compiling with "-ggdb", running in gdb, hitting ctrl + c, type: disas /m

gistfile1.txt

39	int main() {
   0x0000000000406660 <+0>:	push   %r12
   0x000000000040666c <+12>:	push   %rbp
   0x0000000000406675 <+21>:	push   %rbx
   0x000000000040667b <+27>:	sub    $0x20,%rsp

40	  posix_memalign((void**)&particles, 256, sizeof(particle_group) * GROUP_COUNT);
   0x0000000000406662 <+2>:	mov    $0x249f00,%edx
   0x0000000000406667 <+7>:	mov    $0x100,%esi
   0x000000000040666d <+13>:	mov    $0x6238e0,%edi
   0x0000000000406676 <+22>:	mov    $0xa,%ebx
   0x000000000040667f <+31>:	callq  0x405de0 <posix_memalign@plt>
   0x0000000000406684 <+36>:	vmovaps 0x15ea4(%rip),%xmm3        # 0x41c530
   0x000000000040668c <+44>:	vmovaps 0x15eac(%rip),%xmm2        # 0x41c540

41
42	  uint64_t total_time = 0;
   0x0000000000406672 <+18>:	xor    %r12d,%r12d
   0x00000000004066e3 <+131>:	mov    $0x61a8,%ecx
   0x00000000004066e8 <+136>:	nopl   0x0(%rax,%rax,1)

43
44	  for(int k = 0; k < NTESTS; ++k) {
   0x000000000040679f <+319>:	sub    $0x1,%ebx
   0x00000000004067a2 <+322>:	vmovaps (%rsp),%xmm3
   0x00000000004067a7 <+327>:	vmovaps 0x10(%rsp),%xmm2
   0x00000000004067ad <+333>:	jne    0x406694 <main()+52>

45
46	    memset((char*)particles, 0x00, sizeof(particle_group) * GROUP_COUNT);
   0x0000000000406694 <+52>:	mov    0x21d245(%rip),%rdi        # 0x6238e0 <particles>
   0x000000000040669b <+59>:	mov    $0x249f00,%edx
   0x00000000004066a0 <+64>:	xor    %esi,%esi
   0x00000000004066a2 <+66>:	vmovaps %xmm2,0x10(%rsp)
   0x00000000004066a8 <+72>:	vmovaps %xmm3,(%rsp)
   0x00000000004066ad <+77>:	callq  0x405da0 <memset@plt>

47
48	    uint64_t start = uv_hrtime();
   0x00000000004066b2 <+82>:	callq  0x406abd <uv_hrtime>
   0x00000000004066b7 <+87>:	mov    0x21d222(%rip),%rdx        # 0x6238e0 <particles>
   0x00000000004066be <+94>:	vmovaps (%rsp),%xmm3
   0x00000000004066c3 <+99>:	mov    %rax,%rbp
   0x00000000004066c6 <+102>:	mov    $0xbb8,%eax
   0x00000000004066cb <+107>:	vmovaps 0x10(%rsp),%xmm2
   0x00000000004066d1 <+113>:	lea    0x60(%rdx),%rdi
   0x00000000004066d5 <+117>:	lea    0x20(%rdx),%rsi
   0x00000000004066d9 <+121>:	nopl   0x0(%rax)
   0x00000000004066e0 <+128>:	mov    %rsi,%rdx

49	    for(int i = 0; i < NLOOPS; ++i) {
   0x000000000040673e <+222>:	sub    $0x1,%eax
   0x0000000000406741 <+225>:	jne    0x4066e0 <main()+128>
   0x0000000000406743 <+227>:	vmovaps %xmm2,0x10(%rsp)
   0x0000000000406749 <+233>:	vmovaps %xmm3,(%rsp)

50	      add_force(0.0005, 0.03);
51	       step();
52	    }
53	    uint64_t d = uv_hrtime() - start;
   0x000000000040674e <+238>:	callq  0x406abd <uv_hrtime>

54	    printf("Took: %lld, millis: %f, millis per loop: %f\n", d, double(d)/1000000.0, (double(d)/1000000.0)/float(NLOOPS));
   0x0000000000406753 <+243>:	vmovaps (%rsp),%xmm3
   0x0000000000406758 <+248>:	sub    %rbp,%rax
   0x000000000040675b <+251>:	vmovaps 0x10(%rsp),%xmm2
   0x0000000000406761 <+257>:	mov    %rax,%rbp
   0x0000000000406764 <+260>:	js     0x4067ef <main()+399>
   0x000000000040676a <+266>:	vcvtsi2sd %rax,%xmm0,%xmm0
   0x000000000040676f <+271>:	mov    %rbp,%rsi
     0x0000000000406772 <+274>:	mov    $0x41c4e8,%edi
   0x0000000000406777 <+279>:	mov    $0x2,%eax
   0x000000000040677c <+284>:	vmovaps %xmm3,(%rsp)
   0x0000000000406781 <+289>:	vdivsd 0x15dc7(%rip),%xmm0,%xmm0        # 0x41c550
   0x0000000000406792 <+306>:	vdivsd 0x15dbe(%rip),%xmm0,%xmm1        # 0x41c558
   0x000000000040679a <+314>:	callq  0x405d80 <printf@plt>
   0x00000000004067ef <+399>:	mov    %rbp,%rdx
   0x00000000004067f2 <+402>:	shr    %rax
   0x00000000004067f5 <+405>:	and    $0x1,%edx
   0x00000000004067f8 <+408>:	or     %rdx,%rax
   0x00000000004067fb <+411>:	vcvtsi2sd %rax,%xmm0,%xmm0
   0x0000000000406800 <+416>:	vaddsd %xmm0,%xmm0,%xmm0
   0x0000000000406804 <+420>:	jmpq   0x40676f <main()+271>

