpd xmm1, xmm1 mulsd xmm3, [esp + nb312_tsc] mulsd xmm3, xmm0 subsd xmm1, xmm3 movapd xmm0, xmm1 movapd xmm2, xmm1 xorpd xmm3, xmm3 movapd xmm4, xmm3 movapd xmm5, xmm3 mulsd xmm0, [esp + nb312_dxOH1] mulsd xmm1, [esp + nb312_dyOH1] mulsd xmm2, [esp + nb312_dzOH1] subsd xmm3, xmm0 subsd xmm4, xmm1 subsd xmm5, xmm2 addsd xmm0, [esp + nb312_fixO] addsd xmm1, [esp + nb312_fiyO] addsd xmm2, [esp + nb312_fizO] movlpd [esp + nb312_fjxH1], xmm3 movlpd [esp + nb312_fjyH1], xmm4 movlpd [esp + nb312_fjzH1], xmm5 movlpd [esp + nb312_fixO], xmm0 movlpd [esp + nb312_fiyO], xmm1 movlpd [esp + nb312_fizO], xmm2 ;# O-H2 interaction movapd xmm0, [esp + nb312_rinvOH2] movapd xmm1, xmm0 mulsd xmm1, [esp + nb312_rsqOH2] ;# xmm1=r mulsd xmm1, [esp + nb312_tsc] cvttsd2si eax, xmm1 ;# mm6 = lu idx cvtsi2sd xmm6, eax subsd xmm1, xmm6 ;# xmm1=eps movapd xmm2, xmm1 mulsd xmm2, xmm2 ;# xmm2=eps2 shl eax, 2 ;# idx *= 4 mov esi, [ebp + nb312_VFtab] movsd xmm4, [esi + eax*8] ;# Y1 movsd xmm5, [esi + eax*8 + 8] ;# F1 movsd xmm6, [esi + eax*8 + 16] ;# G1 movsd xmm7, [esi + eax*8 + 24] ;# H1 ;# coulomb table ready, in xmm4-xmm7 mulsd xmm6, xmm1 ;# xmm6=Geps mulsd xmm7, xmm2 ;# xmm7=Heps2 addsd xmm5, xmm6 addsd xmm5, xmm7 ;# xmm5=Fp mulsd xmm7, [esp + nb312_two] ;# two*Heps2 movapd xmm3, [esp + nb312_qqOH] addsd xmm7, xmm6 addsd xmm7, xmm5 ;# xmm7=FF mulsd xmm5, xmm1 ;# xmm5=eps*Fp addsd xmm5, xmm4 ;# xmm5=VV mulsd xmm5, xmm3 ;# vcoul=qq*VV mulsd xmm3, xmm7 ;# fijC=FF*qq ;# at this point mm5 contains vcoul and xmm3 fijC addsd xmm5, [esp + nb312_vctot] movlpd [esp + nb312_vctot], xmm5 xorpd xmm1, xmm1 mulsd xmm3, [esp + nb312_tsc] mulsd xmm3, xmm0 subsd xmm1, xmm3 movapd xmm0, xmm1 movapd xmm2, xmm1 xorpd xmm3, xmm3 movapd xmm4, xmm3 movapd xmm5, xmm3 mulsd xmm0, [esp + nb312_dxOH2] mulsd xmm1, [esp + nb312_dyOH2] mulsd xmm2, [esp + nb312_dzOH2] subsd xmm3, xmm0 subsd xmm4, xmm1 subsd xmm5, xmm2 addsd xmm0, [esp + nb312_fixO] addsd xmm1, [esp + nb312_fiyO] addsd xmm2, [esp + nb312_fizO] movlpd [esp + nb312_fjxH2], xmm3 movlpd [esp + nb312_fjyH2], xmm4 movlpd [esp + nb312_fjzH2], xmm5 movlpd [esp + nb312_fixO], xmm0 movlpd [esp + nb312_fiyO], xmm1 movlpd [esp + nb312_fizO], xmm2 ;# H1-O interaction movapd xmm0, [esp + nb312_rinvH1O] movapd