📄 nb_kernel302_ia32_3dnow.s
字号:
#### $Id: nb_kernel302_ia32_3dnow.s,v 1.1.2.2 2007/11/28 09:34:39 spoel Exp $#### Gromacs 4.0 Copyright (c) 1991-2003 ## David van der Spoel, Erik Lindahl#### This program is free software; you can redistribute it and/or## modify it under the terms of the GNU General Public License## as published by the Free Software Foundation; either version 2## of the License, or (at your option) any later version.#### To help us fund GROMACS development, we humbly ask that you cite## the research papers on the package. Check out http://www.gromacs.org## ## And Hey:## Gnomes, ROck Monsters And Chili Sauce##.globl nb_kernel302_ia32_3dnow.globl _nb_kernel302_ia32_3dnownb_kernel302_ia32_3dnow: _nb_kernel302_ia32_3dnow: .set nb302_p_nri, 8.set nb302_iinr, 12.set nb302_jindex, 16.set nb302_jjnr, 20.set nb302_shift, 24.set nb302_shiftvec, 28.set nb302_fshift, 32.set nb302_gid, 36.set nb302_pos, 40.set nb302_faction, 44.set nb302_charge, 48.set nb302_p_facel, 52.set nb302_p_krf, 56.set nb302_p_crf, 60.set nb302_Vc, 64.set nb302_type, 68.set nb302_p_ntype, 72.set nb302_vdwparam, 76.set nb302_Vvdw, 80.set nb302_p_tabscale, 84.set nb302_VFtab, 88.set nb302_invsqrta, 92.set nb302_dvda, 96.set nb302_p_gbtabscale, 100.set nb302_GBtab, 104.set nb302_p_nthreads, 108.set nb302_count, 112.set nb302_mtx, 116.set nb302_outeriter, 120.set nb302_inneriter, 124.set nb302_work, 128 ## stack offsets for local variables .set nb302_is3, 0.set nb302_ii3, 4.set nb302_ixO, 8.set nb302_iyO, 12.set nb302_izO, 16.set nb302_ixH, 20.set nb302_iyH, 28.set nb302_izH, 36.set nb302_qqOO, 44.set nb302_qqOH, 52.set nb302_qqHH, 60.set nb302_two, 68.set nb302_n1, 76.set nb302_tsc, 84.set nb302_vctot, 92.set nb302_innerjjnr, 100.set nb302_innerk, 104.set nb302_fixO, 108.set nb302_fiyO, 112.set nb302_fizO, 116.set nb302_fixH, 120.set nb302_fiyH, 128.set nb302_fizH, 136.set nb302_dxO, 144.set nb302_dyO, 148.set nb302_dzO, 152.set nb302_dxH, 156.set nb302_dyH, 164.set nb302_dzH, 172.set nb302_tmprsqH, 180.set nb302_n, 188 ## idx for outer loop.set nb302_nn1, 192 ## number of outer iterations.set nb302_nri, 196.set nb302_nouter, 200.set nb302_ninner, 204 pushl %ebp movl %esp,%ebp pushl %eax pushl %ebx pushl %ecx pushl %edx pushl %esi pushl %edi subl $208,%esp ## local stack space femms movl nb302_p_nri(%ebp),%ecx movl nb302_p_facel(%ebp),%esi movl nb302_p_tabscale(%ebp),%edi movl (%ecx),%ecx movl %ecx,nb302_nri(%esp) ## zero iteration counters movl $0,%eax movl %eax,nb302_nouter(%esp) movl %eax,nb302_ninner(%esp) ## assume we have at least one i particle - start directly movl nb302_iinr(%ebp),%ecx ## ecx = pointer into iinr[] movl (%ecx),%ebx ## ebx=ii movl nb302_charge(%ebp),%edx movd (%esi),%mm1 ## mm1=facel movd (%edx,%ebx,4),%mm2 ## mm2=charge[ii0] (O) movd 4(%edx,%ebx,4),%mm3 ## mm2=charge[ii0+1] (H) movq %mm2,%mm4 pfmul %mm1,%mm4 movq %mm3,%mm6 pfmul %mm1,%mm6 movq %mm4,%mm5 pfmul %mm2,%mm4 ## mm4=qqOO*facel pfmul %mm3,%mm5 ## mm5=qqOH*facel pfmul %mm3,%mm6 ## mm6=qqHH*facel punpckldq %mm5,%mm5 ## spread to both halves punpckldq %mm6,%mm6 ## spread to both halves movq %mm4,nb302_qqOO(%esp) movq %mm5,nb302_qqOH(%esp) movq %mm6,nb302_qqHH(%esp) movd (%edi),%mm3 punpckldq %mm3,%mm3 movq %mm3,nb302_tsc(%esp) movl $0x40000000,%eax movl %eax,nb302_two(%esp) movl %eax,nb302_two+4(%esp)_nb_kernel302_ia32_3dnow.nb302_threadloop: movl nb302_count(%ebp),%esi ## pointer to sync counter movl (%esi),%eax_nb_kernel302_ia32_3dnow.nb302_spinlock: movl %eax,%ebx ## ebx=*count=nn0 addl $1,%ebx ## ebx=nn1=nn0+10 lock cmpxchgl %ebx,(%esi) ## write nn1 to *counter, ## if it hasnt changed. ## or reread *counter to eax. pause ## -> better p4 performance jnz _nb_kernel302_ia32_3dnow.nb302_spinlock ## if(nn1>nri) nn1=nri movl nb302_nri(%esp),%ecx movl %ecx,%edx subl %ebx,%ecx cmovlel %edx,%ebx ## if(nn1>nri) nn1=nri ## Cleared the spinlock if we got here. ## eax contains nn0, ebx contains nn1. movl %eax,nb302_n(%esp) movl %ebx,nb302_nn1(%esp) subl %eax,%ebx ## calc number of outer lists movl %eax,%esi ## copy n to esi jg _nb_kernel302_ia32_3dnow.nb302_outerstart jmp _nb_kernel302_ia32_3dnow.nb302_end_nb_kernel302_ia32_3dnow.nb302_outerstart: ## ebx contains number of outer iterations addl nb302_nouter(%esp),%ebx movl %ebx,nb302_nouter(%esp)_nb_kernel302_ia32_3dnow.nb302_outer: movl nb302_shift(%ebp),%eax ## eax = pointer into shift[] movl (%eax,%esi,4),%ebx ## ebx=shift[n] leal (%ebx,%ebx,2),%ebx ## ebx=3*is movl %ebx,nb302_is3(%esp) ## store is3 movl nb302_shiftvec(%ebp),%eax ## eax = base of shiftvec[] movq (%eax,%ebx,4),%mm5 ## move shX/shY to mm5 and shZ to mm6. movd 8(%eax,%ebx,4),%mm6 movq %mm5,%mm0 movq %mm5,%mm1 movq %mm6,%mm2 punpckldq %mm0,%mm0 ## also expand shX,Y,Z in mm0--mm2. punpckhdq %mm1,%mm1 punpckldq %mm2,%mm2 movl nb302_iinr(%ebp),%ecx ## ecx = pointer into iinr[] movl (%ecx,%esi,4),%ebx ## ebx=ii leal (%ebx,%ebx,2),%ebx ## ebx = 3*ii=ii3 movl nb302_pos(%ebp),%eax ## eax = base of pos[] pfadd (%eax,%ebx,4),%mm5 ## ix = shX + posX (and iy too) movd 8(%eax,%ebx,4),%mm7 ## cant use direct memory add for 4 bytes (iz) movl %ebx,nb302_ii3(%esp) ## (use mm7 as temp. storage for iz.) pfadd %mm7,%mm6 movq %mm5,nb302_ixO(%esp) movq %mm6,nb302_izO(%esp) movd 12(%eax,%ebx,4),%mm3 movd 16(%eax,%ebx,4),%mm4 movd 20(%eax,%ebx,4),%mm5 punpckldq 24(%eax,%ebx,4),%mm3 punpckldq 28(%eax,%ebx,4),%mm4 punpckldq 32(%eax,%ebx,4),%mm5 ## coords of H1 in low mm3-mm5, H2 in high pfadd %mm3,%mm0 pfadd %mm4,%mm1 pfadd %mm5,%mm2 movq %mm0,nb302_ixH(%esp) movq %mm1,nb302_iyH(%esp) movq %mm2,nb302_izH(%esp) ## clear vctot and i forces pxor %mm7,%mm7 movq %mm7,nb302_vctot(%esp) movq %mm7,nb302_fixO(%esp) movq %mm7,nb302_fizO(%esp) movq %mm7,nb302_fixH(%esp) movq %mm7,nb302_fiyH(%esp) movq %mm7,nb302_fizH(%esp) movl nb302_jindex(%ebp),%eax movl (%eax,%esi,4),%ecx ## jindex[n] movl 4(%eax,%esi,4),%edx ## jindex[n+1] subl %ecx,%edx ## number of innerloop atoms movl %edx,nb302_innerk(%esp) ## number of innerloop atoms addl nb302_ninner(%esp),%edx movl %edx,nb302_ninner(%esp) movl nb302_pos(%ebp),%esi movl nb302_faction(%ebp),%edi movl nb302_jjnr(%ebp),%eax shll $2,%ecx addl %ecx,%eax movl %eax,nb302_innerjjnr(%esp) ## pointer to jjnr[nj0] _nb_kernel302_ia32_3dnow.nb302_inner_loop: ## a single j particle iteration here - compare with the unrolled code for comments. movl nb302_innerjjnr(%esp),%eax movl (%eax),%eax ## eax=jnr offset addl $4,nb302_innerjjnr(%esp) ## advance pointer leal (%eax,%eax,2),%eax movq (%esi,%eax,4),%mm0 movd 8(%esi,%eax,4),%mm1 ## copy & expand to mm2-mm4 for the H interactions movq %mm0,%mm2 movq %mm0,%mm3 movq %mm1,%mm4 punpckldq %mm2,%mm2 punpckhdq %mm3,%mm3 punpckldq %mm4,%mm4 pfsubr nb302_ixO(%esp),%mm0 pfsubr nb302_izO(%esp),%mm1 movq %mm0,nb302_dxO(%esp) pfmul %mm0,%mm0 movd %mm1,nb302_dzO(%esp) pfmul %mm1,%mm1 pfacc %mm0,%mm0 pfadd %mm1,%mm0 ## mm0=rsqO punpckldq %mm2,%mm2 punpckldq %mm3,%mm3 punpckldq %mm4,%mm4 ## mm2-mm4 is jx-jz pfsubr nb302_ixH(%esp),%mm2 pfsubr nb302_iyH(%esp),%mm3 pfsubr nb302_izH(%esp),%mm4 ## mm2-mm4 is dxH-dzH movq %mm2,nb302_dxH(%esp) movq %mm3,nb302_dyH(%esp) movq %mm4,nb302_dzH(%esp) pfmul %mm2,%mm2 pfmul %mm3,%mm3 pfmul %mm4,%mm4 pfadd %mm2,%mm3 pfadd %mm4,%mm3 ## mm3=rsqH movq %mm3,nb302_tmprsqH(%esp) pfrsqrt %mm0,%mm1 movq %mm1,%mm2 pfmul %mm1,%mm1 pfrsqit1 %mm0,%mm1 pfrcpit2 %mm2,%mm1 ## mm1=invsqrt pfmul %mm1,%mm0 ## mm0=rsq pfmul nb302_tsc(%esp),%mm0 pf2iw %mm0,%mm4 movd %mm4,nb302_n1(%esp) pi2fd %mm4,%mm4 pfsub %mm4,%mm0 ## now mm0 is eps and mm4 n0 movq %mm0,%mm2 pfmul %mm2,%mm2 ## mm0 is eps, mm2 eps2 ## coulomb table movl nb302_VFtab(%ebp),%edx movl nb302_n1(%esp),%ecx shll $2,%ecx ## load all values we need movd (%edx,%ecx,4),%mm4 movd 4(%edx,%ecx,4),%mm5 movd 8(%edx,%ecx,4),%mm6 movd 12(%edx,%ecx,4),%mm7 pfmul %mm0,%mm6 ## mm6 = Geps pfmul %mm2,%mm7 ## mm7 = Heps2 pfadd %mm6,%mm5 pfadd %mm7,%mm5 ## mm5 = Fp pfmul nb302_two(%esp),%mm7 ## two*Heps2 pfadd %mm6,%mm7 pfadd %mm5,%mm7 ## mm7=FF pfmul %mm0,%mm5 ## mm5=eps*Fp pfadd %mm4,%mm5 ## mm5= VV pfmul nb302_qqOO(%esp),%mm5 ## vcoul=qq*VV pfmul nb302_qqOO(%esp),%mm7 ## fijC=qq*FF ## update vctot directly, use mm3 for fscal sum. pfadd nb302_vctot(%esp),%mm5 movq %mm5,nb302_vctot(%esp) movq %mm7,%mm3 ## change sign of fscal and multiply with rinv pxor %mm0,%mm0 pfsubr %mm0,%mm3 pfmul nb302_tsc(%esp),%mm3 pfmul %mm1,%mm3 ## mm3 is total fscal (for the oxygen) now ## Ready with the oxygen - potential is updated, fscal is in mm3. ## time for hydrogens! movq nb302_tmprsqH(%esp),%mm0 pfrsqrt %mm0,%mm1 pswapd %mm0,%mm0 pfrsqrt %mm0,%mm2 pswapd %mm0,%mm0 punpckldq %mm2,%mm1 ## seeds are in mm1 now, and rsq in mm0. movq %mm1,%mm2 pfmul %mm1,%mm1 pfrsqit1 %mm0,%mm1 pfrcpit2 %mm2,%mm1 ## mm1=invsqrt pfmul %mm1,%mm0 ## mm0=r pfmul nb302_tsc(%esp),%mm0 pf2iw %mm0,%mm4 movq %mm4,nb302_n1(%esp) pi2fd %mm4,%mm4 pfsub %mm4,%mm0 ## now mm0 is eps and mm4 n0 movq %mm0,%mm2 pfmul %mm2,%mm2 ## mm0 is eps, mm2 eps2 ## coulomb table movl nb302_VFtab(%ebp),%edx movl nb302_n1(%esp),%ecx shll $2,%ecx ## load all values we need movd (%edx,%ecx,4),%mm4 movd 4(%edx,%ecx,4),%mm5 movd 8(%edx,%ecx,4),%mm6 movd 12(%edx,%ecx,4),%mm7 movl nb302_n1+4(%esp),%ecx shll $2,%ecx punpckldq (%edx,%ecx,4),%mm4 punpckldq 4(%edx,%ecx,4),%mm5 punpckldq 8(%edx,%ecx,4),%mm6 punpckldq 12(%edx,%ecx,4),%mm7 pfmul %mm0,%mm6 ## mm6 = Geps pfmul %mm2,%mm7 ## mm7 = Heps2 pfadd %mm6,%mm5 pfadd %mm7,%mm5 ## mm5 = Fp pfmul nb302_two(%esp),%mm7 ## two*Heps2 pfadd %mm6,%mm7 pfadd %mm5,%mm7 ## mm7=FF pfmul %mm0,%mm5 ## mm5=eps*Fp pfadd %mm4,%mm5 ## mm5= VV pfmul nb302_qqOH(%esp),%mm5 ## vcoul=qq*VV pfmul nb302_qqOH(%esp),%mm7 ## fijC=qq*FF ## update vctot pfadd nb302_vctot(%esp),%mm5 movq %mm5,nb302_vctot(%esp) ## change sign of fijC and multiply by rinv pxor %mm4,%mm4 pfsub %mm7,%mm4 pfmul nb302_tsc(%esp),%mm4 pfmul %mm1,%mm4 ## mm4 is total fscal (for the hydrogens) now ⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -