Hi Everyone,
I'm trying to use _addcarryx_u64. GCC 7.1 is still generating ADD and
ADC instead of ADCX or ADOX:
$ objdump --disassemble test.o | c++filt | egrep 'adc|adcx|adox'
a4: 48 11 c8 adc %rcx,%rax
10f: 48 11 c8 adc %rcx,%rax
17a: 48 11 c8 adc %rcx,%rax
1e5: 48 11 c8 adc %rcx,%rax
24a: 48 11 c8 adc %rcx,%rax
29d: 48 11 c8 adc %rcx,%rax
2f0: 48 11 c8 adc %rcx,%rax
337: 48 11 c8 adc %rcx,%rax
Below is the test program, the objdump and the cpu features as
reported in /proc/cpuinfo. Unnecessary stuff was removed from objdump,
like Print and static initalizers.
The bug reporter is still hanging, so I cannot search for bug reports.
It would be nice if someone fixed the web server problem, too. Its
been a problem of 3 or 6 months now.
Any ideas what I am doing wrong with respect to ADX?
Thanks in advance
****************************************
$ g++ -Wall -march=native test.cxx -c
****************************************
skylake:cryptopp-fork$ objdump --disassemble test.o | c++filt
test.o: file format elf64-x86-64
Disassembly of section .text:
0000000000000000 <unsigned char Add<8u>(unsigned long*, unsigned long
const*, unsigned long const*)>:
0: 55 push %rbp
1: 48 89 e5 mov %rsp,%rbp
4: 48 81 ec c0 00 00 00 sub $0xc0,%rsp
b: 48 89 bd d8 fe ff ff mov %rdi,-0x128(%rbp)
12: 48 89 b5 d0 fe ff ff mov %rsi,-0x130(%rbp)
19: 48 89 95 c8 fe ff ff mov %rdx,-0x138(%rbp)
20: 48 8b 85 d0 fe ff ff mov -0x130(%rbp),%rax
27: 48 89 45 f8 mov %rax,-0x8(%rbp)
2b: 48 8b 85 c8 fe ff ff mov -0x138(%rbp),%rax
32: 48 89 45 f0 mov %rax,-0x10(%rbp)
36: 48 8b 85 d8 fe ff ff mov -0x128(%rbp),%rax
3d: 48 89 45 e8 mov %rax,-0x18(%rbp)
41: c6 45 e7 00 movb $0x0,-0x19(%rbp)
45: 48 8b 45 e8 mov -0x18(%rbp),%rax
49: 48 8d 70 38 lea 0x38(%rax),%rsi
4d: 48 8b 45 f0 mov -0x10(%rbp),%rax
51: 48 83 c0 38 add $0x38,%rax
55: 48 8b 00 mov (%rax),%rax
58: 48 8b 55 f8 mov -0x8(%rbp),%rdx
5c: 48 83 c2 38 add $0x38,%rdx
60: 48 8b 12 mov (%rdx),%rdx
63: 0f b6 4d e7 movzbl -0x19(%rbp),%ecx
67: 88 8d 07 ff ff ff mov %cl,-0xf9(%rbp)
6d: 48 89 95 f8 fe ff ff mov %rdx,-0x108(%rbp)
74: 48 89 85 f0 fe ff ff mov %rax,-0x110(%rbp)