xmm1, xmm0 mulsd xmm1, [esp + nb312_rsqH1O] ;# xmm1=r mulsd xmm1, [esp + nb312_tsc] cvttsd2si eax, xmm1 ;# mm6 = lu idx cvtsi2sd xmm6, eax subsd xmm1, xmm6 ;# xmm1=eps movapd xmm2, xmm1 mulsd xmm2, xmm2 ;# xmm2=eps2 shl eax, 2 ;# idx *= 4 mov esi, [ebp + nb312_VFtab] movsd xmm4, [esi + eax*8] ;# Y1 movsd xmm5, [esi + eax*8 + 8] ;# F1 movsd xmm6, [esi + eax*8 + 16] ;# G1 movsd xmm7, [esi + eax*8 + 24] ;# H1 ;# coulomb table ready, in xmm4-xmm7 mulsd xmm6, xmm1 ;# xmm6=Geps mulsd xmm7, xmm2 ;# xmm7=Heps2 addsd xmm5, xmm6 addsd xmm5, xmm7 ;# xmm5=Fp mulsd xmm7, [esp + nb312_two] ;# two*Heps2 movapd xmm3, [esp + nb312_qqOH] addsd xmm7, xmm6 addsd xmm7, xmm5 ;# xmm7=FF mulsd xmm5, xmm1 ;# xmm5=eps*Fp addsd xmm5, xmm4 ;# xmm5=VV mulsd xmm5, xmm3 ;# vcoul=qq*VV mulsd xmm3, xmm7 ;# fijC=FF*qq ;# at this point mm5 contains vcoul and xmm3 fijC addsd xmm5, [esp + nb312_vctot] movlpd [esp + nb312_vctot], xmm5 xorpd xmm1, xmm1 mulsd xmm3, [esp + nb312_tsc] mulsd xmm3, xmm0 subsd xmm1, xmm3 movapd xmm0, xmm1 movapd xmm2, xmm1 movapd xmm3, [esp + nb312_fjxO] movapd xmm4, [esp + nb312_fjyO] movapd xmm5, [esp + nb312_fjzO] mulsd xmm0, [esp + nb312_dxH1O] mulsd xmm1, [esp + nb312_dyH1O] mulsd xmm2, [esp + nb312_dzH1O] subsd xmm3, xmm0 subsd xmm4, xmm1 subsd xmm5, xmm2 addsd xmm0, [esp + nb312_fixH1] addsd xmm1, [esp + nb312_fiyH1] addsd xmm2, [esp + nb312_fizH1] movlpd [esp + nb312_fjxO], xmm3 movlpd [esp + nb312_fjyO], xmm4 movlpd [esp + nb312_fjzO], xmm5 movlpd [esp + nb312_fixH1], xmm0 movlpd [esp + nb312_fiyH1], xmm1 movlpd [esp + nb312_fizH1], xmm2 ;# H1-H1 interaction movapd xmm0, [esp + nb312_rinvH1H1] movapd xmm1, xmm0 mulsd xmm1, [esp + nb312_rsqH1H1] ;# xmm1=r mulsd xmm1, [esp + nb312_tsc] cvttsd2si eax, xmm1 ;# mm6 = lu idx cvtsi2sd xmm6, eax subsd xmm1, xmm6 ;# xmm1=eps movapd xmm2, xmm1 mulsd xmm2, xmm2 ;# xmm2=eps2 shl eax, 2 ;# idx *= 4 mov esi, [ebp + nb312_VFtab] movsd xmm4, [esi + eax*8] ;# Y1 movsd xmm5, [esi + eax*8 + 8] ;# F1 movsd xmm6, [esi + eax*8 + 16] ;# G1 movsd xmm7, [esi + eax*8 + 24] ;# H1 ;# coulomb table ready, in xmm4-xmm7 mulsd xmm6, xmm1 ;# xmm6=Geps mulsd xmm7, xmm2 ;# xmm7=Heps2 addsd xmm5, xmm6 addsd xmm5, xmm7 ;# xmm5=Fp ⌨️ 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?