55
56	    total_time += d;
   0x0000000000406789 <+297>:	add    %rbp,%r12
   0x000000000040678c <+300>:	vmovaps %xmm2,0x10(%rsp)

57	  }
58
59	  for(int i = 0; i < 10; ++i) {
60	    int dx = i * 2;
61	    //     printf("force_x: %f, force_y: %f, pos_x: %f, pos_y: %f\n", forces[dx + 0], forces[dx + 1], positions[dx + 0], positions[dx + 1]);
62	  }
63
64	  double avg = (double(total_time)/1000000.0)/double(NTESTS);
   0x00000000004067b3 <+339>:	test   %r12,%r12
   0x00000000004067b6 <+342>:	js     0x406809 <main()+425>
   0x00000000004067b8 <+344>:	vcvtsi2sd %r12,%xmm0,%xmm0
   0x00000000004067c7 <+359>:	vdivsd 0x15d81(%rip),%xmm0,%xmm0        # 0x41c550
   0x00000000004067cf <+367>:	vdivsd 0x15d89(%rip),%xmm0,%xmm0        # 0x41c560
   0x0000000000406809 <+425>:	mov    %r12,%rax
   0x000000000040680c <+428>:	and    $0x1,%r12d
   0x0000000000406810 <+432>:	shr    %rax
   0x0000000000406813 <+435>:	or     %r12,%rax
   0x0000000000406816 <+438>:	vcvtsi2sd %rax,%xmm0,%xmm0
   0x000000000040681b <+443>:	vaddsd %xmm0,%xmm0,%xmm0
   0x000000000040681f <+447>:	jmp    0x4067bd <main()+349>
   0x0000000000406821:	data32 data32 data32 data32 data32 nopw %cs:0x0(%rax,%rax,1)

65	  printf("Avarage: %f, ms: %f\n", avg, avg/float(NLOOPS));
   0x00000000004067bd <+349>:	mov    $0x41c518,%edi
   0x00000000004067c2 <+354>:	mov    $0x2,%eax
   0x00000000004067d7 <+375>:	vdivsd 0x15d79(%rip),%xmm0,%xmm1        # 0x41c558
   0x00000000004067df <+383>:	callq  0x405d80 <printf@plt>

66	}
   0x00000000004067e4 <+388>:	add    $0x20,%rsp
   0x00000000004067e8 <+392>:	xor    %eax,%eax
   0x00000000004067ea <+394>:	pop    %rbx
   0x00000000004067eb <+395>:	pop    %rbp
   0x00000000004067ec <+396>:	pop    %r12
   0x00000000004067ee <+398>:	retq

67
68	// ----------------------------------------------------------------------
69	/*
70	  prefecth heuristics:
71	  _mm_prefetch is a hint to fill the cache
72
73	*/
74
75	void add_force(float x, float y) {
76	  __m128 force_x = _mm_set_ps(x, x, x, x);
77	  __m128 force_y = _mm_set_ps(y, y, y, y);
78
79	  int const prefetch_count = power_of_two_below(CACHE_LINE_SIZE / sizeof(particles[0]));
80	  __m128 tmp_x;
81	  __m128 tmp_y;
82	  for(int i = 0; i < GROUP_COUNT; ++i){
   0x0000000000406707 <+167>:	sub    $0x1,%ecx
   0x000000000040670a <+170>:	jne    0x4066f0 <main()+144>
   0x000000000040670c <+172>:	mov    %rdi,%rcx
   0x000000000040670f <+175>:	xor    %edx,%edx
   0x0000000000406711 <+177>:	jmp    0x406727 <main()+199>
   0x0000000000406713 <+179>:	nopl   0x0(%rax,%rax,1)

83
84	    if(i % 256 == 0) {
85	      // _mm_prefetch(particles + i + 1, _MM_HINT_T0);
86	      // _mm_prefetch(particles + i + prefetch_count, _MM_HINT_T0);
87	    }
88	    tmp_x = particles[i].force_x;
89	    tmp_y = particles[i].force_y;
90
91	    particles[i].force_x = _mm_add_ps(tmp_x, force_x);
   0x00000000004066f8 <+152>:	vmovaps %xmm0,-0x60(%rdx)

92	    particles[i].force_y = _mm_add_ps(tmp_y, force_y);
   0x0000000000406702 <+162>:	vmovaps %xmm0,-0x50(%rdx)

93
94	  }
95	}
96
97	void step() {
98	  __m128 const null_force = _mm_set_ps(0.0f, 0.0f, 0.0f, 0.0f);
99	  __m128 const drag = _mm_set_ps(0.99f, 0.99f, 0.99f, 0.99f);
100	  __m128 pos_x;
101	  __m128 pos_y;
102	  __m128 vel_x;
103	  __m128 vel_y;
104	  int const prefetch_count = power_of_two_below(CACHE_LINE_SIZE / sizeof(particles[0]));
105	  uint32_t i = 0;
106	  uint32_t total = GROUP_COUNT;
107	  for (i = 0; i < total; ++i){
   0x0000000000406718 <+184>:	add    $0x1,%edx
   0x000000000040671b <+187>:	add    $0x60,%rcx
   0x000000000040671f <+191>:	cmp    $0x61a8,%edx
   0x0000000000406725 <+197>:	je     0x40673e <main()+222>
   0x000000000040672c <+204>:	add    $0x1,%edx
   0x0000000000406736 <+214>:	cmp    $0x61a8,%edx
   0x000000000040673c <+220>:	jne    0x406727 <main()+199>

108
109	      if( (i % 2) == 0) {
   0x0000000000406727 <+199>:	test   $0x1,%dl
   0x000000000040672a <+202>:	jne    0x406718 <main()+184>