2104 lines
68 KiB
ArmAsm
2104 lines
68 KiB
ArmAsm
.text
|
|
.file "force_lj.c"
|
|
.section .rodata.cst8,"aM",@progbits,8
|
|
.p2align 3 # -- Begin function computeForceLJ_ref
|
|
.LCPI0_0:
|
|
.quad 0x4048000000000000 # 48
|
|
.LCPI0_2:
|
|
.quad 0xbfe0000000000000 # -0.5
|
|
.LCPI0_3:
|
|
.quad 0x3fe0000000000000 # 0.5
|
|
.section .rodata.cst4,"aM",@progbits,4
|
|
.p2align 2
|
|
.LCPI0_1:
|
|
.long 0x3f800000 # 1
|
|
.text
|
|
.globl computeForceLJ_ref
|
|
.p2align 4, 0x90
|
|
.type computeForceLJ_ref,@function
|
|
computeForceLJ_ref: #
|
|
.cfi_startproc
|
|
# %bb.0:
|
|
pushq %rbp
|
|
.cfi_def_cfa_offset 16
|
|
pushq %r15
|
|
.cfi_def_cfa_offset 24
|
|
pushq %r14
|
|
.cfi_def_cfa_offset 32
|
|
pushq %r13
|
|
.cfi_def_cfa_offset 40
|
|
pushq %r12
|
|
.cfi_def_cfa_offset 48
|
|
pushq %rbx
|
|
.cfi_def_cfa_offset 56
|
|
subq $152, %rsp
|
|
.cfi_def_cfa_offset 208
|
|
.cfi_offset %rbx, -56
|
|
.cfi_offset %r12, -48
|
|
.cfi_offset %r13, -40
|
|
.cfi_offset %r14, -32
|
|
.cfi_offset %r15, -24
|
|
.cfi_offset %rbp, -16
|
|
movq %rcx, 16(%rsp) # 8-byte Spill
|
|
movq %rdx, 32(%rsp) # 8-byte Spill
|
|
movq %rsi, %r14
|
|
movq %rdi, %r12
|
|
movl $.L.str, %edi
|
|
xorl %eax, %eax
|
|
callq debug_printf
|
|
vmovss 108(%r12), %xmm0 # xmm0 = mem[0],zero,zero,zero
|
|
vmovss %xmm0, 8(%rsp) # 4-byte Spill
|
|
vmovss 40(%r12), %xmm0 # xmm0 = mem[0],zero,zero,zero
|
|
vmovss %xmm0, (%rsp) # 4-byte Spill
|
|
vmovss 48(%r12), %xmm2 # xmm2 = mem[0],zero,zero,zero
|
|
movl 20(%r14), %r11d
|
|
testl %r11d, %r11d
|
|
jle .LBB0_5
|
|
# %bb.1:
|
|
movq 176(%r14), %r9
|
|
movq 192(%r14), %r10
|
|
decq %r11
|
|
leaq 64(%r9), %r8
|
|
xorl %edi, %edi
|
|
vxorps %xmm0, %xmm0, %xmm0
|
|
vxorps %xmm1, %xmm1, %xmm1
|
|
jmp .LBB0_2
|
|
.p2align 5, 0x90
|
|
.LBB0_79: # in Loop: Header=BB0_2 Depth=1
|
|
cmpq %r11, %rdi
|
|
leaq 1(%rdi), %rdi
|
|
je .LBB0_5
|
|
.LBB0_2: # =>This Loop Header: Depth=1
|
|
# Child Loop BB0_74 Depth 2
|
|
# Child Loop BB0_78 Depth 2
|
|
leaq (%rdi,%rdi,8), %rax
|
|
leaq (%rax,%rax,2), %rax
|
|
addq %rdi, %rax
|
|
movl (%r10,%rax), %esi
|
|
testl %esi, %esi
|
|
jle .LBB0_79
|
|
# %bb.3: # in Loop: Header=BB0_2 Depth=1
|
|
leal (,%rdi,4), %ebx
|
|
movl %ebx, %eax
|
|
andl $2147483640, %eax # imm = 0x7FFFFFF8
|
|
leal (%rax,%rax,2), %eax
|
|
andl $4, %ebx
|
|
orl %eax, %ebx
|
|
cmpl $7, %esi
|
|
ja .LBB0_73
|
|
# %bb.4: # in Loop: Header=BB0_2 Depth=1
|
|
movl %esi, %ebp
|
|
andl $-8, %ebp
|
|
cmpq %rsi, %rbp
|
|
jae .LBB0_79
|
|
jmp .LBB0_77
|
|
.p2align 5, 0x90
|
|
.LBB0_73: # in Loop: Header=BB0_2 Depth=1
|
|
leaq (,%rsi,4), %rbp
|
|
andq $-32, %rbp
|
|
movl %ebx, %ecx
|
|
leaq (%r9,%rcx,4), %rdx
|
|
xorl %eax, %eax
|
|
.p2align 4, 0x90
|
|
.LBB0_74: # Parent Loop BB0_2 Depth=1
|
|
# => This Inner Loop Header: Depth=2
|
|
vmovups %ymm0, (%rdx,%rax)
|
|
addq $32, %rax
|
|
cmpq %rax, %rbp
|
|
jne .LBB0_74
|
|
# %bb.75: # in Loop: Header=BB0_2 Depth=1
|
|
movl %esi, %ebp
|
|
andl $-8, %ebp
|
|
addq %rbp, %rcx
|
|
vmovups %zmm1, (%r9,%rcx,4)
|
|
cmpq %rsi, %rbp
|
|
jae .LBB0_79
|
|
.LBB0_77: # in Loop: Header=BB0_2 Depth=1
|
|
movl %ebx, %eax
|
|
leaq (%r8,%rax,4), %rcx
|
|
.p2align 4, 0x90
|
|
.LBB0_78: # Parent Loop BB0_2 Depth=1
|
|
# => This Inner Loop Header: Depth=2
|
|
movl $0, -64(%rcx,%rbp,4)
|
|
movl $0, -32(%rcx,%rbp,4)
|
|
movl $0, (%rcx,%rbp,4)
|
|
incq %rbp
|
|
cmpq %rbp, %rsi
|
|
jne .LBB0_78
|
|
jmp .LBB0_79
|
|
.p2align 5, 0x90
|
|
.LBB0_5:
|
|
vmovss %xmm2, 28(%rsp) # 4-byte Spill
|
|
xorl %eax, %eax
|
|
vzeroupper
|
|
callq getTimeStamp
|
|
vmovsd %xmm0, 56(%rsp) # 8-byte Spill
|
|
movl $.L.str.1, %edi
|
|
callq likwid_markerStartRegion
|
|
movl 20(%r14), %esi
|
|
testl %esi, %esi
|
|
jle .LBB0_17
|
|
# %bb.6:
|
|
vmovss 8(%rsp), %xmm0 # 4-byte Reload
|
|
# xmm0 = mem[0],zero,zero,zero
|
|
vmulss %xmm0, %xmm0, %xmm13
|
|
movq 16(%rsp), %rax # 8-byte Reload
|
|
leaq 32(%rax), %r15
|
|
vmovss (%rsp), %xmm0 # 4-byte Reload
|
|
# xmm0 = mem[0],zero,zero,zero
|
|
vcvtss2sd %xmm0, %xmm0, %xmm0
|
|
vmulsd .LCPI0_0(%rip), %xmm0, %xmm12
|
|
leaq 24(%rax), %rdx
|
|
movq 160(%r14), %rdi
|
|
movq 176(%r14), %rbp
|
|
movq 32(%rsp), %rcx # 8-byte Reload
|
|
movq 8(%rcx), %rbx
|
|
movq %rbx, 72(%rsp) # 8-byte Spill
|
|
movq 24(%rcx), %rbx
|
|
movq %rbx, 96(%rsp) # 8-byte Spill
|
|
movslq 16(%rcx), %rcx
|
|
movq %rcx, 64(%rsp) # 8-byte Spill
|
|
vmovdqu 8(%rax), %xmm9
|
|
leal -1(%rsi), %ecx
|
|
addq (%rax), %rcx
|
|
movq %rcx, 48(%rsp) # 8-byte Spill
|
|
movq %rbp, 80(%rsp) # 8-byte Spill
|
|
leaq 64(%rbp), %rax
|
|
movq %rax, 128(%rsp) # 8-byte Spill
|
|
movq %rdi, 40(%rsp) # 8-byte Spill
|
|
leaq 64(%rdi), %rax
|
|
movq %rax, 120(%rsp) # 8-byte Spill
|
|
xorl %edi, %edi
|
|
vmovss .LCPI0_1(%rip), %xmm10 # xmm10 = mem[0],zero,zero,zero
|
|
vmovsd .LCPI0_2(%rip), %xmm11 # xmm11 = mem[0],zero
|
|
vmovsd .LCPI0_3(%rip), %xmm8 # xmm8 = mem[0],zero
|
|
vmovss 28(%rsp), %xmm20 # 4-byte Reload
|
|
# xmm20 = mem[0],zero,zero,zero
|
|
movq %rsi, 88(%rsp) # 8-byte Spill
|
|
jmp .LBB0_7
|
|
.p2align 5, 0x90
|
|
.LBB0_19: # in Loop: Header=BB0_7 Depth=1
|
|
movq 88(%rsp), %rsi # 8-byte Reload
|
|
movq 112(%rsp), %rdi # 8-byte Reload
|
|
movq 104(%rsp), %rbp # 8-byte Reload
|
|
.LBB0_20: # in Loop: Header=BB0_7 Depth=1
|
|
vcvtsi2sd %ebp, %xmm21, %xmm0
|
|
vmulsd %xmm0, %xmm8, %xmm0
|
|
vcvttsd2si %xmm0, %rax
|
|
vmovq %rax, %xmm0
|
|
vmovq %rcx, %xmm1
|
|
vpunpcklqdq %xmm0, %xmm1, %xmm0 # xmm0 = xmm1[0],xmm0[0]
|
|
vpaddq %xmm0, %xmm9, %xmm9
|
|
incq %rdi
|
|
cmpq %rsi, %rdi
|
|
je .LBB0_16
|
|
.LBB0_7: # =>This Loop Header: Depth=1
|
|
# Child Loop BB0_9 Depth 2
|
|
# Child Loop BB0_10 Depth 3
|
|
# Child Loop BB0_12 Depth 4
|
|
movq 96(%rsp), %rax # 8-byte Reload
|
|
movslq (%rax,%rdi,4), %rbp
|
|
movq %rbp, %rcx
|
|
testq %rbp, %rbp
|
|
jle .LBB0_20
|
|
# %bb.8: # in Loop: Header=BB0_7 Depth=1
|
|
movl %edi, %r13d
|
|
shrl %r13d
|
|
leal (,%r13,8), %eax
|
|
leal (%rax,%rax,2), %eax
|
|
leal (,%rdi,4), %ecx
|
|
andl $4, %ecx
|
|
orl %ecx, %eax
|
|
movq 40(%rsp), %rsi # 8-byte Reload
|
|
leaq (%rsi,%rax,4), %r8
|
|
movq 80(%rsp), %rsi # 8-byte Reload
|
|
leaq (%rsi,%rax,4), %r11
|
|
movq %rdi, 112(%rsp) # 8-byte Spill
|
|
movq %rdi, %rax
|
|
imulq 64(%rsp), %rax # 8-byte Folded Reload
|
|
movq 72(%rsp), %rsi # 8-byte Reload
|
|
leaq (%rsi,%rax,4), %rax
|
|
movq %rax, 136(%rsp) # 8-byte Spill
|
|
movq 32(%rsp), %rax # 8-byte Reload
|
|
movl 32(%rax), %eax
|
|
movl %eax, (%rsp) # 4-byte Spill
|
|
movl %ecx, %r12d
|
|
movq %rbp, 104(%rsp) # 8-byte Spill
|
|
movl %ebp, %ecx
|
|
xorl %esi, %esi
|
|
movq %rcx, 144(%rsp) # 8-byte Spill
|
|
jmp .LBB0_9
|
|
.p2align 5, 0x90
|
|
.LBB0_18: # in Loop: Header=BB0_9 Depth=2
|
|
movq 8(%rsp), %rsi # 8-byte Reload
|
|
incq %rsi
|
|
movq 144(%rsp), %rcx # 8-byte Reload
|
|
cmpq %rcx, %rsi
|
|
je .LBB0_19
|
|
.LBB0_9: # Parent Loop BB0_7 Depth=1
|
|
# => This Loop Header: Depth=2
|
|
# Child Loop BB0_10 Depth 3
|
|
# Child Loop BB0_12 Depth 4
|
|
movq 136(%rsp), %rax # 8-byte Reload
|
|
movq %rsi, 8(%rsp) # 8-byte Spill
|
|
movslq (%rax,%rsi,4), %r10
|
|
movq %r10, %rax
|
|
shlq $5, %rax
|
|
leaq (%rax,%rax,2), %rdi
|
|
movq 40(%rsp), %rax # 8-byte Reload
|
|
addq %rdi, %rax
|
|
movq 128(%rsp), %rcx # 8-byte Reload
|
|
leaq (%rcx,%rdi), %rsi
|
|
addq 120(%rsp), %rdi # 8-byte Folded Reload
|
|
xorl %r9d, %r9d
|
|
xorl %r14d, %r14d
|
|
jmp .LBB0_10
|
|
.p2align 5, 0x90
|
|
.LBB0_67: # in Loop: Header=BB0_10 Depth=3
|
|
vdivss %xmm3, %xmm10, %xmm3
|
|
vmulss %xmm20, %xmm3, %xmm14
|
|
vmulss %xmm3, %xmm3, %xmm4
|
|
vmulss %xmm4, %xmm14, %xmm4
|
|
vcvtss2sd %xmm4, %xmm4, %xmm4
|
|
vaddsd %xmm4, %xmm11, %xmm14
|
|
vcvtss2sd %xmm3, %xmm3, %xmm3
|
|
vmulsd %xmm3, %xmm12, %xmm3
|
|
vmulsd %xmm4, %xmm14, %xmm4
|
|
vmulsd %xmm3, %xmm4, %xmm3
|
|
vcvtsd2ss %xmm3, %xmm3, %xmm3
|
|
vfmadd231ss %xmm2, %xmm3, %xmm7 # xmm7 = (xmm3 * xmm2) + xmm7
|
|
vfmadd231ss %xmm1, %xmm3, %xmm5 # xmm5 = (xmm3 * xmm1) + xmm5
|
|
vfmadd231ss %xmm0, %xmm3, %xmm19 # xmm19 = (xmm3 * xmm0) + xmm19
|
|
movl $1, %r14d
|
|
movq %rdx, %rbx
|
|
.LBB0_68: # in Loop: Header=BB0_10 Depth=3
|
|
incq (%rbx)
|
|
.p2align 4, 0x90
|
|
.LBB0_69: # in Loop: Header=BB0_10 Depth=3
|
|
xorl %ecx, %ecx
|
|
testl %r14d, %r14d
|
|
sete %cl
|
|
movq 16(%rsp), %rbx # 8-byte Reload
|
|