7b: 48 89 b5 e8 fe ff ff mov %rsi,-0x118(%rbp)
82: 0f b6 85 07 ff ff ff movzbl -0xf9(%rbp),%eax
89: 89 c6 mov %eax,%esi
8b: 48 8b 85 f8 fe ff ff mov -0x108(%rbp),%rax
92: 48 8b 8d f0 fe ff ff mov -0x110(%rbp),%rcx
99: 48 8b 95 e8 fe ff ff mov -0x118(%rbp),%rdx
a0: 40 80 c6 ff add $0xff,%sil
a4: 48 11 c8 adc %rcx,%rax
a7: 0f 92 c1 setb %cl
aa: 48 89 02 mov %rax,(%rdx)
ad: 88 4d e7 mov %cl,-0x19(%rbp)
b0: 48 8b 45 e8 mov -0x18(%rbp),%rax
b4: 48 8d 70 30 lea 0x30(%rax),%rsi
b8: 48 8b 45 f0 mov -0x10(%rbp),%rax
bc: 48 83 c0 30 add $0x30,%rax
c0: 48 8b 00 mov (%rax),%rax
c3: 48 8b 55 f8 mov -0x8(%rbp),%rdx
c7: 48 83 c2 30 add $0x30,%rdx
cb: 48 8b 12 mov (%rdx),%rdx
ce: 0f b6 4d e7 movzbl -0x19(%rbp),%ecx
d2: 88 8d 27 ff ff ff mov %cl,-0xd9(%rbp)
d8: 48 89 95 18 ff ff ff mov %rdx,-0xe8(%rbp)
df: 48 89 85 10 ff ff ff mov %rax,-0xf0(%rbp)
e6: 48 89 b5 08 ff ff ff mov %rsi,-0xf8(%rbp)
ed: 0f b6 85 27 ff ff ff movzbl -0xd9(%rbp),%eax
f4: 89 c6 mov %eax,%esi
f6: 48 8b 85 18 ff ff ff mov -0xe8(%rbp),%rax
fd: 48 8b 8d 10 ff ff ff mov -0xf0(%rbp),%rcx
104: 48 8b 95 08 ff ff ff mov -0xf8(%rbp),%rdx
10b: 40 80 c6 ff add $0xff,%sil
10f: 48 11 c8 adc %rcx,%rax
112: 0f 92 c1 setb %cl
115: 48 89 02 mov %rax,(%rdx)
118: 88 4d e7 mov %cl,-0x19(%rbp)
11b: 48 8b 45 e8 mov -0x18(%rbp),%rax
11f: 48 8d 70 28 lea 0x28(%rax),%rsi
123: 48 8b 45 f0 mov -0x10(%rbp),%rax
127: 48 83 c0 28 add $0x28,%rax
12b: 48 8b 00 mov (%rax),%rax
12e: 48 8b 55 f8 mov -0x8(%rbp),%rdx
132: 48 83 c2 28 add $0x28,%rdx
136: 48 8b 12 mov (%rdx),%rdx
139: 0f b6 4d e7 movzbl -0x19(%rbp),%ecx
13d: 88 8d 47 ff ff ff mov %cl,-0xb9(%rbp)
143: 48 89 95 38 ff ff ff mov %rdx,-0xc8(%rbp)
14a: 48 89 85 30 ff ff ff mov %rax,-0xd0(%rbp)
151: 48 89 b5 28 ff ff ff mov %rsi,-0xd8(%rbp)
158: 0f b6 85 47 ff ff ff movzbl -0xb9(%rbp),%eax
15f: 89 c6 mov %eax,%esi
161: 48 8b 85 38 ff ff ff mov -0xc8(%rbp),%rax
168: 48 8b 8d 30 ff ff ff mov -0xd0(%rbp),%rcx
16f: 48 8b 95 28 ff ff ff mov -0xd8(%rbp),%rdx
176: 40 80 c6 ff add $0xff,%sil
17a: 48 11 c8 adc %rcx,%rax
17d: 0f 92 c1 setb %cl
180: 48 89 02 mov %rax,(%rdx)
183: 88 4d e7 mov %cl,-0x19(%rbp)
186: 48 8b 45 e8 mov -0x18(%rbp),%rax
