nb_kernel332_ia32_sse2.intel_syntax.s
来自「最著名最快的分子模拟软件」· S 代码 · 共 2,183 行 · 第 1/5 页
S
2,183 行
movapd xmm0, [esp + nb332_ixH2] movapd xmm1, [esp + nb332_iyH2] movapd xmm2, [esp + nb332_izH2] movapd xmm3, [esp + nb332_ixH2] movapd xmm4, [esp + nb332_iyH2] movapd xmm5, [esp + nb332_izH2] subsd xmm0, [esp + nb332_jxO] subsd xmm1, [esp + nb332_jyO] subsd xmm2, [esp + nb332_jzO] subsd xmm3, [esp + nb332_jxH1] subsd xmm4, [esp + nb332_jyH1] subsd xmm5, [esp + nb332_jzH1] movapd [esp + nb332_dxH2O], xmm0 movapd [esp + nb332_dyH2O], xmm1 movapd [esp + nb332_dzH2O], xmm2 mulsd xmm0, xmm0 mulsd xmm1, xmm1 mulsd xmm2, xmm2 movapd [esp + nb332_dxH2H1], xmm3 movapd [esp + nb332_dyH2H1], xmm4 movapd [esp + nb332_dzH2H1], xmm5 mulsd xmm3, xmm3 mulsd xmm4, xmm4 mulsd xmm5, xmm5 addsd xmm0, xmm1 addsd xmm0, xmm2 addsd xmm4, xmm3 addsd xmm4, xmm5 movapd [esp + nb332_rsqH2O], xmm0 movapd [esp + nb332_rsqH2H1], xmm4 movapd xmm0, [esp + nb332_ixH2] movapd xmm1, [esp + nb332_iyH2] movapd xmm2, [esp + nb332_izH2] subsd xmm0, [esp + nb332_jxH2] subsd xmm1, [esp + nb332_jyH2] subsd xmm2, [esp + nb332_jzH2] movapd [esp + nb332_dxH2H2], xmm0 movapd [esp + nb332_dyH2H2], xmm1 movapd [esp + nb332_dzH2H2], xmm2 mulsd xmm0, xmm0 mulsd xmm1, xmm1 mulsd xmm2, xmm2 addsd xmm0, xmm1 addsd xmm0, xmm2 movapd [esp + nb332_rsqH2H2], xmm0 ;# start doing invsqrt use rsq values in xmm0, xmm4 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 + nb332_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 + nb332_half] ;# iter1 mulsd xmm7, [esp + nb332_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 + nb332_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 + nb332_half] ;# rinv mulsd xmm5, [esp + nb332_half] ;# rinv movapd [esp + nb332_rinvH2H2], xmm1 movapd [esp + nb332_rinvH2H1], xmm5 movapd xmm0, [esp + nb332_rsqOO] movapd xmm4, [esp + nb332_rsqOH1] 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 + nb332_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 + nb332_half] ;# iter1 of mulsd xmm7, [esp + nb332_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 + nb332_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 + nb332_half] ;# rinv mulsd xmm5, [esp + nb332_half] ;# rinv movapd [esp + nb332_rinvOO], xmm1 movapd [esp + nb332_rinvOH1], xmm5 movapd xmm0, [esp + nb332_rsqOH2] movapd xmm4, [esp + nb332_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 + nb332_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 + nb332_half] ;# iter1 mulsd xmm7, [esp + nb332_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 + nb332_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 + nb332_half] ;# rinv mulsd xmm5, [esp + nb332_half] ;# rinv movapd [esp + nb332_rinvOH2], xmm1 movapd [esp + nb332_rinvH1O], xmm5 movapd xmm0, [esp + nb332_rsqH1H1] movapd xmm4, [esp + nb332_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 + nb332_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 + nb332_half] ;# iter1a mulsd xmm7, [esp + nb332_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 + nb332_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 + nb332_half] ;# rinv mulsd xmm5, [esp + nb332_half] ;# rinv movapd [esp + nb332_rinvH1H1], xmm1 movapd [esp + nb332_rinvH1H2], xmm5 movapd xmm0, [esp + nb332_rsqH2O] cvtsd2ss xmm1, xmm0 rsqrtss xmm1, xmm1 cvtss2sd xmm1, xmm1 movapd xmm2, xmm1 ;# copy of luA mulsd xmm1, xmm1 ;# luA*luA movapd xmm3, [esp + nb332_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 + nb332_half] ;# iter1 movapd xmm2, xmm3 ;# copy of luA mulsd xmm3, xmm3 ;# luA*luA movapd xmm1, [esp + nb332_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 + nb332_half] ;# rinv movapd [esp + nb332_rinvH2O], xmm1 ;# start with OO interaction movapd xmm0, [esp + nb332_rinvOO] movapd xmm1, xmm0 mulsd xmm1, [esp + nb332_rsqOO] ;# xmm1=r mulsd xmm1, [esp + nb332_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 + nb332_VFtab] lea eax, [eax + eax*2] ;# idx*=3 (12 total now) 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 + nb332_two] ;# two*Heps2 movapd xmm3, [esp + nb332_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 + nb332_vctot] movlpd [esp + nb332_vctot], xmm5 ;# put scalar force on stack temporarily movapd [esp + nb332_fscal], xmm3 ;# Dispersion movsd xmm4, [esi + eax*8 + 32] ;# Y1 movsd xmm5, [esi + eax*8 + 40] ;# F1 movsd xmm6, [esi + eax*8 + 48] ;# G1 movsd xmm7, [esi + eax*8 + 56] ;# H1 ;# Dispersion 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 + nb332_two] ;# two*Heps2 addsd xmm7, xmm6 addsd xmm7, xmm5 ;# xmm7=FF mulsd xmm5, xmm1 ;# xmm5=eps*Fp addsd xmm5, xmm4 ;# xmm5=VV movapd xmm4, [esp + nb332_c6] mulsd xmm7, xmm4 ;# fijD mulsd xmm5, xmm4 ;# Vvdw6 addsd xmm7, [esp + nb332_fscal] ;# add to fscal ;# put scalar force back on stack Update Vvdwtot directly addsd xmm5, [esp + nb332_Vvdwtot] movapd [esp + nb332_fscal], xmm7 movlpd [esp + nb332_Vvdwtot], xmm5 ;# Repulsion movsd xmm4, [esi + eax*8 + 64] ;# Y1 movsd xmm5, [esi + eax*8 + 72] ;# F1 movsd xmm6, [esi + eax*8 + 80] ;# G1 movsd xmm7, [esi + eax*8 + 88] ;# H1 ;# Repulsion 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 + nb332_two] ;# two*Heps2 addsd xmm7, xmm6 addsd xmm7, xmm5 ;# xmm7=FF mulsd xmm5, xmm1 ;# xmm5=eps*Fp addsd xmm5, xmm4 ;# xmm5=VV movapd xmm4, [esp + nb332_c12] mulsd xmm7, xmm4 ;# fijR mulsd xmm5, xmm4 ;# Vvdw12 addsd xmm7, [esp + nb332_fscal] addsd xmm5, [esp + nb332_Vvdwtot] movlpd [esp + nb332_Vvdwtot], xmm5 xorpd xmm4, xmm4 mulsd xmm7, [esp + nb332_tsc] mulsd xmm7, [esp + nb332_rinvOO] subsd xmm4, xmm7 movapd xmm0, xmm4 movapd xmm1, xmm4 movapd xmm2, xmm4 xorpd xmm3, xmm3 movapd xmm4, xmm3 movapd xmm5, xmm3 mulpd xmm0, [esp + nb332_dxOO] mulpd xmm1, [esp + nb332_dyOO] mulpd xmm2, [esp + nb332_dzOO] subsd xmm3, xmm0 subsd xmm4, xmm1 subsd xmm5, xmm2 addsd xmm0, [esp + nb332_fixO] addsd xmm1, [esp + nb332_fiyO] addsd xmm2, [esp + nb332_fizO] movlpd [esp + nb332_fjxO], xmm3 movlpd [esp + nb332_fjyO], xmm4 movlpd [esp + nb332_fjzO], xmm5 movlpd [esp + nb332_fixO], xmm0 movlpd [esp + nb332_fiyO], xmm1 movlpd [esp + nb332_fizO], xmm2 ;# O-H1 interaction movapd xmm0, [esp + nb332_rinvOH1] movapd xmm1, xmm0 mulsd xmm1, [esp + nb332_rsqOH1] ;# xmm1=r mulsd xmm1, [esp + nb332_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 + nb332_VFtab] lea eax, [eax + eax*2] ;# idx*=3 (12 total now) 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 + nb332_two] ;# two*Heps2 movapd xmm3, [esp + nb332_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 + nb332_vctot] movlpd [esp + nb332_vctot], xmm5 xorpd xmm1, xmm1 mulsd xmm3, [esp + nb332_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 + nb332_dxOH1] mulsd xmm1, [esp + nb332_dyOH1] mulsd xmm2, [esp + nb332_dzOH1] subsd xmm3, xmm0 subsd xmm4, xmm1 subsd xmm5, xmm2 addsd xmm0, [esp + nb332_fixO] addsd xmm1, [esp + nb332_fiyO] addsd xmm2, [esp + nb332_fizO] movlpd [esp + nb332_fjxH1], xmm3 movlpd [esp + nb332_fjyH1], xmm4 movlpd [esp + nb332_fjzH1], xmm5 movlpd [esp + nb332_fixO], xmm0 movlpd [esp + nb332_fiyO], xmm1 movlpd [esp + nb332_fizO], xmm2 ;# O-H2 interaction movapd xmm0, [esp + nb332_rinvOH2] movapd xmm1, xmm0 mulsd xmm1, [esp + nb332_rsqOH2] ;# xmm1=r mulsd xmm1, [esp + nb332_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 + nb332_VFtab] lea eax, [eax + eax*2] ;# idx*=3⌨️ 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?