incq 40(%rbx,%rcx,8)
|
|
vaddss (%r11,%r9,4), %xmm7, %xmm0
|
|
vmovss %xmm0, (%r11,%r9,4)
|
|
vaddss 32(%r11,%r9,4), %xmm5, %xmm0
|
|
vmovss %xmm0, 32(%r11,%r9,4)
|
|
vaddss 64(%r11,%r9,4), %xmm19, %xmm0
|
|
vmovss %xmm0, 64(%r11,%r9,4)
|
|
incq %r9
|
|
cmpq $4, %r9
|
|
je .LBB0_18
|
|
.LBB0_10: # Parent Loop BB0_7 Depth=1
|
|
# Parent Loop BB0_9 Depth=2
|
|
# => This Loop Header: Depth=3
|
|
# Child Loop BB0_12 Depth 4
|
|
vmovss (%r8,%r9,4), %xmm14 # xmm14 = mem[0],zero,zero,zero
|
|
leaq (%r9,%r12), %rbp
|
|
vmovss 32(%r8,%r9,4), %xmm16 # xmm16 = mem[0],zero,zero,zero
|
|
vmovss 64(%r8,%r9,4), %xmm18 # xmm18 = mem[0],zero,zero,zero
|
|
cmpl $0, (%rsp) # 4-byte Folded Reload
|
|
je .LBB0_21
|
|
# %bb.11: # in Loop: Header=BB0_10 Depth=3
|
|
vxorps %xmm7, %xmm7, %xmm7
|
|
xorl %ebx, %ebx
|
|
vxorps %xmm5, %xmm5, %xmm5
|
|
vxorps %xmm19, %xmm19, %xmm19
|
|
jmp .LBB0_12
|
|
.p2align 5, 0x90
|
|
.LBB0_70: # in Loop: Header=BB0_12 Depth=4
|
|
vdivss %xmm3, %xmm10, %xmm3
|
|
vmulss %xmm20, %xmm3, %xmm4
|
|
vmulss %xmm3, %xmm3, %xmm6
|
|
vmulss %xmm4, %xmm6, %xmm4
|
|
vcvtss2sd %xmm4, %xmm4, %xmm4
|
|
vaddsd %xmm4, %xmm11, %xmm6
|
|
vcvtss2sd %xmm3, %xmm3, %xmm3
|
|
vmulsd %xmm3, %xmm12, %xmm3
|
|
vmulsd %xmm4, %xmm6, %xmm4
|
|
vmulsd %xmm3, %xmm4, %xmm3
|
|
vcvtsd2ss %xmm3, %xmm3, %xmm3
|
|
vmovss -32(%rsi,%rbx,4), %xmm4 # xmm4 = mem[0],zero,zero,zero
|
|
vmovss -64(%rsi,%rbx,4), %xmm6 # xmm6 = mem[0],zero,zero,zero
|
|
vfnmadd231ss %xmm2, %xmm3, %xmm6 # xmm6 = -(xmm3 * xmm2) + xmm6
|
|
vmovss %xmm6, -64(%rsi,%rbx,4)
|
|
vfnmadd231ss %xmm0, %xmm3, %xmm4 # xmm4 = -(xmm3 * xmm0) + xmm4
|
|
vmovss %xmm4, -32(%rsi,%rbx,4)
|
|
vmovss (%rsi,%rbx,4), %xmm4 # xmm4 = mem[0],zero,zero,zero
|
|
vfnmadd231ss %xmm1, %xmm3, %xmm4 # xmm4 = -(xmm3 * xmm1) + xmm4
|
|
vmovss %xmm4, (%rsi,%rbx,4)
|
|
vfmadd231ss %xmm2, %xmm3, %xmm7 # xmm7 = (xmm3 * xmm2) + xmm7
|
|
vfmadd231ss %xmm0, %xmm3, %xmm5 # xmm5 = (xmm3 * xmm0) + xmm5
|
|
vfmadd231ss %xmm1, %xmm3, %xmm19 # xmm19 = (xmm3 * xmm1) + xmm19
|
|
movl $1, %r14d
|
|
movq %rdx, %rcx
|
|
.LBB0_71: # in Loop: Header=BB0_12 Depth=4
|
|
incq (%rcx)
|
|
.LBB0_72: # in Loop: Header=BB0_12 Depth=4
|
|
incq %rbx
|
|
cmpq $8, %rbx
|
|
je .LBB0_69
|
|
.LBB0_12: # Parent Loop BB0_7 Depth=1
|
|
# Parent Loop BB0_9 Depth=2
|
|
# Parent Loop BB0_10 Depth=3
|
|
# => This Inner Loop Header: Depth=4
|
|
cmpl %r10d, %r13d
|
|
jne .LBB0_14
|
|
# %bb.13: # in Loop: Header=BB0_12 Depth=4
|
|
cmpq %rbx, %rbp
|
|
jae .LBB0_72
|
|
.LBB0_14: # in Loop: Header=BB0_12 Depth=4
|
|
vsubss -64(%rdi,%rbx,4), %xmm14, %xmm2
|
|
vsubss -32(%rdi,%rbx,4), %xmm16, %xmm0
|
|
vsubss (%rdi,%rbx,4), %xmm18, %xmm1
|
|
vmulss %xmm2, %xmm2, %xmm3
|
|
vfmadd231ss %xmm0, %xmm0, %xmm3 # xmm3 = (xmm0 * xmm0) + xmm3
|
|
vfmadd231ss %xmm1, %xmm1, %xmm3 # xmm3 = (xmm1 * xmm1) + xmm3
|
|
vucomiss %xmm13, %xmm3
|
|
jb .LBB0_70
|
|
# %bb.15: # in Loop: Header=BB0_12 Depth=4
|
|
movq %r15, %rcx
|
|
jmp .LBB0_71
|
|
.p2align 5, 0x90
|
|
.LBB0_21: # in Loop: Header=BB0_10 Depth=3
|
|
cmpl %r10d, %r13d
|
|
jne .LBB0_24
|
|
# %bb.22: # in Loop: Header=BB0_10 Depth=3
|
|
vxorps %xmm19, %xmm19, %xmm19
|
|
testq %rbp, %rbp
|
|
jne .LBB0_24
|
|
# %bb.23: # in Loop: Header=BB0_10 Depth=3
|
|
vxorps %xmm5, %xmm5, %xmm5
|
|
vxorps %xmm7, %xmm7, %xmm7
|
|
cmpl %r10d, %r13d
|
|
je .LBB0_28
|
|
jmp .LBB0_29
|
|
.p2align 5, 0x90
|
|
.LBB0_24: # in Loop: Header=BB0_10 Depth=3
|
|
vsubss (%rax), %xmm14, %xmm15
|
|
vsubss 32(%rax), %xmm16, %xmm1
|
|
vsubss 64(%rax), %xmm18, %xmm2
|
|
vmulss %xmm15, %xmm15, %xmm0
|
|
vfmadd231ss %xmm1, %xmm1, %xmm0 # xmm0 = (xmm1 * xmm1) + xmm0
|
|
vfmadd231ss %xmm2, %xmm2, %xmm0 # xmm0 = (xmm2 * xmm2) + xmm0
|
|
vxorps %xmm19, %xmm19, %xmm19
|
|
vucomiss %xmm13, %xmm0
|
|
movq %r15, %rbx
|
|
vxorps %xmm5, %xmm5, %xmm5
|
|
vxorps %xmm7, %xmm7, %xmm7
|
|
jae .LBB0_26
|
|
# %bb.25: # in Loop: Header=BB0_10 Depth=3
|
|
vdivss %xmm0, %xmm10, %xmm0
|
|
vmulss %xmm20, %xmm0, %xmm3
|
|
vmulss %xmm0, %xmm0, %xmm5
|
|
vmulss %xmm3, %xmm5, %xmm3
|
|
vcvtss2sd %xmm3, %xmm3, %xmm3
|
|
vaddsd %xmm3, %xmm11, %xmm5
|
|
vcvtss2sd %xmm0, %xmm0, %xmm0
|
|
vmulsd %xmm0, %xmm12, %xmm0
|
|
vmulsd %xmm3, %xmm5, %xmm3
|
|
vmulsd %xmm0, %xmm3, %xmm0
|
|
vcvtsd2ss %xmm0, %xmm0, %xmm0
|
|
vmulss %xmm0, %xmm15, %xmm7
|
|
vmulss %xmm0, %xmm1, %xmm5
|
|
vmulss %xmm0, %xmm2, %xmm19
|
|
movl $1, %r14d
|
|
movq %rdx, %rbx
|
|
.LBB0_26: # in Loop: Header=BB0_10 Depth=3
|
|
incq (%rbx)
|
|
cmpl %r10d, %r13d
|
|
jne .LBB0_29
|
|
.LBB0_28: # in Loop: Header=BB0_10 Depth=3
|
|
cmpq $1, %rbp
|
|
je .LBB0_33
|
|
.LBB0_29: # in Loop: Header=BB0_10 Depth=3
|
|
vsubss 4(%rax), %xmm14, %xmm2
|
|
vsubss 36(%rax), %xmm16, %xmm1
|
|
vsubss 68(%rax), %xmm18, %xmm15
|
|
vmulss %xmm2, %xmm2, %xmm0
|
|
vfmadd231ss %xmm1, %xmm1, %xmm0 # xmm0 = (xmm1 * xmm1) + xmm0
|
|
vfmadd231ss %xmm15, %xmm15, %xmm0 # xmm0 = (xmm15 * xmm15) + xmm0
|
|
vucomiss %xmm13, %xmm0
|
|
jae .LBB0_30
|
|
# %bb.31: # in Loop: Header=BB0_10 Depth=3
|
|
vdivss %xmm0, %xmm10, %xmm0
|
|
vmulss %xmm20, %xmm0, %xmm17
|
|
vmulss %xmm0, %xmm0, %xmm3
|
|
vmulss %xmm17, %xmm3, %xmm3
|
|
vcvtss2sd %xmm3, %xmm3, %xmm3
|
|
vaddsd %xmm11, %xmm3, %xmm17
|
|
vcvtss2sd %xmm0, %xmm0, %xmm0
|
|
vmulsd %xmm0, %xmm12, %xmm0
|
|
vmulsd %xmm3, %xmm17, %xmm3
|
|
vmulsd %xmm0, %xmm3, %xmm0
|
|
vcvtsd2ss %xmm0, %xmm0, %xmm0
|
|
vfmadd231ss %xmm2, %xmm0, %xmm7 # xmm7 = (xmm0 * xmm2) + xmm7
|
|
vfmadd231ss %xmm1, %xmm0, %xmm5 # xmm5 = (xmm0 * xmm1) + xmm5
|
|
vfmadd231ss %xmm15, %xmm0, %xmm19 # xmm19 = (xmm0 * xmm15) + xmm19
|
|
movl $1, %r14d
|
|
movq %rdx, %rbx
|
|
jmp .LBB0_32
|
|
.p2align 5, 0x90
|
|
.LBB0_30: # in Loop: Header=BB0_10 Depth=3
|
|
movq %r15, %rbx
|
|
.LBB0_32: # in Loop: Header=BB0_10 Depth=3
|
|
incq (%rbx)
|
|
.LBB0_33: # in Loop: Header=BB0_10 Depth=3
|
|
cmpl %r10d, %r13d
|
|
jne .LBB0_35
|
|
# %bb.34: # in Loop: Header=BB0_10 Depth=3
|
|
cmpq $2, %rbp
|
|
je .LBB0_39
|
|
.LBB0_35: # in Loop: Header=BB0_10 Depth=3
|
|
vsubss 8(%rax), %xmm14, %xmm2
|
|
vsubss 40(%rax), %xmm16, %xmm1
|
|
vsubss 72(%rax), %xmm18, %xmm15
|
|
vmulss %xmm2, %xmm2, %xmm0
|
|
vfmadd231ss %xmm1, %xmm1, %xmm0 # xmm0 = (xmm1 * xmm1) + xmm0
|
|
vfmadd231ss %xmm15, %xmm15, %xmm0 # xmm0 = (xmm15 * xmm15) + xmm0
|
|
vucomiss %xmm13, %xmm0
|
|
jae .LBB0_36
|
|
# %bb.37: # in Loop: Header=BB0_10 Depth=3
|
|
vdivss %xmm0, %xmm10, %xmm0
|
|
vmulss %xmm20, %xmm0, %xmm17
|
|
vmulss %xmm0, %xmm0, %xmm3
|
|
vmulss %xmm17, %xmm3, %xmm3
|
|
vcvtss2sd %xmm3, %xmm3, %xmm3
|
|
vaddsd %xmm11, %xmm3, %xmm17
|
|
vcvtss2sd %xmm0, %xmm0, %xmm0
|
|
vmulsd %xmm0, %xmm12, %xmm0
|
|
vmulsd %xmm3, %xmm17, %xmm3
|
|
vmulsd %xmm0, %xmm3, %xmm0
|
|
vcvtsd2ss %xmm0, %xmm0, %xmm0
|
|
vfmadd231ss %xmm2, %xmm0, %xmm7 # xmm7 = (xmm0 * xmm2) + xmm7
|
|
vfmadd231ss %xmm1, %xmm0, %xmm5 # xmm5 = (xmm0 * xmm1) + xmm5
|
|
vfmadd231ss %xmm15, %xmm0, %xmm19 # xmm19 = (xmm0 * xmm15) + xmm19
|
|
movl $1, %r14d
|
|
movq %rdx, %rbx
|
|
jmp .LBB0_38
|
|
.p2align 5, 0x90
|
|
.LBB0_36: # in Loop: Header=BB0_10 Depth=3
|
|
movq %r15, %rbx
|
|
.LBB0_38: # in Loop: Header=BB0_10 Depth=3
|
|
incq (%rbx)
|
|
.LBB0_39: # in Loop: Header=BB0_10 Depth=3
|
|
cmpl %r10d, %r13d
|
|
jne .LBB0_41
|
|
# %bb.40: # in Loop: Header=BB0_10 Depth=3
|
|
cmpq $3, %rbp
|
|
je .LBB0_45
|
|
.LBB0_41: # in Loop: Header=BB0_10 Depth=3
|
|
vsubss 12(%rax), %xmm14, %xmm2
|
|
vsubss 44(%rax), %xmm16, %xmm1
|
|
vsubss 76(%rax), %xmm18, %xmm15
|
|
vmulss %xmm2, %xmm2, %xmm0
|
|
vfmadd231ss %xmm1, %xmm1, %xmm0 # xmm0 = (xmm1 * xmm1) + xmm0
|
|
vfmadd231ss %xmm15, %xmm15, %xmm0 # xmm0 = (xmm15 * xmm15) + xmm0
|
|
vucomiss %xmm13, %xmm0
|
|
jae .LBB0_42
|
|
# %bb.43: # in Loop: Header=BB0_10 Depth=3
|
|
vdivss %xmm0, %xmm10, %xmm0
|
|
vmulss %xmm20, %xmm0, %xmm17
|
|
vmulss %xmm0, %xmm0, %xmm3
|
|
vmulss %xmm17, %xmm3, %xmm3
|
|
vcvtss2sd %xmm3, %xmm3, %xmm3
|
|
vaddsd %xmm11, %xmm3, %xmm17
|
|
vcvtss2sd %xmm0, %xmm0, %xmm0
|
|
vmulsd %xmm0, %xmm12, %xmm0
|
|
vmulsd %xmm3, %xmm17, %xmm3
|
|
vmulsd %xmm0, %xmm3, %xmm0
|
|
vcvtsd2ss %xmm0, %xmm0, %xmm0
|
|
vfmadd231ss %xmm2, %xmm0, %xmm7 # xmm7 = (xmm0 * xmm2) + xmm7
|
|
vfmadd231ss %xmm1, %xmm0, %xmm5 # xmm5 = (xmm0 * xmm1) + xmm5
|
|
vfmadd231ss %xmm15, %xmm0, %xmm19 # xmm19 = (xmm0 * xmm15) + xmm19
|
|
movl $1, %r14d
|
|
movq %rdx, %rbx
|
|
jmp .LBB0_44
|
|
.p2align 5, 0x90
|
|
.LBB0_42: # in Loop: Header=BB0_10 Depth=3
|
|
movq %r15, %rbx
|
|
.LBB0_44: # in Loop: Header=BB0_10 Depth=3
|
|
incq (%rbx)
|
|
.LBB0_45: # in Loop: Header=BB0_10 Depth=3
|
|
cmpl %r10d, %r13d
|
|
jne .LBB0_47
|
|
# %bb.46: # in Loop: Header=BB0_10 Depth=3
|
|
cmpq $4, %rbp
|
|
je .LBB0_51
|
|
.LBB0_47: # in Loop: Header=BB0_10 Depth=3
|
|
vsubss 16(%rax), %xmm14, %xmm2
|
|
vsubss 48(%rax), %xmm16, %xmm1
|
|
vsubss 80(%rax), %xmm18, %xmm15
|
|
vmulss %xmm2, %xmm2, %xmm0
|
|
vfmadd231ss %xmm1, %xmm1, %xmm0 # xmm0 = (xmm1 * xmm1) + xmm0
|
|
vfmadd231ss %xmm15, %xmm15, %xmm0 # xmm0 = (xmm15 * xmm15) + xmm0
|
|
vucomiss %xmm13, %xmm0
|
|
jae .LBB0_48
|
|
# %bb.49: # in Loop: Header=BB0_10 Depth=3
|
|
vdivss %xmm0, %xmm10, %xmm0
|
|
vmulss %xmm20, %xmm0, %xmm17
|
|
vmulss %xmm0, %xmm0, %xmm3
|
|
vmulss %xmm17, %xmm3, %xmm3
|
|
vcvtss2sd %xmm3, %xmm3, %xmm3
|
|
vaddsd %xmm11, %xmm3, %xmm17
|
|
vcvtss2sd %xmm0, %xmm0, %xmm0
|
|
vmulsd %xmm0, %xmm12, %xmm0
|
|
vmulsd %xmm3, %xmm17, %xmm3
|
|
vmulsd %xmm0, %xmm3, %xmm0
|
|
vcvtsd2ss %xmm0, %xmm0, %xmm0
|
|
vfmadd231ss %xmm2, %xmm0, %xmm7 # xmm7 = (xmm0 * xmm2) + xmm7
|
|
vfmadd231ss %xmm1, %xmm0, %xmm5 # xmm5 = (xmm0 * xmm1) + xmm5
|
|
vfmadd231ss %xmm15, %xmm0, %xmm19 # xmm19 = (xmm0 * xmm15) + xmm19
|
|
movl $1, %r14d
|
|
movq %rdx, %rbx
|
|
jmp .LBB0_50
|
|
.p2align 5, 0x90
|
|
.LBB0_48: # in Loop: Header=BB0_10 Depth=3
|
|
movq %r15, %rbx
|
|
.LBB0_50: # in Loop: Header=BB0_10 Depth=3
|
|
incq (%rbx)
|
|
.LBB0_51: # in Loop: Header=BB0_10 Depth=3
|
|
cmpl %r10d, %r13d
|
|
jne .LBB0_53
|
|
# %bb.52: # in Loop: Header=BB0_10 Depth=3
|
|
cmpq $5, %rbp
|
|
je .LBB0_57
|
|
.LBB0_53: # in Loop: Header=BB0_10 Depth=3
|
|
vsubss 20(%rax), %xmm14, %xmm2
|
|
vsubss 52(%rax), %xmm16, %xmm1
|
|
vsubss 84(%rax), %xmm18, %xmm15
|
|
vmulss %xmm2, %xmm2, %xmm0
|
|
vfmadd231ss %xmm1, %xmm1, %xmm0 # xmm0 = (xmm1 * xmm1) + xmm0
|
|
vfmadd231ss %xmm15, %xmm15, %xmm0 # xmm0 = (xmm15 * xmm15) + xmm0
|
|
vucomiss %xmm13, %xmm0
|
|
jae .LBB0_54
|
|
# %bb.55: # in Loop: Header=BB0_10 Depth=3
|
|
vdivss %xmm0, %xmm10, %xmm0
|
|
vmulss %xmm20, %xmm0, %xmm17
|
|
vmulss %xmm0, %xmm0, %xmm3
|
|
vmulss %xmm17, %xmm3, %xmm3
|
|
vcvtss2sd %xmm3, %xmm3, %xmm3
|
|
vaddsd %xmm11, %xmm3, %xmm17
|
|
vcvtss2sd %xmm0, %xmm0, %xmm0
|
|
vmulsd %xmm0, %xmm12, %xmm0
|
|
vmulsd %xmm3, %xmm17, %xmm3
|
|
vmulsd %xmm0, %xmm3, %xmm0
|
|
vcvtsd2ss %xmm0, %xmm0, %xmm0
|
|
vfmadd231ss %xmm2, %xmm0, %xmm7 # xmm7 = (xmm0 * xmm2) + xmm7
|
|
vfmadd231ss %xmm1, %xmm0, %xmm5 # xmm5 = (xmm0 * xmm1) + xmm5
|
|
vfmadd231ss %xmm15, %xmm0, %xmm19 # xmm19 = (xmm0 * xmm15) + xmm19
|
|
movl $1, %r14d
|
|
movq %rdx, %rbx
|
|
jmp .LBB0_56
|
|
.p2align 5, 0x90
|
|
.LBB0_54: # in Loop: Header=BB0_10 Depth=3
|
|
movq %r15, %rbx
|
|
.LBB0_56: # in Loop: Header=BB0_10 Depth=3
|
|
incq (%rbx)
|
|
.LBB0_57: # in Loop: Header=BB0_10 Depth=3
|
|
cmpl %r10d, %r13d
|
|
jne .LBB0_59
|
|
# %bb.58: # in Loop: Header=BB0_10 Depth=3
|
|
cmpq $6, %rbp
|
|
je .LBB0_63
|
|
.LBB0_59: # in Loop: Header=BB0_10 Depth=3
|
|
vsubss 24(%rax), %xmm14, %xmm2
|
|
vsubss 56(%rax), %xmm16, %xmm1
|
|
vsubss 88(%rax), %xmm18, %xmm15
|
|
vmulss %xmm2, %xmm2, %xmm0
|
|
vfmadd231ss %xmm1, %xmm1, %xmm0 # xmm0 = (xmm1 * xmm1) + xmm0
|
|
vfmadd231ss %xmm15, %xmm15, %xmm0 # xmm0 = (xmm15 * xmm15) + xmm0
|
|
vucomiss %xmm13, %xmm0
|
|
jae .LBB0_60
|
|
# %bb.61: # in Loop: Header=BB0_10 Depth=3
|
|
vdivss %xmm0, %xmm10, %xmm0
|
|
vmulss %xmm20, %xmm0, %xmm17
|
|
vmulss %xmm0, %xmm0, %xmm3
|
|
vmulss %xmm17, %xmm3, %xmm3
|
|
vcvtss2sd %xmm3, %xmm3, %xmm3
|
|
vaddsd %xmm11, %xmm3, %xmm17
|
|
vcvtss2sd %xmm0, %xmm0, %xmm0
|
|
vmulsd %xmm0, %xmm12, %xmm0
|
|
vmulsd %xmm3, %xmm17, %xmm3
|
|
vmulsd %xmm0, %xmm3, %xmm0
|
|
vcvtsd2ss %xmm0, %xmm0, %xmm0
|
|
vfmadd231ss %xmm2, %xmm0, %xmm7 # xmm7 = (xmm0 * xmm2) + xmm7
|
|
vfmadd231ss %xmm1, %xmm0, %xmm5 # xmm5 = (xmm0 * xmm1) + xmm5
|
|
vfmadd231ss %xmm15, %xmm0, %xmm19 # xmm19 = (xmm0 * xmm15) + xmm19
|
|
movl $1, %r14d
|
|
movq %rdx, %rbx
|
|
jmp .LBB0_62
|
|
.p2align 5, 0x90
|
|
.LBB0_60: # in Loop: Header=BB0_10 Depth=3
|
|
movq %r15, %rbx
|
|
.LBB0_62: # in Loop: Header=BB0_10 Depth=3
|
|
incq (%rbx)
|
|
.LBB0_63: # in Loop: Header=BB0_10 Depth=3
|
|
cmpl %r10d, %r13d
|
|
jne .LBB0_65
|
|
# %bb.64: # in Loop: Header=BB0_10 Depth=3
|
|
cmpq $7, %rbp
|
|
je .LBB0_69
|
|
.LBB0_65: # in Loop: Header=BB0_10 Depth=3
|
|
vsubss 28(%rax), %xmm14, %xmm2
|
|
vsubss 60(%rax), %xmm16, %xmm1
|
|
vsubss 92(%rax), %xmm18, %xmm0
|
|
vmulss %xmm2, %xmm2, %xmm3
|
|
vfmadd231ss %xmm1, %xmm1, %xmm3 # xmm3 = (xmm1 * xmm1) + xmm3
|
|
vfmadd231ss %xmm0, %xmm0, %xmm3 # xmm3 = (xmm0 * xmm0) + xmm3
|
|
vucomiss %xmm13, %xmm3
|
|
jb .LBB0_67
|
|
# %bb.66: # in Loop: Header=BB0_10 Depth=3
|
|
movq %r15, %rbx
|
|
jmp .LBB0_68
|
|
.p2align 5, 0x90
|
|
.LBB0_16:
|
|
movq 48(%rsp), %rcx # 8-byte Reload
|
|
incq %rcx
|
|
movq 16(%rsp), %rax # 8-byte Reload
|
|
movq %rcx, (%rax)
|
|
vmovdqu %xmm9, 8(%rax)
|
|
.LBB0_17:
|
|
movl $.L.str.1, %edi
|
|
callq likwid_markerStopRegion
|
|
xorl %eax, %eax
|
|
callq getTimeStamp
|
|
vmovsd %xmm0, (%rsp) # 8-byte Spill
|
|
movl $.L.str.2, %edi
|
|
xorl %eax, %eax
|
|
callq debug_printf
|
|
vmovsd (%rsp), %xmm0 # 8-byte Reload
|
|
# xmm0 = mem[0],zero
|
|
vsubsd 56(%rsp), %xmm0, %xmm0 # 8-byte Folded Reload
|
|
addq $152, %rsp
|
|
.cfi_def_cfa_offset 56
|
|
popq %rbx
|
|
.cfi_def_cfa_offset 48
|
|
popq %r12
|
|
.cfi_def_cfa_offset 40
|
|
popq %r13
|
|
.cfi_def_cfa_offset 32
|
|
popq %r14
|
|
.cfi_def_cfa_offset 24
|
|
popq %r15
|
|
.cfi_def_cfa_offset 16
|
|
popq %rbp
|
|
.cfi_def_cfa_offset 8
|
|
retq
|
|
.Lfunc_end0:
|
|
.size computeForceLJ_ref, .Lfunc_end0-computeForceLJ_ref
|
|
.cfi_endproc
|
|
# -- End function
|
|
.section .rodata.cst4,"aM",@progbits,4
|
|
.p2align 2 # -- Begin function computeForceLJ_2xnn_half
|
|
.LCPI1_0:
|
|
.long 0xbf000000 # -0.5
|
|
.LCPI1_1:
|
|
.long 0x42400000 # 48
|
|
.section .rodata.cst16,"aM",@progbits,16
|
|
.p2align 4
|
|
.LCPI1_2:
|
|
.quad 1 # 0x1
|
|
.zero 8
|
|
.section .rodata.cst8,"aM",@progbits,8
|
|
.p2align 3
|
|
.LCPI1_3:
|
|
.quad 0x3fe0000000000000 # 0.5
|
|
.LCPI1_4:
|
|
.quad 0x41cdcd6500000000 # 1.0E+9
|
|
.text
|
|
.globl computeForceLJ_2xnn_half
|
|
.p2align 4, 0x90
|
|
.type computeForceLJ_2xnn_half,@function
|
|
computeForceLJ_2xnn_half: #
|
|
.cfi_startproc
|
|
# %bb.0:
|
|
pushq %rbp
|
|
.cfi_def_cfa_offset 16
|
|
pushq %r15
|
|
.cfi_def_cfa_offset 24
|
|
pushq %r14
|
|
.cfi_def_cfa_offset 32
|
|
pushq %r13
|
|
.cfi_def_cfa_offset 40
|
|
pushq %r12
|
|
.cfi_def_cfa_offset 48
|
|
pushq %rbx
|
|
.cfi_def_cfa_offset 56
|
|
subq $232, %rsp
|
|
.cfi_def_cfa_offset 288
|
|
.cfi_offset %rbx, -56
|
|
.cfi_offset %r12, -48
|
|
.cfi_offset %r13, -40
|
|
.cfi_offset %r14, -32
|
|
.cfi_offset %r15, -24
|
|
.cfi_offset %rbp, -16
|
|
movq %rcx, 48(%rsp) # 8-byte Spill
|
|
movq %rdx, %r12
|
|
movq %rsi, %r15
|
|
movq %rdi, %rbp
|
|
xorl %ebx, %ebx
|
|
movl $.L.str.3, %edi
|
|
xorl %eax, %eax
|
|
callq debug_printf
|
|
vmovss 108(%rbp), %xmm0 # xmm0 = mem[0],zero,zero,zero
|
|
vmovss %xmm0, (%rsp) # 4-byte Spill
|
|
vbroadcastss 48(%rbp), %zmm2
|
|
movq %rbp, 32(%rsp) # 8-byte Spill
|
|
vpbroadcastd 40(%rbp), %zmm3
|
|
movl 20(%r15), %r11d
|
|
testl %r11d, %r11d
|
|
jle .LBB1_5
|
|
# %bb.1:
|
|
movq 176(%r15), %r9
|
|
movq 192(%r15), %r10
|
|
decq %r11
|
|
leaq 64(%r9), %r8
|
|
xorl %edi, %edi
|
|
vxorps %xmm0, %xmm0, %xmm0
|
|
vxorps %xmm1, %xmm1, %xmm1
|
|
jmp .LBB1_2
|
|
.p2align 5, 0x90
|
|
.LBB1_21: # in Loop: Header=BB1_2 Depth=1
|
|
cmpq %r11, %rdi
|
|
leaq 1(%rdi), %rdi
|
|
je .LBB1_5
|
|
.LBB1_2: # =>This Loop Header: Depth=1
|
|
# Child Loop BB1_16 Depth 2
|
|
# Child Loop BB1_20 Depth 2
|
|
leaq (%rdi,%rdi,8), %rax
|
|
leaq (%rax,%rax,2), %rax
|
|
addq %rdi, %rax
|
|
movl (%r10,%rax), %ecx
|
|
testl %ecx, %ecx
|
|
jle .LBB1_21
|
|
# %bb.3: # in Loop: Header=BB1_2 Depth=1
|
|
leal (,%rdi,4), %esi
|
|
movl %esi, %eax
|
|
andl $2147483640, %eax # imm = 0x7FFFFFF8
|
|
leal (%rax,%rax,2), %eax
|
|
andl $4, %esi
|
|
orl %eax, %esi
|
|
cmpl $7, %ecx
|
|
ja .LBB1_15
|
|
# %bb.4: # in Loop: Header=BB1_2 Depth=1
|
|
movl %ecx, %ebp
|
|
andl $-8, %ebp
|
|
cmpq %rcx, %rbp
|
|
jae .LBB1_21
|
|
jmp .LBB1_19
|
|
.p2align 5, 0x90
|
|
.LBB1_15: # in Loop: Header=BB1_2 Depth=1
|
|
leaq (,%rcx,4), %rbp
|
|
andq $-32, %rbp
|
|
movl %esi, %r14d
|
|
leaq (%r9,%r14,4), %rax
|
|
xorl %edx, %edx
|
|
.p2align 4, 0x90
|
|
.LBB1_16: # Parent Loop BB1_2 Depth=1
|
|
# => This Inner Loop Header: Depth=2
|
|
vmovups %ymm0, (%rax,%rdx)
|
|
addq $32, %rdx
|
|
cmpq %rdx, %rbp
|
|
jne .LBB1_16
|
|
# %bb.17: # in Loop: Header=BB1_2 Depth=1
|
|
movl %ecx, %ebp
|
|
andl $-8, %ebp
|
|
addq %rbp, %r14
|
|
vmovups %zmm1, (%r9,%r14,4)
|
|
cmpq %rcx, %rbp
|
|
jae .LBB1_21
|
|
.LBB1_19: # in Loop: Header=BB1_2 Depth=1
|
|
movl %esi, %eax
|
|
leaq (%r8,%rax,4), %rdx
|
|
.p2align 4, 0x90
|
|
.LBB1_20: # Parent Loop BB1_2 Depth=1
|
|
# => This Inner Loop Header: Depth=2
|
|
movl $0, -64(%rdx,%rbp,4)
|
|
movl $0, -32(%rdx,%rbp,4)
|
|
movl $0, (%rdx,%rbp,4)
|
|
incq %rbp
|
|
cmpq %rbp, %rcx
|
|
jne .LBB1_20
|
|
jmp .LBB1_21
|
|
.p2align 5, 0x90
|
|
.LBB1_5:
|
|
xorl %eax, %eax
|
|
vmovups %zmm2, 160(%rsp) # 64-byte Spill
|
|
vmovdqu64 %zmm3, 96(%rsp) # 64-byte Spill
|
|
vzeroupper
|
|
callq getTimeStamp
|
|
vmovsd %xmm0, 24(%rsp) # 8-byte Spill
|
|
movl $.L.str.1, %edi
|
|
callq likwid_markerStartRegion
|
|
vmovups 96(%rsp), %zmm31 # 64-byte Reload
|
|
vmovups 160(%rsp), %zmm30 # 64-byte Reload
|
|
cmpl $0, 20(%r15)
|
|
jle .LBB1_10
|
|
# %bb.6:
|
|
vmovss (%rsp), %xmm0 # 4-byte Reload
|
|
# xmm0 = mem[0],zero,zero,zero
|
|
vmulss %xmm0, %xmm0, %xmm0
|
|
vbroadcastss %xmm0, %zmm0
|
|
xorl %r11d, %r11d
|
|
vbroadcastss .LCPI1_0(%rip), %zmm1 # zmm1 = [-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1]
|
|
vbroadcastss .LCPI1_1(%rip), %zmm2 # zmm2 = [4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1]
|
|
movw $4369, %cx # imm = 0x1111
|
|
kmovw %ecx, %k1
|
|
vmovdqu .LCPI1_2(%rip), %xmm3 # xmm3 = <1,u>
|
|
# AlignMOV convert to UnAlignMOV
|
|
vmovsd .LCPI1_3(%rip), %xmm4 # xmm4 = mem[0],zero
|
|
xorl %ebx, %ebx
|
|
movq %r12, 40(%rsp) # 8-byte Spill
|
|
movq %r15, 16(%rsp) # 8-byte Spill
|
|
jmp .LBB1_7
|
|
.p2align 5, 0x90
|
|
.LBB1_13: # in Loop: Header=BB1_7 Depth=1
|
|
movl 12(%rsp), %ebx # 4-byte Reload
|
|
movq 40(%rsp), %r12 # 8-byte Reload
|
|
movq 16(%rsp), %r15 # 8-byte Reload
|
|
movq 80(%rsp), %rax # 8-byte Reload
|
|
movq 72(%rsp), %rsi # 8-byte Reload
|
|
movq 64(%rsp), %r10 # 8-byte Reload
|
|
movq 56(%rsp), %rcx # 8-byte Reload
|
|
.LBB1_9: # in Loop: Header=BB1_7 Depth=1
|
|
vshuff64x2 $136, %zmm14, %zmm12, %zmm7 # zmm7 = zmm12[0,1,4,5],zmm14[0,1,4,5]
|
|
vshuff64x2 $221, %zmm14, %zmm12, %zmm10 # zmm10 = zmm12[2,3,6,7],zmm14[2,3,6,7]
|
|
vaddps %zmm10, %zmm7, %zmm7
|
|
vpermilpd $85, %zmm7, %zmm10 # zmm10 = zmm7[1,0,3,2,5,4,7,6]
|
|
vaddps %zmm10, %zmm7, %zmm7
|
|
vpermilps $177, %zmm7, %zmm10 # zmm10 = zmm7[1,0,3,2,5,4,7,6,9,8,11,10,13,12,15,14]
|
|
vaddps %zmm10, %zmm7, %zmm7
|
|
vcompressps %zmm7, %zmm7 {%k1} {z}
|
|
vaddps (%r10,%rax,4), %xmm7, %xmm7
|
|
vmovups %xmm7, (%r10,%rax,4) # AlignMOV convert to UnAlignMOV
|
|
vshuff64x2 $136, %zmm9, %zmm8, %zmm7 # zmm7 = zmm8[0,1,4,5],zmm9[0,1,4,5]
|
|
vshuff64x2 $221, %zmm9, %zmm8, %zmm8 # zmm8 = zmm8[2,3,6,7],zmm9[2,3,6,7]
|
|
vaddps %zmm8, %zmm7, %zmm7
|
|
vpermilpd $85, %zmm7, %zmm8 # zmm8 = zmm7[1,0,3,2,5,4,7,6]
|
|
vaddps %zmm8, %zmm7, %zmm7
|
|
vpermilps $177, %zmm7, %zmm8 # zmm8 = zmm7[1,0,3,2,5,4,7,6,9,8,11,10,13,12,15,14]
|
|
vaddps %zmm8, %zmm7, %zmm7
|
|
vcompressps %zmm7, %zmm7 {%k1} {z}
|
|
vaddps 32(%r10,%rax,4), %xmm7, %xmm7
|
|
vmovups %xmm7, 32(%r10,%rax,4) # AlignMOV convert to UnAlignMOV
|
|
vshuff64x2 $136, %zmm5, %zmm6, %zmm7 # zmm7 = zmm6[0,1,4,5],zmm5[0,1,4,5]
|
|
vshuff64x2 $221, %zmm5, %zmm6, %zmm5 # zmm5 = zmm6[2,3,6,7],zmm5[2,3,6,7]
|
|
vaddps %zmm5, %zmm7, %zmm5
|
|
vpermilpd $85, %zmm5, %zmm6 # zmm6 = zmm5[1,0,3,2,5,4,7,6]
|
|
vaddps %zmm6, %zmm5, %zmm5
|
|
vpermilps $177, %zmm5, %zmm6 # zmm6 = zmm5[1,0,3,2,5,4,7,6,9,8,11,10,13,12,15,14]
|
|
vaddps %zmm6, %zmm5, %zmm5
|
|
vcompressps %zmm5, %zmm5 {%k1} {z}
|
|
vaddps 64(%r10,%rax,4), %xmm5, %xmm5
|
|
vmovups %xmm5, 64(%r10,%rax,4) # AlignMOV convert to UnAlignMOV
|
|
vpinsrq $1, %rcx, %xmm3, %xmm5
|
|
movq 48(%rsp), %rdx # 8-byte Reload
|
|
vpaddq (%rdx), %xmm5, %xmm5
|
|
vmovdqu %xmm5, (%rdx)
|
|
addl %esi, %ebx
|
|
vcvtsi2sd %esi, %xmm11, %xmm5
|
|
vmulsd %xmm4, %xmm5, %xmm5
|
|
vcvttsd2si %xmm5, %rcx
|
|
addq %rcx, 16(%rdx)
|
|
incq %r11
|
|
movslq 20(%r15), %rcx
|
|
cmpq %rcx, %r11
|
|
jge .LBB1_10
|
|
.LBB1_7: # =>This Loop Header: Depth=1
|
|
# Child Loop BB1_12 Depth 2
|
|
leal (,%r11,4), %eax
|
|
movl %eax, %ecx
|
|
andl $2147483640, %ecx # imm = 0x7FFFFFF8
|
|
leal (%rcx,%rcx,2), %ecx
|
|
andl $4, %eax
|
|
orl %ecx, %eax
|
|
movq 176(%r15), %r10
|
|
movq 24(%r12), %rcx
|
|
movslq (%rcx,%r11,4), %rsi
|
|
testq %rsi, %rsi
|
|
jle .LBB1_8
|
|
# %bb.11: # in Loop: Header=BB1_7 Depth=1
|
|
movl %ebx, 12(%rsp) # 4-byte Spill
|
|
movq 160(%r15), %r15
|
|
vbroadcastss (%r15,%rax,4), %ymm5
|
|
movq 8(%r12), %rcx
|
|
vbroadcastss 4(%r15,%rax,4), %ymm6
|
|
vinsertf64x4 $1, %ymm6, %zmm5, %zmm7
|
|
vbroadcastss 8(%r15,%rax,4), %ymm5
|
|
vbroadcastss 12(%r15,%rax,4), %ymm6
|
|
vbroadcastss 32(%r15,%rax,4), %ymm8
|
|
vbroadcastss 36(%r15,%rax,4), %ymm9
|
|
vinsertf64x4 $1, %ymm6, %zmm5, %zmm10
|
|
vinsertf64x4 $1, %ymm9, %zmm8, %zmm11
|
|
vbroadcastss 40(%r15,%rax,4), %ymm5
|
|
vbroadcastss 44(%r15,%rax,4), %ymm6
|
|
vinsertf64x4 $1, %ymm6, %zmm5, %zmm13
|
|
vbroadcastss 64(%r15,%rax,4), %ymm5
|
|
vbroadcastss 68(%r15,%rax,4), %ymm6
|
|
vinsertf64x4 $1, %ymm6, %zmm5, %zmm15
|
|
vbroadcastss 72(%r15,%rax,4), %ymm5
|
|
movq %rax, 80(%rsp) # 8-byte Spill
|
|
vbroadcastss 76(%r15,%rax,4), %ymm6
|
|
vinsertf64x4 $1, %ymm6, %zmm5, %zmm16
|
|
movq %rsi, 72(%rsp) # 8-byte Spill
|
|
movl %esi, %eax
|
|
movl 16(%r12), %edx
|
|
imull %r11d, %edx
|
|
movslq %edx, %rdx
|
|
leaq (%rcx,%rdx,4), %rcx
|
|
movq %rcx, (%rsp) # 8-byte Spill
|
|
movq %rax, 56(%rsp) # 8-byte Spill
|
|
decq %rax
|
|
movq %rax, 88(%rsp) # 8-byte Spill
|
|
vxorps %xmm12, %xmm12, %xmm12
|
|
movq %r10, 64(%rsp) # 8-byte Spill
|
|
xorl %ecx, %ecx
|
|
vxorps %xmm8, %xmm8, %xmm8
|
|
vxorps %xmm6, %xmm6, %xmm6
|
|
vxorps %xmm14, %xmm14, %xmm14
|
|
vxorps %xmm9, %xmm9, %xmm9
|
|
vxorps %xmm5, %xmm5, %xmm5
|
|
.p2align 4, 0x90
|
|
.LBB1_12: # Parent Loop BB1_7 Depth=1
|
|
# => This Inner Loop Header: Depth=2
|
|
movq (%rsp), %rax # 8-byte Reload
|
|
movslq (%rax,%rcx,4), %rdx
|
|
leal (%rdx,%rdx), %esi
|
|
xorl %ebx, %ebx
|
|
cmpq %rsi, %r11
|
|
leal 1(%rdx,%rdx), %edi
|
|
setne %bl
|
|
leal (%rbx,%rbx,2), %ebx
|
|
movl $255, %ebp
|
|
movl $248, %eax
|
|
cmovel %eax, %ebp
|
|
orl $252, %ebx
|
|
leal -127(%rbp), %r8d
|
|
cmpq %rdi, %r11
|
|
cmovnel %ebp, %r8d
|
|
leal 193(%rbx), %r14d
|
|
xorl %r13d, %r13d
|
|
cmpq %rdi, %r11
|
|
cmovnel %ebx, %r14d
|
|
sete %r13b
|
|
movl $0, %r9d
|
|
movl $-31, %eax
|
|
cmovel %eax, %r9d
|
|
leal 240(%r13), %edi
|
|
addl $255, %r13d
|
|
xorl %ebx, %ebx
|
|
cmpq %rsi, %r11
|
|
cmovel %edi, %r13d
|
|
sete %bl
|
|
shlq $5, %rdx
|
|
leaq (%rdx,%rdx,2), %r12
|
|
vmovupd (%r15,%r12), %zmm17
|
|
vbroadcastf64x4 (%r15,%r12), %zmm18 # zmm18 = mem[0,1,2,3,0,1,2,3]
|
|
vbroadcastf64x4 64(%r15,%r12), %zmm19 # zmm19 = mem[0,1,2,3,0,1,2,3]
|
|
subl %ebx, %r9d
|
|
addl $255, %r9d
|
|
shll $8, %r14d
|
|
orl %r9d, %r14d
|
|
kmovd %r14d, %k2
|
|
shll $8, %r13d
|
|
orl %r8d, %r13d
|
|
kmovd %r13d, %k3
|
|
vshuff64x2 $238, %zmm17, %zmm17, %zmm17 # zmm17 = zmm17[4,5,6,7,4,5,6,7]
|
|
vsubps %zmm18, %zmm7, %zmm20
|
|
vsubps %zmm17, %zmm11, %zmm21
|
|
vsubps %zmm19, %zmm15, %zmm22
|
|
vsubps %zmm18, %zmm10, %zmm18
|
|
vsubps %zmm17, %zmm13, %zmm17
|
|
vsubps %zmm19, %zmm16, %zmm19
|
|
vmulps %zmm22, %zmm22, %zmm23
|
|
vfmadd231ps %zmm21, %zmm21, %zmm23 # zmm23 = (zmm21 * zmm21) + zmm23
|
|
vfmadd231ps %zmm20, %zmm20, %zmm23 # zmm23 = (zmm20 * zmm20) + zmm23
|
|
vmulps %zmm19, %zmm19, %zmm24
|
|
vrcp14ps %zmm23, %zmm25
|
|
vfmadd231ps %zmm17, %zmm17, %zmm24 # zmm24 = (zmm17 * zmm17) + zmm24
|
|
vfmadd231ps %zmm18, %zmm18, %zmm24 # zmm24 = (zmm18 * zmm18) + zmm24
|
|
vrcp14ps %zmm24, %zmm26
|
|
vmulps %zmm25, %zmm30, %zmm27
|
|
vmulps %zmm27, %zmm25, %zmm27
|
|
vmulps %zmm27, %zmm25, %zmm27
|
|
vmulps %zmm26, %zmm30, %zmm28
|
|
vmulps %zmm28, %zmm26, %zmm28
|
|
vmulps %zmm28, %zmm26, %zmm28
|
|
vaddps %zmm1, %zmm27, %zmm29
|
|
vmulps %zmm25, %zmm31, %zmm25
|
|
vmulps %zmm29, %zmm25, %zmm25
|
|
vmulps %zmm25, %zmm27, %zmm25
|
|
vmulps %zmm2, %zmm25, %zmm25
|
|
vaddps %zmm1, %zmm28, %zmm27
|
|
vmulps %zmm26, %zmm31, %zmm26
|
|
vmulps %zmm27, %zmm26, %zmm26
|
|
vmulps %zmm26, %zmm28, %zmm26
|
|
vmulps %zmm2, %zmm26, %zmm26
|
|
vcmpltps %zmm0, %zmm23, %k2 {%k2}
|
|
vmulps %zmm25, %zmm20, %zmm20 {%k2} {z}
|
|
vmulps %zmm25, %zmm21, %zmm21 {%k2} {z}
|
|
vmulps %zmm25, %zmm22, %zmm22 {%k2} {z}
|
|
vcmpltps %zmm0, %zmm24, %k2 {%k3}
|
|
vmulps %zmm26, %zmm18, %zmm18 {%k2} {z}
|
|
vmulps %zmm26, %zmm17, %zmm17 {%k2} {z}
|
|
vmulps %zmm26, %zmm19, %zmm19 {%k2} {z}
|
|
vaddps %zmm18, %zmm20, %zmm23
|
|
vaddps %zmm17, %zmm21, %zmm24
|
|
vextractf64x4 $1, %zmm23, %ymm25
|
|
vaddps %ymm25, %ymm23, %ymm23
|
|
vmovups (%r10,%r12), %ymm25 # AlignMOV convert to UnAlignMOV
|
|
vsubps %ymm23, %ymm25, %ymm23
|
|
vmovups 32(%r10,%r12), %ymm25 # AlignMOV convert to UnAlignMOV
|
|
vmovups 64(%r10,%r12), %ymm26 # AlignMOV convert to UnAlignMOV
|
|
vmovups %ymm23, (%r10,%r12) # AlignMOV convert to UnAlignMOV
|
|
vaddps %zmm19, %zmm22, %zmm23
|
|
vextractf64x4 $1, %zmm24, %ymm27
|
|
vaddps %ymm27, %ymm24, %ymm24
|
|
vsubps %ymm24, %ymm25, %ymm24
|
|
vmovups %ymm24, 32(%r10,%r12) # AlignMOV convert to UnAlignMOV
|
|
vextractf64x4 $1, %zmm23, %ymm24
|
|
vaddps %ymm24, %ymm23, %ymm23
|
|
vsubps %ymm23, %ymm26, %ymm23
|
|
vmovups %ymm23, 64(%r10,%r12) # AlignMOV convert to UnAlignMOV
|
|
vaddps %zmm20, %zmm12, %zmm12
|
|
vaddps %zmm21, %zmm8, %zmm8
|
|
vaddps %zmm22, %zmm6, %zmm6
|
|
vaddps %zmm18, %zmm14, %zmm14
|
|
vaddps %zmm17, %zmm9, %zmm9
|
|
vaddps %zmm19, %zmm5, %zmm5
|
|
cmpq %rcx, 88(%rsp) # 8-byte Folded Reload
|
|
je .LBB1_13
|
|
# %bb.14: # in Loop: Header=BB1_12 Depth=2
|
|
movq 16(%rsp), %rdx # 8-byte Reload
|
|
movq 160(%rdx), %r15
|
|
movq 176(%rdx), %r10
|
|
incq %rcx
|
|
jmp .LBB1_12
|
|
.p2align 5, 0x90
|
|
.LBB1_8: # in Loop: Header=BB1_7 Depth=1
|
|
vxorps %xmm5, %xmm5, %xmm5
|
|
movq %rsi, %rcx
|
|
vxorps %xmm9, %xmm9, %xmm9
|
|
vxorps %xmm14, %xmm14, %xmm14
|
|
vxorps %xmm6, %xmm6, %xmm6
|
|
vxorps %xmm8, %xmm8, %xmm8
|
|
vxorps %xmm12, %xmm12, %xmm12
|
|
jmp .LBB1_9
|
|
.p2align 5, 0x90
|
|
.LBB1_10:
|
|
movl $.L.str.1, %edi
|
|
vzeroupper
|
|
callq likwid_markerStopRegion
|
|
xorl %eax, %eax
|
|
callq getTimeStamp
|
|
movq 32(%rsp), %rax # 8-byte Reload
|
|
vmovsd 184(%rax), %xmm3 # xmm3 = mem[0],zero
|
|
vsubsd 24(%rsp), %xmm0, %xmm1 # 8-byte Folded Reload
|
|
vmovsd %xmm1, (%rsp) # 8-byte Spill
|
|
vmulsd .LCPI1_4(%rip), %xmm3, %xmm0
|
|
vmulsd %xmm1, %xmm0, %xmm0
|
|
vcvtusi2sd %ebx, %xmm11, %xmm2
|
|
vdivsd %xmm2, %xmm0, %xmm2
|
|
movl $.L.str.4, %edi
|
|
movl %ebx, %esi
|
|
vmovapd %xmm3, %xmm0
|
|
movb $3, %al
|
|
callq printf
|
|
movl $.L.str.5, %edi
|
|
xorl %eax, %eax
|
|
callq debug_printf
|
|
vmovsd (%rsp), %xmm0 # 8-byte Reload
|
|
# xmm0 = mem[0],zero
|
|
addq $232, %rsp
|
|
.cfi_def_cfa_offset 56
|
|
popq %rbx
|
|
.cfi_def_cfa_offset 48
|
|
popq %r12
|
|
.cfi_def_cfa_offset 40
|
|
popq %r13
|
|
.cfi_def_cfa_offset 32
|
|
popq %r14
|
|
.cfi_def_cfa_offset 24
|
|
popq %r15
|
|
.cfi_def_cfa_offset 16
|
|
popq %rbp
|
|
.cfi_def_cfa_offset 8
|
|
retq
|
|
.Lfunc_end1:
|
|
.size computeForceLJ_2xnn_half, .Lfunc_end1-computeForceLJ_2xnn_half
|
|
.cfi_endproc
|
|
# -- End function
|
|
.section .rodata.cst4,"aM",@progbits,4
|
|
.p2align 2 # -- Begin function computeForceLJ_2xnn_full
|
|
.LCPI2_0:
|
|
.long 0xbf000000 # -0.5
|
|
.LCPI2_1:
|
|
.long 0x42400000 # 48
|
|
.section .rodata.cst16,"aM",@progbits,16
|
|
.p2align 4
|
|
.LCPI2_2:
|
|
.quad 1 # 0x1
|
|
.zero 8
|
|
.text
|
|
.globl computeForceLJ_2xnn_full
|
|
.p2align 4, 0x90
|
|
.type computeForceLJ_2xnn_full,@function
|
|
computeForceLJ_2xnn_full: #
|
|
.cfi_startproc
|
|
# %bb.0:
|
|
pushq %rbp
|
|
.cfi_def_cfa_offset 16
|
|
pushq %r15
|
|
.cfi_def_cfa_offset 24
|
|
pushq %r14
|
|
.cfi_def_cfa_offset 32
|
|
pushq %r13
|
|
.cfi_def_cfa_offset 40
|
|
pushq %r12
|
|
.cfi_def_cfa_offset 48
|
|
pushq %rbx
|
|
.cfi_def_cfa_offset 56
|
|
subq $168, %rsp
|
|
.cfi_def_cfa_offset 224
|
|
.cfi_offset %rbx, -56
|
|
.cfi_offset %r12, -48
|
|
.cfi_offset %r13, -40
|
|
.cfi_offset %r14, -32
|
|
.cfi_offset %r15, -24
|
|
.cfi_offset %rbp, -16
|
|
movq %rcx, %r13
|
|
movq %rdx, %r14
|
|
movq %rsi, %rbp
|
|
movq %rdi, %r12
|
|
movl $.L.str.3, %edi
|
|
xorl %eax, %eax
|
|
callq debug_printf
|
|
vmovss 108(%r12), %xmm0 # xmm0 = mem[0],zero,zero,zero
|
|
vmovss %xmm0, 20(%rsp) # 4-byte Spill
|
|
vbroadcastss 48(%r12), %zmm2
|
|
vpbroadcastd 40(%r12), %zmm3
|
|
movl 20(%rbp), %r11d
|
|
testl %r11d, %r11d
|
|
jle .LBB2_5
|
|
# %bb.1:
|
|
movq 176(%rbp), %r9
|
|
movq 192(%rbp), %r10
|
|
decq %r11
|
|
leaq 64(%r9), %r8
|
|
xorl %edi, %edi
|
|
vxorps %xmm0, %xmm0, %xmm0
|
|
vxorps %xmm1, %xmm1, %xmm1
|
|
jmp .LBB2_2
|
|
.p2align 5, 0x90
|
|
.LBB2_20: # in Loop: Header=BB2_2 Depth=1
|
|
cmpq %r11, %rdi
|
|
leaq 1(%rdi), %rdi
|
|
je .LBB2_5
|
|
.LBB2_2: # =>This Loop Header: Depth=1
|
|
# Child Loop BB2_15 Depth 2
|
|
# Child Loop BB2_19 Depth 2
|
|
leaq (%rdi,%rdi,8), %rax
|
|
leaq (%rax,%rax,2), %rax
|
|
addq %rdi, %rax
|
|
movl (%r10,%rax), %ecx
|
|
testl %ecx, %ecx
|
|
jle .LBB2_20
|
|
# %bb.3: # in Loop: Header=BB2_2 Depth=1
|
|
leal (,%rdi,4), %esi
|
|
movl %esi, %eax
|
|
andl $2147483640, %eax # imm = 0x7FFFFFF8
|
|
leal (%rax,%rax,2), %eax
|
|
andl $4, %esi
|
|
orl %eax, %esi
|
|
cmpl $7, %ecx
|
|
ja .LBB2_14
|
|
# %bb.4: # in Loop: Header=BB2_2 Depth=1
|
|
movl %ecx, %ebx
|
|
andl $-8, %ebx
|
|
cmpq %rcx, %rbx
|
|
jae .LBB2_20
|
|
jmp .LBB2_18
|
|
.p2align 5, 0x90
|
|
.LBB2_14: # in Loop: Header=BB2_2 Depth=1
|
|
leaq (,%rcx,4), %rbx
|
|
andq $-32, %rbx
|
|
movl %esi, %r12d
|
|
leaq (%r9,%r12,4), %rax
|
|
xorl %edx, %edx
|
|
.p2align 4, 0x90
|
|
.LBB2_15: # Parent Loop BB2_2 Depth=1
|
|
# => This Inner Loop Header: Depth=2
|
|
vmovups %ymm0, (%rax,%rdx)
|
|
addq $32, %rdx
|
|
cmpq %rdx, %rbx
|
|
jne .LBB2_15
|
|
# %bb.16: # in Loop: Header=BB2_2 Depth=1
|
|
movl %ecx, %ebx
|
|
andl $-8, %ebx
|
|
addq %rbx, %r12
|
|
vmovups %zmm1, (%r9,%r12,4)
|
|
cmpq %rcx, %rbx
|
|
jae .LBB2_20
|
|
.LBB2_18: # in Loop: Header=BB2_2 Depth=1
|
|
movl %esi, %eax
|
|
leaq (%r8,%rax,4), %rdx
|
|
.p2align 4, 0x90
|
|
.LBB2_19: # Parent Loop BB2_2 Depth=1
|
|
# => This Inner Loop Header: Depth=2
|
|
movl $0, -64(%rdx,%rbx,4)
|
|
movl $0, -32(%rdx,%rbx,4)
|
|
movl $0, (%rdx,%rbx,4)
|
|
incq %rbx
|
|
cmpq %rbx, %rcx
|
|
jne .LBB2_19
|
|
jmp .LBB2_20
|
|
.p2align 5, 0x90
|
|
.LBB2_5:
|
|
xorl %r12d, %r12d
|
|
xorl %eax, %eax
|
|
vmovups %zmm2, 96(%rsp) # 64-byte Spill
|
|
vmovdqu64 %zmm3, 32(%rsp) # 64-byte Spill
|
|
vzeroupper
|
|
callq getTimeStamp
|
|
vmovsd %xmm0, 24(%rsp) # 8-byte Spill
|
|
movl $.L.str.1, %edi
|
|
callq likwid_markerStartRegion
|
|
vmovups 32(%rsp), %zmm27 # 64-byte Reload
|
|
vmovups 96(%rsp), %zmm26 # 64-byte Reload
|
|
cmpl $0, 20(%rbp)
|
|
jle .LBB2_10
|
|
# %bb.6:
|
|
vmovss 20(%rsp), %xmm0 # 4-byte Reload
|
|
# xmm0 = mem[0],zero,zero,zero
|
|
vmulss %xmm0, %xmm0, %xmm0
|
|
vbroadcastss %xmm0, %zmm0
|
|
movl $224, %r8d
|
|
vbroadcastss .LCPI2_0(%rip), %zmm1 # zmm1 = [-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1]
|
|
vbroadcastss .LCPI2_1(%rip), %zmm2 # zmm2 = [4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1]
|
|
movw $4369, %ax # imm = 0x1111
|
|
kmovw %eax, %k1
|
|
vmovdqu .LCPI2_2(%rip), %xmm3 # xmm3 = <1,u>
|
|
# AlignMOV convert to UnAlignMOV
|
|
movq %rbp, 8(%rsp) # 8-byte Spill
|
|
jmp .LBB2_7
|
|
.p2align 5, 0x90
|
|
.LBB2_8: # in Loop: Header=BB2_7 Depth=1
|
|
vxorps %xmm4, %xmm4, %xmm4
|
|
vxorps %xmm8, %xmm8, %xmm8
|
|
vxorps %xmm13, %xmm13, %xmm13
|
|
vxorps %xmm5, %xmm5, %xmm5
|
|
vxorps %xmm9, %xmm9, %xmm9
|
|
vxorps %xmm12, %xmm12, %xmm12
|
|
.LBB2_9: # in Loop: Header=BB2_7 Depth=1
|
|
movq 176(%rbp), %rax
|
|
vshuff64x2 $136, %zmm13, %zmm12, %zmm6 # zmm6 = zmm12[0,1,4,5],zmm13[0,1,4,5]
|
|
vshuff64x2 $221, %zmm13, %zmm12, %zmm7 # zmm7 = zmm12[2,3,6,7],zmm13[2,3,6,7]
|
|
vaddps %zmm7, %zmm6, %zmm6
|
|
vpermilpd $85, %zmm6, %zmm7 # zmm7 = zmm6[1,0,3,2,5,4,7,6]
|
|
vaddps %zmm7, %zmm6, %zmm6
|
|
vpermilps $177, %zmm6, %zmm7 # zmm7 = zmm6[1,0,3,2,5,4,7,6,9,8,11,10,13,12,15,14]
|
|
vaddps %zmm7, %zmm6, %zmm6
|
|
vcompressps %zmm6, %zmm6 {%k1} {z}
|
|
vaddps (%rax,%r9,4), %xmm6, %xmm6
|
|
vmovups %xmm6, (%rax,%r9,4) # AlignMOV convert to UnAlignMOV
|
|
vshuff64x2 $136, %zmm8, %zmm9, %zmm6 # zmm6 = zmm9[0,1,4,5],zmm8[0,1,4,5]
|
|
vshuff64x2 $221, %zmm8, %zmm9, %zmm7 # zmm7 = zmm9[2,3,6,7],zmm8[2,3,6,7]
|
|
vaddps %zmm7, %zmm6, %zmm6
|
|
vpermilpd $85, %zmm6, %zmm7 # zmm7 = zmm6[1,0,3,2,5,4,7,6]
|
|
vaddps %zmm7, %zmm6, %zmm6
|
|
vpermilps $177, %zmm6, %zmm7 # zmm7 = zmm6[1,0,3,2,5,4,7,6,9,8,11,10,13,12,15,14]
|
|
vaddps %zmm7, %zmm6, %zmm6
|
|
vcompressps %zmm6, %zmm6 {%k1} {z}
|
|
vaddps 32(%rax,%r9,4), %xmm6, %xmm6
|
|
vmovups %xmm6, 32(%rax,%r9,4) # AlignMOV convert to UnAlignMOV
|
|
vshuff64x2 $136, %zmm4, %zmm5, %zmm6 # zmm6 = zmm5[0,1,4,5],zmm4[0,1,4,5]
|
|
vshuff64x2 $221, %zmm4, %zmm5, %zmm4 # zmm4 = zmm5[2,3,6,7],zmm4[2,3,6,7]
|
|
vaddps %zmm4, %zmm6, %zmm4
|
|
vpermilpd $85, %zmm4, %zmm5 # zmm5 = zmm4[1,0,3,2,5,4,7,6]
|
|
vaddps %zmm5, %zmm4, %zmm4
|
|
vpermilps $177, %zmm4, %zmm5 # zmm5 = zmm4[1,0,3,2,5,4,7,6,9,8,11,10,13,12,15,14]
|
|
vaddps %zmm5, %zmm4, %zmm4
|
|
vcompressps %zmm4, %zmm4 {%k1} {z}
|
|
vaddps 64(%rax,%r9,4), %xmm4, %xmm4
|
|
vmovups %xmm4, 64(%rax,%r9,4) # AlignMOV convert to UnAlignMOV
|
|
vpinsrq $1, %r10, %xmm3, %xmm4
|
|
vpaddq (%r13), %xmm4, %xmm4
|
|
vmovdqu %xmm4, (%r13)
|
|
addq %r10, 16(%r13)
|
|
incq %r12
|
|
movslq 20(%rbp), %rax
|
|
cmpq %rax, %r12
|
|
jge .LBB2_10
|
|
.LBB2_7: # =>This Loop Header: Depth=1
|
|
# Child Loop BB2_12 Depth 2
|
|
leal (,%r12,4), %r9d
|
|
movl %r9d, %eax
|
|
andl $2147483640, %eax # imm = 0x7FFFFFF8
|
|
leal (%rax,%rax,2), %eax
|
|
andl $4, %r9d
|
|
orl %eax, %r9d
|
|
movq 24(%r14), %rax
|
|
movslq (%rax,%r12,4), %r10
|
|
testq %r10, %r10
|
|
jle .LBB2_8
|
|
# %bb.11: # in Loop: Header=BB2_7 Depth=1
|
|
movq 160(%rbp), %rsi
|
|
movq 8(%r14), %rax
|
|
vbroadcastss (%rsi,%r9,4), %ymm4
|
|
vbroadcastss 4(%rsi,%r9,4), %ymm5
|
|
vbroadcastss 8(%rsi,%r9,4), %ymm7
|
|
vinsertf64x4 $1, %ymm5, %zmm4, %zmm6
|
|
vbroadcastss 12(%rsi,%r9,4), %ymm4
|
|
vinsertf64x4 $1, %ymm4, %zmm7, %zmm7
|
|
vbroadcastss 32(%rsi,%r9,4), %ymm4
|
|
vbroadcastss 36(%rsi,%r9,4), %ymm5
|
|
vbroadcastss 40(%rsi,%r9,4), %ymm8
|
|
vbroadcastss 44(%rsi,%r9,4), %ymm9
|
|
vinsertf64x4 $1, %ymm5, %zmm4, %zmm10
|
|
vinsertf64x4 $1, %ymm9, %zmm8, %zmm11
|
|
vbroadcastss 64(%rsi,%r9,4), %ymm4
|
|
vbroadcastss 68(%rsi,%r9,4), %ymm5
|
|
vinsertf64x4 $1, %ymm5, %zmm4, %zmm14
|
|
vbroadcastss 72(%rsi,%r9,4), %ymm4
|
|
vbroadcastss 76(%rsi,%r9,4), %ymm5
|
|
vinsertf64x4 $1, %ymm5, %zmm4, %zmm15
|
|
movl %r10d, %r10d
|
|
movq %r14, %r15
|
|
movl 16(%r14), %ecx
|
|
imull %r12d, %ecx
|
|
movslq %ecx, %rcx
|
|
leaq (%rax,%rcx,4), %r11
|
|
vxorps %xmm12, %xmm12, %xmm12
|
|
xorl %eax, %eax
|
|
vxorps %xmm9, %xmm9, %xmm9
|
|
vxorps %xmm5, %xmm5, %xmm5
|
|
vxorps %xmm13, %xmm13, %xmm13
|
|
vxorps %xmm8, %xmm8, %xmm8
|
|
vxorps %xmm4, %xmm4, %xmm4
|
|
.p2align 4, 0x90
|
|
movl $111, %ebx # OSACA START MARKER
|
|
.byte 100 # OSACA START MARKER
|
|
.byte 103 # OSACA START MARKER
|
|
.byte 144 # OSACA START MARKER
|
|
# LLVM-MCA-BEGIN
|
|
# pointer_increment=256 a23042eac7d8a1e13e9ff886fc02a80e
|
|
.LBB2_12: # Parent Loop BB2_7 Depth=1
|
|
# => This Inner Loop Header: Depth=2
|
|
movslq (%r11,%rax,4), %rcx
|
|
leaq (%rcx,%rcx,2), %rdx
|
|
shlq $5, %rdx
|
|
vmovupd (%rsi,%rdx), %zmm16
|
|
vbroadcastf64x4 64(%rsi,%rdx), %zmm20 # zmm20 = mem[0,1,2,3,0,1,2,3]
|
|
vbroadcastf64x4 (%rsi,%rdx), %zmm19 # zmm19 = mem[0,1,2,3,0,1,2,3]
|
|
vshuff64x2 $238, %zmm16, %zmm16, %zmm21 # zmm21 = zmm16[4,5,6,7,4,5,6,7]
|
|
vsubps %zmm19, %zmm6, %zmm18
|
|
vsubps %zmm21, %zmm10, %zmm17
|
|
vsubps %zmm20, %zmm14, %zmm16
|
|
vmulps %zmm16, %zmm16, %zmm22
|
|
vfmadd231ps %zmm17, %zmm17, %zmm22 # zmm22 = (zmm17 * zmm17) + zmm22
|
|
vfmadd231ps %zmm18, %zmm18, %zmm22 # zmm22 = (zmm18 * zmm18) + zmm22
|
|
vrcp14ps %zmm22, %zmm23
|
|
vmulps %zmm23, %zmm26, %zmm24
|
|
vmulps %zmm24, %zmm23, %zmm24
|
|
vmulps %zmm24, %zmm23, %zmm24
|
|
vaddps %zmm1, %zmm24, %zmm25
|
|
vmulps %zmm23, %zmm27, %zmm23
|
|
vmulps %zmm25, %zmm23, %zmm23
|
|
vmulps %zmm23, %zmm24, %zmm23
|
|
leal (%rcx,%rcx), %edx
|
|
xorl %edi, %edi
|
|
xorl %ebp, %ebp
|
|
cmpq %rdx, %r12
|
|
setne %dil
|
|
leal 1(%rcx,%rcx), %ecx
|
|
sete %bpl
|
|
xorl %edx, %edx
|
|
xorl %ebx, %ebx
|
|
cmpq %rcx, %r12
|
|
sete %dl
|
|
movl $0, %ecx
|
|
setne %bl
|
|
cmovel %r8d, %ecx
|
|
movl %ebx, %r14d
|
|
shll $4, %r14d
|
|
subl %ebp, %r14d
|
|
leal (%rcx,%rdi,2), %ecx
|
|
shll $8, %ecx
|
|
addl $239, %r14d
|
|
addl $-768, %ecx # imm = 0xFD00
|
|
orl %r14d, %ecx
|
|
kmovd %ecx, %k2
|
|
vcmpltps %zmm0, %zmm22, %k2 {%k2}
|
|
vsubps %zmm21, %zmm11, %zmm21
|
|
vsubps %zmm20, %zmm15, %zmm20
|
|
vsubps %zmm19, %zmm7, %zmm19
|
|
vmulps %zmm2, %zmm23, %zmm22
|
|
vfmadd231ps %zmm22, %zmm18, %zmm12 {%k2} # zmm12 {%k2} = (zmm18 * zmm22) + zmm12
|
|
vmulps %zmm20, %zmm20, %zmm18
|
|
vfmadd231ps %zmm21, %zmm21, %zmm18 # zmm18 = (zmm21 * zmm21) + zmm18
|
|
vfmadd231ps %zmm19, %zmm19, %zmm18 # zmm18 = (zmm19 * zmm19) + zmm18
|
|
vfmadd231ps %zmm22, %zmm17, %zmm9 {%k2} # zmm9 {%k2} = (zmm17 * zmm22) + zmm9
|
|
vrcp14ps %zmm18, %zmm17
|
|
vfmadd231ps %zmm22, %zmm16, %zmm5 {%k2} # zmm5 {%k2} = (zmm16 * zmm22) + zmm5
|
|
vmulps %zmm17, %zmm26, %zmm16
|
|
vmulps %zmm16, %zmm17, %zmm16
|
|
vmulps %zmm16, %zmm17, %zmm16
|
|
vaddps %zmm1, %zmm16, %zmm22
|
|
vmulps %zmm17, %zmm27, %zmm17
|
|
vmulps %zmm22, %zmm17, %zmm17
|
|
vmulps %zmm17, %zmm16, %zmm16
|
|
shll $6, %ebx
|
|
leal (%rbx,%rdi,4), %ecx
|
|
shll $7, %edx
|
|
leal (%rdx,%rdi,8), %edx
|
|
shll $8, %edx
|
|
addl %edx, %ecx
|
|
addl $-2117, %ecx # imm = 0xF7BB
|
|
kmovd %ecx, %k2
|
|
vcmpltps %zmm0, %zmm18, %k2 {%k2}
|
|
vmulps %zmm2, %zmm16, %zmm16
|
|
vfmadd231ps %zmm16, %zmm19, %zmm13 {%k2} # zmm13 {%k2} = (zmm19 * zmm16) + zmm13
|
|
vfmadd231ps %zmm16, %zmm21, %zmm8 {%k2} # zmm8 {%k2} = (zmm21 * zmm16) + zmm8
|
|
vfmadd231ps %zmm16, %zmm20, %zmm4 {%k2} # zmm4 {%k2} = (zmm20 * zmm16) + zmm4
|
|
incq %rax
|
|
cmpq %rax, %r10
|
|
jne .LBB2_12
|
|
# LLVM-MCA-END
|
|
movl $222, %ebx # OSACA END MARKER
|
|
.byte 100 # OSACA END MARKER
|
|
.byte 103 # OSACA END MARKER
|
|
.byte 144 # OSACA END MARKER
|
|
# %bb.13: # in Loop: Header=BB2_7 Depth=1
|
|
movq %r15, %r14
|
|
movq 8(%rsp), %rbp # 8-byte Reload
|
|
jmp .LBB2_9
|
|
.p2align 5, 0x90
|
|
.LBB2_10:
|
|
movl $.L.str.1, %edi
|
|
vzeroupper
|
|
callq likwid_markerStopRegion
|
|
xorl %eax, %eax
|
|
callq getTimeStamp
|
|
vmovsd %xmm0, 8(%rsp) # 8-byte Spill
|
|
movl $.L.str.5, %edi
|
|
xorl %eax, %eax
|
|
callq debug_printf
|
|
vmovsd 8(%rsp), %xmm0 # 8-byte Reload
|
|
# xmm0 = mem[0],zero
|
|
vsubsd 24(%rsp), %xmm0, %xmm0 # 8-byte Folded Reload
|
|
addq $168, %rsp
|
|
.cfi_def_cfa_offset 56
|
|
popq %rbx
|
|
.cfi_def_cfa_offset 48
|
|
popq %r12
|
|
.cfi_def_cfa_offset 40
|
|
popq %r13
|
|
.cfi_def_cfa_offset 32
|
|
popq %r14
|
|
.cfi_def_cfa_offset 24
|
|
popq %r15
|
|
.cfi_def_cfa_offset 16
|
|
popq %rbp
|
|
.cfi_def_cfa_offset 8
|
|
retq
|
|
.Lfunc_end2:
|
|
.size computeForceLJ_2xnn_full, .Lfunc_end2-computeForceLJ_2xnn_full
|
|
.cfi_endproc
|
|
# -- End function
|
|
.globl computeForceLJ_2xnn # -- Begin function computeForceLJ_2xnn
|
|
.p2align 4, 0x90
|
|
.type computeForceLJ_2xnn,@function
|
|
computeForceLJ_2xnn: #
|
|
.cfi_startproc
|
|
# %bb.0:
|
|
cmpl $0, 32(%rdx)
|
|
je .LBB3_2
|
|
# %bb.1:
|
|
jmp computeForceLJ_2xnn_half # TAILCALL
|
|
.p2align 5, 0x90
|
|
.LBB3_2:
|
|
jmp computeForceLJ_2xnn_full # TAILCALL
|
|
.Lfunc_end3:
|
|
.size computeForceLJ_2xnn, .Lfunc_end3-computeForceLJ_2xnn
|
|
.cfi_endproc
|
|
# -- End function
|
|
.section .rodata.cst8,"aM",@progbits,8
|
|
.p2align 3 # -- Begin function computeForceLJ_4xn_half
|
|
.LCPI4_0:
|
|
.quad 0x41cdcd6500000000 # 1.0E+9
|
|
.section .rodata.cst4,"aM",@progbits,4
|
|
.p2align 2
|
|
.LCPI4_1:
|
|
.long 0xbf000000 # -0.5
|
|
.LCPI4_2:
|
|
.long 0x42400000 # 48
|
|
.text
|
|
.globl computeForceLJ_4xn_half
|
|
.p2align 4, 0x90
|
|
.type computeForceLJ_4xn_half,@function
|
|
computeForceLJ_4xn_half: #
|
|
.cfi_startproc
|
|
# %bb.0:
|
|
pushq %r15
|
|
.cfi_def_cfa_offset 16
|
|
pushq %r14
|
|
.cfi_def_cfa_offset 24
|
|
pushq %r13
|
|
.cfi_def_cfa_offset 32
|
|
pushq %r12
|
|
.cfi_def_cfa_offset 40
|
|
pushq %rbx
|
|
.cfi_def_cfa_offset 48
|
|
subq $576, %rsp # imm = 0x240
|
|
.cfi_def_cfa_offset 624
|
|
.cfi_offset %rbx, -48
|
|
.cfi_offset %r12, -40
|
|
.cfi_offset %r13, -32
|
|
.cfi_offset %r14, -24
|
|
.cfi_offset %r15, -16
|
|
movq %rdx, %r14
|
|
movq %rsi, %r15
|
|
movq %rdi, %r12
|
|
movl $.L.str.6, %edi
|
|
xorl %eax, %eax
|
|
callq debug_printf
|
|
vmovss 108(%r12), %xmm0 # xmm0 = mem[0],zero,zero,zero
|
|
vmovss %xmm0, 64(%rsp) # 4-byte Spill
|
|
vbroadcastss 48(%r12), %zmm0
|
|
vmovups %zmm0, 512(%rsp) # 64-byte Spill
|
|
vbroadcastss 40(%r12), %zmm0
|
|
vmovupd %zmm0, 448(%rsp) # 64-byte Spill
|
|
movl 20(%r15), %r11d
|
|
testl %r11d, %r11d
|
|
jle .LBB4_5
|
|
# %bb.1:
|
|
movq 176(%r15), %r9
|
|
movq 192(%r15), %r10
|
|
decq %r11
|
|
leaq 64(%r9), %r8
|
|
xorl %edi, %edi
|
|
vxorpd %xmm0, %xmm0, %xmm0
|
|
vxorps %xmm1, %xmm1, %xmm1
|
|
jmp .LBB4_2
|
|
.p2align 5, 0x90
|
|
.LBB4_16: # in Loop: Header=BB4_2 Depth=1
|
|
cmpq %r11, %rdi
|
|
leaq 1(%rdi), %rdi
|
|
je .LBB4_5
|
|
.LBB4_2: # =>This Loop Header: Depth=1
|
|
# Child Loop BB4_11 Depth 2
|
|
# Child Loop BB4_15 Depth 2
|
|
leaq (%rdi,%rdi,8), %rax
|
|
leaq (%rax,%rax,2), %rax
|
|
addq %rdi, %rax
|
|
movl (%r10,%rax), %ecx
|
|
testl %ecx, %ecx
|
|
jle .LBB4_16
|
|
# %bb.3: # in Loop: Header=BB4_2 Depth=1
|
|
leal (,%rdi,4), %esi
|
|
movl %esi, %eax
|
|
andl $2147483640, %eax # imm = 0x7FFFFFF8
|
|
leal (%rax,%rax,2), %eax
|
|
andl $4, %esi
|
|
orl %eax, %esi
|
|
cmpl $7, %ecx
|
|
ja .LBB4_10
|
|
# %bb.4: # in Loop: Header=BB4_2 Depth=1
|
|
movl %ecx, %ebx
|
|
andl $-8, %ebx
|
|
cmpq %rcx, %rbx
|
|
jae .LBB4_16
|
|
jmp .LBB4_14
|
|
.p2align 5, 0x90
|
|
.LBB4_10: # in Loop: Header=BB4_2 Depth=1
|
|
leaq (,%rcx,4), %rbx
|
|
andq $-32, %rbx
|
|
movl %esi, %r13d
|
|
leaq (%r9,%r13,4), %rax
|
|
xorl %edx, %edx
|
|
.p2align 4, 0x90
|
|
.LBB4_11: # Parent Loop BB4_2 Depth=1
|
|
# => This Inner Loop Header: Depth=2
|
|
vmovupd %ymm0, (%rax,%rdx)
|
|
addq $32, %rdx
|
|
cmpq %rdx, %rbx
|
|
jne .LBB4_11
|
|
# %bb.12: # in Loop: Header=BB4_2 Depth=1
|
|
movl %ecx, %ebx
|
|
andl $-8, %ebx
|
|
addq %rbx, %r13
|
|
vmovups %zmm1, (%r9,%r13,4)
|
|
cmpq %rcx, %rbx
|
|
jae .LBB4_16
|
|
.LBB4_14: # in Loop: Header=BB4_2 Depth=1
|
|
movl %esi, %eax
|
|
leaq (%r8,%rax,4), %rdx
|
|
.p2align 4, 0x90
|
|
.LBB4_15: # Parent Loop BB4_2 Depth=1
|
|
# => This Inner Loop Header: Depth=2
|
|
movl $0, -64(%rdx,%rbx,4)
|
|
movl $0, -32(%rdx,%rbx,4)
|
|
movl $0, (%rdx,%rbx,4)
|
|
incq %rbx
|
|
cmpq %rbx, %rcx
|
|
jne .LBB4_15
|
|
jmp .LBB4_16
|
|
.p2align 5, 0x90
|
|
.LBB4_5:
|
|
xorl %eax, %eax
|
|
vzeroupper
|
|
callq getTimeStamp
|
|
vmovsd %xmm0, (%rsp) # 8-byte Spill
|
|
movl $.L.str.1, %edi
|
|
callq likwid_markerStartRegion
|
|
cmpl $0, 20(%r15)
|
|
jg .LBB4_6
|
|
# %bb.17:
|
|
movl $.L.str.1, %edi
|
|
callq likwid_markerStopRegion
|
|
xorl %eax, %eax
|
|
callq getTimeStamp
|
|
vmovsd 184(%r12), %xmm3 # xmm3 = mem[0],zero
|
|
vsubsd (%rsp), %xmm0, %xmm1 # 8-byte Folded Reload
|
|
vmovsd %xmm1, (%rsp) # 8-byte Spill
|
|
vmulsd .LCPI4_0(%rip), %xmm3, %xmm0
|
|
vmulsd %xmm1, %xmm0, %xmm0
|
|
vxorpd %xmm2, %xmm2, %xmm2
|
|
vdivsd %xmm2, %xmm0, %xmm2
|
|
movl $.L.str.4, %edi
|
|
xorl %esi, %esi
|
|
vmovapd %xmm3, %xmm0
|
|
movb $3, %al
|
|
callq printf
|
|
movl $.L.str.7, %edi
|
|
xorl %eax, %eax
|
|
callq debug_printf
|
|
vmovsd (%rsp), %xmm0 # 8-byte Reload
|
|
# xmm0 = mem[0],zero
|
|
addq $576, %rsp # imm = 0x240
|
|
.cfi_def_cfa_offset 48
|
|
popq %rbx
|
|
.cfi_def_cfa_offset 40
|
|
popq %r12
|
|
.cfi_def_cfa_offset 32
|
|
popq %r13
|
|
.cfi_def_cfa_offset 24
|
|
popq %r14
|
|
.cfi_def_cfa_offset 16
|
|
popq %r15
|
|
.cfi_def_cfa_offset 8
|
|
retq
|
|
.p2align 5, 0x90
|
|
.LBB4_6:
|
|
.cfi_def_cfa_offset 624
|
|
movq 24(%r14), %rax
|
|
movl (%rax), %r10d
|
|
testl %r10d, %r10d
|
|
jle .LBB4_18
|
|
# %bb.7:
|
|
vmovss 64(%rsp), %xmm0 # 4-byte Reload
|
|
# xmm0 = mem[0],zero,zero,zero
|
|
vmulss %xmm0, %xmm0, %xmm0
|
|
vbroadcastss %xmm0, %zmm0
|
|
movq 160(%r15), %rdi
|
|
movq 8(%r14), %r11
|
|
vbroadcastss (%rdi), %zmm1
|
|
vmovups %zmm1, (%rsp) # 64-byte Spill
|
|
vbroadcastss 4(%rdi), %zmm1
|
|
vmovups %zmm1, 64(%rsp) # 64-byte Spill
|
|
vbroadcastss 8(%rdi), %zmm1
|
|
vmovups %zmm1, 384(%rsp) # 64-byte Spill
|
|
vbroadcastss 12(%rdi), %zmm1
|
|
vmovups %zmm1, 320(%rsp) # 64-byte Spill
|
|
vbroadcastss 32(%rdi), %zmm1
|
|
vmovups %zmm1, 256(%rsp) # 64-byte Spill
|
|
vbroadcastss 36(%rdi), %zmm1
|
|
vmovups %zmm1, 192(%rsp) # 64-byte Spill
|
|
vbroadcastss 40(%rdi), %zmm1
|
|
vmovups %zmm1, 128(%rsp) # 64-byte Spill
|
|
vbroadcastss 44(%rdi), %zmm8
|
|
vbroadcastss 64(%rdi), %zmm9
|
|
vbroadcastss 68(%rdi), %zmm10
|
|
vbroadcastss 72(%rdi), %zmm11
|
|
vbroadcastss 76(%rdi), %zmm12
|
|
decq %r10
|
|
xorl %edx, %edx
|
|
movl $248, %r8d
|
|
movl $240, %r9d
|
|
vbroadcastss .LCPI4_1(%rip), %zmm13 # zmm13 = [-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1]
|
|
vbroadcastss .LCPI4_2(%rip), %zmm14 # zmm14 = [4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1,4.8E+1]
|
|
vmovups 512(%rsp), %zmm6 # 64-byte Reload
|
|
vmovups 448(%rsp), %zmm7 # 64-byte Reload
|
|
.p2align 4, 0x90
|
|
.LBB4_8: # =>This Inner Loop Header: Depth=1
|
|
movslq (%r11,%rdx,4), %rax
|
|
movq %rax, %rsi
|
|
shlq $5, %rsi
|
|
leaq (%rsi,%rsi,2), %rbx
|
|
vmovups (%rdi,%rbx), %zmm15 # AlignMOV convert to UnAlignMOV
|
|
vmovups 32(%rdi,%rbx), %zmm16 # AlignMOV convert to UnAlignMOV
|
|
vmovups 64(%rdi,%rbx), %zmm27 # AlignMOV convert to UnAlignMOV
|
|
vmovups (%rsp), %zmm1 # 64-byte Reload
|
|
vsubps %zmm15, %zmm1, %zmm24
|
|
vmovups 256(%rsp), %zmm1 # 64-byte Reload
|
|
vsubps %zmm16, %zmm1, %zmm25
|
|
vsubps %zmm27, %zmm9, %zmm26
|
|
vmovups 64(%rsp), %zmm1 # 64-byte Reload
|
|
vsubps %zmm15, %zmm1, %zmm21
|
|
vmovups 192(%rsp), %zmm1 # 64-byte Reload
|
|
vsubps %zmm16, %zmm1, %zmm22
|
|
vsubps %zmm27, %zmm10, %zmm23
|
|
vmovups 384(%rsp), %zmm1 # 64-byte Reload
|
|
vsubps %zmm15, %zmm1, %zmm17
|
|
vmovups 128(%rsp), %zmm1 # 64-byte Reload
|
|
vsubps %zmm16, %zmm1, %zmm19
|
|
vsubps %zmm27, %zmm11, %zmm20
|
|
vmovups 320(%rsp), %zmm1 # 64-byte Reload
|
|
vsubps %zmm15, %zmm1, %zmm18
|
|
vsubps %zmm16, %zmm8, %zmm16
|
|
vsubps %zmm27, %zmm12, %zmm15
|
|
vmulps %zmm26, %zmm26, %zmm27
|
|
vfmadd231ps %zmm25, %zmm25, %zmm27 # zmm27 = (zmm25 * zmm25) + zmm27
|
|
vfmadd231ps %zmm24, %zmm24, %zmm27 # zmm27 = (zmm24 * zmm24) + zmm27
|
|
vmulps %zmm23, %zmm23, %zmm28
|
|
vfmadd231ps %zmm22, %zmm22, %zmm28 # zmm28 = (zmm22 * zmm22) + zmm28
|
|
vfmadd231ps %zmm21, %zmm21, %zmm28 # zmm28 = (zmm21 * zmm21) + zmm28
|
|
vmulps %zmm20, %zmm20, %zmm29
|
|
vfmadd231ps %zmm19, %zmm19, %zmm29 # zmm29 = (zmm19 * zmm19) + zmm29
|
|
vfmadd231ps %zmm17, %zmm17, %zmm29 # zmm29 = (zmm17 * zmm17) + zmm29
|
|
vmulps %zmm15, %zmm15, %zmm30
|
|
vfmadd231ps %zmm16, %zmm16, %zmm30 # zmm30 = (zmm16 * zmm16) + zmm30
|
|
vrcp14ps %zmm27, %zmm31
|
|
vrcp14ps %zmm28, %zmm1
|
|
vrcp14ps %zmm29, %zmm2
|
|
vfmadd231ps %zmm18, %zmm18, %zmm30 # zmm30 = (zmm18 * zmm18) + zmm30
|
|
vrcp14ps %zmm30, %zmm3
|
|
vmulps %zmm31, %zmm6, %zmm4
|
|
vmulps %zmm4, %zmm31, %zmm4
|
|
vmulps %zmm4, %zmm31, %zmm4
|
|
vaddps %zmm13, %zmm4, %zmm5
|
|
vmulps %zmm31, %zmm7, %zmm31
|
|
vmulps %zmm5, %zmm31, %zmm5
|
|
vmulps %zmm1, %zmm6, %zmm31
|
|
vmulps %zmm31, %zmm1, %zmm31
|
|
vmulps %zmm31, %zmm1, %zmm31
|
|
vmulps %zmm5, %zmm4, %zmm4
|
|
vaddps %zmm13, %zmm31, %zmm5
|
|
vmulps %zmm1, %zmm7, %zmm1
|
|
vmulps %zmm5, %zmm1, %zmm1
|
|
vmulps %zmm2, %zmm6, %zmm5
|
|
vmulps %zmm5, %zmm2, %zmm5
|
|
vmulps %zmm5, %zmm2, %zmm5
|
|
vmulps %zmm1, %zmm31, %zmm1
|
|
vaddps %zmm13, %zmm5, %zmm31
|
|
vmulps %zmm2, %zmm7, %zmm2
|
|
vmulps %zmm31, %zmm2, %zmm2
|
|
vmulps %zmm3, %zmm6, %zmm31
|
|
vmulps %zmm31, %zmm3, %zmm31
|
|
vmulps %zmm31, %zmm3, %zmm31
|
|
vmulps %zmm2, %zmm5, %zmm2
|
|
vaddps %zmm13, %zmm31, %zmm5
|
|
vmulps %zmm3, %zmm7, %zmm3
|
|
vmulps %zmm5, %zmm3, %zmm3
|
|
vmulps %zmm3, %zmm31, %zmm3
|
|
xorl %esi, %esi
|
|
xorl %edi, %edi
|
|
testl $2147483647, %eax # imm = 0x7FFFFFFF
|
|
sete %sil
|
|
setne %dil
|
|
movl $255, %eax
|
|
cmovel %r8d, %eax
|
|
movl $255, %ecx
|
|
cmovel %r9d, %ecx
|
|
xorl $255, %esi
|
|
kmovd %esi, %k1
|
|
vcmpltps %zmm0, %zmm27, %k1 {%k1}
|
|
vmulps %zmm14, %zmm4, %zmm4
|
|
vmulps %zmm4, %zmm24, %zmm5 {%k1} {z}
|
|
vmulps %zmm4, %zmm25, %zmm24 {%k1} {z}
|
|
vmulps %zmm4, %zmm26, %zmm4 {%k1} {z}
|
|
leal (%rdi,%rdi,2), %esi
|
|
orl $252, %esi
|
|
kmovd %esi, %k1
|
|
vcmpltps %zmm0, %zmm28, %k1 {%k1}
|
|
vmulps %zmm14, %zmm1, %zmm1
|
|
vmulps %zmm1, %zmm21, %zmm21 {%k1} {z}
|
|
vaddps %zmm21, %zmm5, %zmm5
|
|
vmulps %zmm1, %zmm22, %zmm21 {%k1} {z}
|
|
vaddps %zmm21, %zmm24, %zmm21
|
|
vmulps %zmm1, %zmm23, %zmm1 {%k1} {z}
|
|
vaddps %zmm1, %zmm4, %zmm1
|
|
kmovd %eax, %k1
|
|
vcmpltps %zmm0, %zmm29, %k1 {%k1}
|
|
vmulps %zmm14, %zmm2, %zmm2
|
|
vmulps %zmm2, %zmm17, %zmm4 {%k1} {z}
|
|
vmulps %zmm2, %zmm19, %zmm17 {%k1} {z}
|
|
vmulps %zmm2, %zmm20, %zmm2 {%k1} {z}
|
|
kmovd %ecx, %k1
|
|
vcmpltps %zmm0, %zmm30, %k1 {%k1}
|
|
vmulps %zmm14, %zmm3, %zmm3
|
|
vmulps %zmm3, %zmm18, %zmm18 {%k1} {z}
|
|
vaddps %zmm18, %zmm4, %zmm4
|
|
vaddps %zmm4, %zmm5, %zmm4
|
|
vmulps %zmm3, %zmm16, %zmm5 {%k1} {z}
|
|
vaddps %zmm5, %zmm17, %zmm5
|
|
vaddps %zmm5, %zmm21, %zmm5
|
|
vmulps %zmm3, %zmm15, %zmm3 {%k1} {z}
|
|
movq 176(%r15), %rax
|
|
vaddps %zmm3, %zmm2, %zmm2
|
|
vmovups (%rax,%rbx), %zmm3 # AlignMOV convert to UnAlignMOV
|
|
vsubps %zmm4, %zmm3, %zmm3
|
|
vmovups %zmm3, (%rax,%rbx) # AlignMOV convert to UnAlignMOV
|
|
vaddps %zmm2, %zmm1, %zmm1
|
|
vmovups 32(%rax,%rbx), %zmm2 # AlignMOV convert to UnAlignMOV
|
|
vsubps %zmm5, %zmm2, %zmm2
|
|
vmovups %zmm2, 32(%rax,%rbx) # AlignMOV convert to UnAlignMOV
|
|
vmovups 64(%rax,%rbx), %zmm2 # AlignMOV convert to UnAlignMOV
|
|
vsubps %zmm1, %zmm2, %zmm1
|
|
vmovups %zmm1, 64(%rax,%rbx) # AlignMOV convert to UnAlignMOV
|
|
cmpq %rdx, %r10
|
|
je .LBB4_18
|
|
# %bb.9: # in Loop: Header=BB4_8 Depth=1
|
|
movq 160(%r15), %rdi
|
|
incq %rdx
|
|
jmp .LBB4_8
|
|
.p2align 5, 0x90
|
|
.LBB4_18:
|
|
vzeroupper
|
|
callq simd_incr_reduced_sum
|
|
.Lfunc_end4:
|
|
.size computeForceLJ_4xn_half, .Lfunc_end4-computeForceLJ_4xn_half
|
|
.cfi_endproc
|
|
# -- End function
|
|
.p2align 4, 0x90 # -- Begin function simd_incr_reduced_sum
|
|
.type simd_incr_reduced_sum,@function
|
|
simd_incr_reduced_sum: #
|
|
.cfi_startproc
|
|
# %bb.0:
|
|
pushq %rax
|
|
.cfi_def_cfa_offset 16
|
|
movq stderr(%rip), %rcx
|
|
movl $.L.str.8, %edi
|
|
movl $92, %esi
|
|
movl $1, %edx
|
|
callq fwrite@PLT
|
|
movl $-1, %edi
|
|
callq exit
|
|
.Lfunc_end5:
|
|
.size simd_incr_reduced_sum, .Lfunc_end5-simd_incr_reduced_sum
|
|
.cfi_endproc
|
|
# -- End function
|
|
.section .rodata.cst8,"aM",@progbits,8
|
|
.p2align 3 # -- Begin function computeForceLJ_4xn_full
|
|
.LCPI6_0:
|
|
.quad 0x41cdcd6500000000 # 1.0E+9
|
|
.text
|
|
.globl computeForceLJ_4xn_full
|
|
.p2align 4, 0x90
|
|
.type computeForceLJ_4xn_full,@function
|
|
computeForceLJ_4xn_full: #
|
|
.cfi_startproc
|
|
# %bb.0:
|
|
pushq %r15
|
|
.cfi_def_cfa_offset 16
|
|
pushq %r14
|
|
.cfi_def_cfa_offset 24
|
|
pushq %r12
|
|
.cfi_def_cfa_offset 32
|
|
pushq %rbx
|
|
.cfi_def_cfa_offset 40
|
|
pushq %rax
|
|
.cfi_def_cfa_offset 48
|
|
.cfi_offset %rbx, -40
|
|
.cfi_offset %r12, -32
|
|
.cfi_offset %r14, -24
|
|
.cfi_offset %r15, -16
|
|
movq %rsi, %r15
|
|
movq %rdi, %r14
|
|
movl $.L.str.6, %edi
|
|
xorl %eax, %eax
|
|
callq debug_printf
|
|
movl 20(%r15), %r11d
|
|
testl %r11d, %r11d
|
|
jle .LBB6_5
|
|
# %bb.1:
|
|
movq 176(%r15), %r9
|
|
movq 192(%r15), %r10
|
|
decq %r11
|
|
leaq 64(%r9), %r8
|
|
xorl %edi, %edi
|
|
vxorpd %xmm0, %xmm0, %xmm0
|
|
vxorps %xmm1, %xmm1, %xmm1
|
|
jmp .LBB6_2
|
|
.p2align 5, 0x90
|
|
.LBB6_13: # in Loop: Header=BB6_2 Depth=1
|
|
cmpq %r11, %rdi
|
|
leaq 1(%rdi), %rdi
|
|
je .LBB6_5
|
|
.LBB6_2: # =>This Loop Header: Depth=1
|
|
# Child Loop BB6_8 Depth 2
|
|
# Child Loop BB6_12 Depth 2
|
|
leaq (%rdi,%rdi,8), %rax
|
|
leaq (%rax,%rax,2), %rax
|
|
addq %rdi, %rax
|
|
movl (%r10,%rax), %ecx
|
|
testl %ecx, %ecx
|
|
jle .LBB6_13
|
|
# %bb.3: # in Loop: Header=BB6_2 Depth=1
|
|
leal (,%rdi,4), %esi
|
|
movl %esi, %eax
|
|
andl $2147483640, %eax # imm = 0x7FFFFFF8
|
|
leal (%rax,%rax,2), %eax
|
|
andl $4, %esi
|
|
orl %eax, %esi
|
|
cmpl $7, %ecx
|
|
ja .LBB6_7
|
|
# %bb.4: # in Loop: Header=BB6_2 Depth=1
|
|
movl %ecx, %ebx
|
|
andl $-8, %ebx
|
|
cmpq %rcx, %rbx
|
|
jae .LBB6_13
|
|
jmp .LBB6_11
|
|
.p2align 5, 0x90
|
|
.LBB6_7: # in Loop: Header=BB6_2 Depth=1
|
|
leaq (,%rcx,4), %rbx
|
|
andq $-32, %rbx
|
|
movl %esi, %r12d
|
|
leaq (%r9,%r12,4), %rax
|
|
xorl %edx, %edx
|
|
.p2align 4, 0x90
|
|
.LBB6_8: # Parent Loop BB6_2 Depth=1
|
|
# => This Inner Loop Header: Depth=2
|
|
vmovupd %ymm0, (%rax,%rdx)
|
|
addq $32, %rdx
|
|
cmpq %rdx, %rbx
|
|
jne .LBB6_8
|
|
# %bb.9: # in Loop: Header=BB6_2 Depth=1
|
|
movl %ecx, %ebx
|
|
andl $-8, %ebx
|
|
addq %rbx, %r12
|
|
vmovups %zmm1, (%r9,%r12,4)
|
|
cmpq %rcx, %rbx
|
|
jae .LBB6_13
|
|
.LBB6_11: # in Loop: Header=BB6_2 Depth=1
|
|
movl %esi, %eax
|
|
leaq (%r8,%rax,4), %rdx
|
|
.p2align 4, 0x90
|
|
.LBB6_12: # Parent Loop BB6_2 Depth=1
|
|
# => This Inner Loop Header: Depth=2
|
|
movl $0, -64(%rdx,%rbx,4)
|
|
movl $0, -32(%rdx,%rbx,4)
|
|
movl $0, (%rdx,%rbx,4)
|
|
incq %rbx
|
|
cmpq %rbx, %rcx
|
|
jne .LBB6_12
|
|
jmp .LBB6_13
|
|
.p2align 5, 0x90
|
|
.LBB6_5:
|
|
xorl %eax, %eax
|
|
vzeroupper
|
|
callq getTimeStamp
|
|
vmovsd %xmm0, (%rsp) # 8-byte Spill
|
|
movl $.L.str.1, %edi
|
|
callq likwid_markerStartRegion
|
|
cmpl $0, 20(%r15)
|
|
jg .LBB6_6
|
|
# %bb.14:
|
|
movl $.L.str.1, %edi
|
|
callq likwid_markerStopRegion
|
|
xorl %eax, %eax
|
|
callq getTimeStamp
|
|
vmovsd 184(%r14), %xmm3 # xmm3 = mem[0],zero
|
|
vsubsd (%rsp), %xmm0, %xmm1 # 8-byte Folded Reload
|
|
vmovsd %xmm1, (%rsp) # 8-byte Spill
|
|
vmulsd .LCPI6_0(%rip), %xmm3, %xmm0
|
|
vmulsd %xmm1, %xmm0, %xmm0
|
|
vxorpd %xmm2, %xmm2, %xmm2
|
|
vdivsd %xmm2, %xmm0, %xmm2
|
|
movl $.L.str.4, %edi
|
|
xorl %esi, %esi
|
|
vmovapd %xmm3, %xmm0
|
|
movb $3, %al
|
|
callq printf
|
|
movl $.L.str.7, %edi
|
|
xorl %eax, %eax
|
|
callq debug_printf
|
|
vmovsd (%rsp), %xmm0 # 8-byte Reload
|
|
# xmm0 = mem[0],zero
|
|
addq $8, %rsp
|
|
.cfi_def_cfa_offset 40
|
|
popq %rbx
|
|
.cfi_def_cfa_offset 32
|
|
popq %r12
|
|
.cfi_def_cfa_offset 24
|
|
popq %r14
|
|
.cfi_def_cfa_offset 16
|
|
popq %r15
|
|
.cfi_def_cfa_offset 8
|
|
retq
|
|
.p2align 5, 0x90
|
|
.LBB6_6:
|
|
.cfi_def_cfa_offset 48
|
|
callq simd_incr_reduced_sum
|
|
.Lfunc_end6:
|
|
.size computeForceLJ_4xn_full, .Lfunc_end6-computeForceLJ_4xn_full
|
|
.cfi_endproc
|
|
# -- End function
|
|
.globl computeForceLJ_4xn # -- Begin function computeForceLJ_4xn
|
|
.p2align 4, 0x90
|
|
.type computeForceLJ_4xn,@function
|
|
computeForceLJ_4xn: #
|
|
.cfi_startproc
|
|
# %bb.0:
|
|
cmpl $0, 32(%rdx)
|
|
je .LBB7_2
|
|
# %bb.1:
|
|
jmp computeForceLJ_4xn_half # TAILCALL
|
|
.p2align 5, 0x90
|
|
.LBB7_2:
|
|
jmp computeForceLJ_4xn_full # TAILCALL
|
|
.Lfunc_end7:
|
|
.size computeForceLJ_4xn, .Lfunc_end7-computeForceLJ_4xn
|
|
.cfi_endproc
|
|
# -- End function
|
|
.type .L.str,@object #
|
|
.section .rodata.str1.1,"aMS",@progbits,1
|
|
.L.str:
|
|
.asciz "computeForceLJ begin\n"
|
|
.size .L.str, 22
|
|
.type .L.str.1,@object #
|
|
.L.str.1:
|
|
.asciz "force"
|
|
.size .L.str.1, 6
|
|
.type .L.str.2,@object #
|
|
.L.str.2:
|
|
.asciz "computeForceLJ end\n"
|
|
.size .L.str.2, 20
|
|
.type .L.str.3,@object #
|
|
.L.str.3:
|
|
.asciz "computeForceLJ_2xnn begin\n"
|
|
.size .L.str.3, 27
|
|
.type .L.str.4,@object #
|
|
.L.str.4:
|
|
.asciz "Its: %u Freq: %f Time: %f\nCy/it: %f\n"
|
|
.size .L.str.4, 39
|
|
.type .L.str.5,@object #
|
|
.L.str.5:
|
|
.asciz "computeForceLJ_2xnn end\n"
|
|
.size .L.str.5, 25
|
|
.type .L.str.6,@object #
|
|
.L.str.6:
|
|
.asciz "computeForceLJ_4xn begin\n"
|
|
.size .L.str.6, 26
|
|
.type .L.str.7,@object #
|
|
.L.str.7:
|
|
.asciz "computeForceLJ_4xn end\n"
|
|
.size .L.str.7, 24
|
|
.type .L.str.8,@object #
|
|
.L.str.8:
|
|
.asciz "simd_h_reduce_sum(): Called with AVX512 intrinsics and single-precision which is not valid!\n"
|
|
.size .L.str.8, 93
|
|
.ident "Intel(R) oneAPI DPC++/C++ Compiler 2022.1.0 (2022.1.0.20220316)"
|
|
.section ".note.GNU-stack","",@progbits
|