18a: 48 8d 70 20 lea 0x20(%rax),%rsi
18e: 48 8b 45 f0 mov -0x10(%rbp),%rax
192: 48 83 c0 20 add $0x20,%rax
196: 48 8b 00 mov (%rax),%rax
199: 48 8b 55 f8 mov -0x8(%rbp),%rdx
19d: 48 83 c2 20 add $0x20,%rdx
1a1: 48 8b 12 mov (%rdx),%rdx
1a4: 0f b6 4d e7 movzbl -0x19(%rbp),%ecx
1a8: 88 8d 67 ff ff ff mov %cl,-0x99(%rbp)
1ae: 48 89 95 58 ff ff ff mov %rdx,-0xa8(%rbp)
1b5: 48 89 85 50 ff ff ff mov %rax,-0xb0(%rbp)
1bc: 48 89 b5 48 ff ff ff mov %rsi,-0xb8(%rbp)
1c3: 0f b6 85 67 ff ff ff movzbl -0x99(%rbp),%eax
1ca: 89 c6 mov %eax,%esi
1cc: 48 8b 85 58 ff ff ff mov -0xa8(%rbp),%rax
1d3: 48 8b 8d 50 ff ff ff mov -0xb0(%rbp),%rcx
1da: 48 8b 95 48 ff ff ff mov -0xb8(%rbp),%rdx
1e1: 40 80 c6 ff add $0xff,%sil
1e5: 48 11 c8 adc %rcx,%rax
1e8: 0f 92 c1 setb %cl
1eb: 48 89 02 mov %rax,(%rdx)
1ee: 88 4d e7 mov %cl,-0x19(%rbp)
1f1: 48 8b 45 e8 mov -0x18(%rbp),%rax
1f5: 48 8d 70 18 lea 0x18(%rax),%rsi
1f9: 48 8b 45 f0 mov -0x10(%rbp),%rax
1fd: 48 83 c0 18 add $0x18,%rax
201: 48 8b 00 mov (%rax),%rax
204: 48 8b 55 f8 mov -0x8(%rbp),%rdx
208: 48 83 c2 18 add $0x18,%rdx
20c: 48 8b 12 mov (%rdx),%rdx
20f: 0f b6 4d e7 movzbl -0x19(%rbp),%ecx
213: 88 4d 87 mov %cl,-0x79(%rbp)
216: 48 89 95 78 ff ff ff mov %rdx,-0x88(%rbp)
21d: 48 89 85 70 ff ff ff mov %rax,-0x90(%rbp)
224: 48 89 b5 68 ff ff ff mov %rsi,-0x98(%rbp)
22b: 0f b6 45 87 movzbl -0x79(%rbp),%eax
22f: 89 c6 mov %eax,%esi
231: 48 8b 85 78 ff ff ff mov -0x88(%rbp),%rax
238: 48 8b 8d 70 ff ff ff mov -0x90(%rbp),%rcx
23f: 48 8b 95 68 ff ff ff mov -0x98(%rbp),%rdx
246: 40 80 c6 ff add $0xff,%sil
24a: 48 11 c8 adc %rcx,%rax
24d: 0f 92 c1 setb %cl
250: 48 89 02 mov %rax,(%rdx)
253: 88 4d e7 mov %cl,-0x19(%rbp)
256: 48 8b 45 e8 mov -0x18(%rbp),%rax
25a: 48 8d 70 10 lea 0x10(%rax),%rsi
25e: 48 8b 45 f0 mov -0x10(%rbp),%rax
262: 48 83 c0 10 add $0x10,%rax
266: 48 8b 00 mov (%rax),%rax
269: 48 8b 55 f8 mov -0x8(%rbp),%rdx
26d: 48 83 c2 10 add $0x10,%rdx
271: 48 8b 12 mov (%rdx),%rdx
274: 0f b6 4d e7 movzbl -0x19(%rbp),%ecx
278: 88 4d a7 mov %cl,-0x59(%rbp)
27b: 48 89 55 98 mov %rdx,-0x68(%rbp)
27f: 48 89 45 90 mov %rax,-0x70(%rbp)
283: 48 89 75 88 mov %rsi,-0x78(%rbp)
287: 0f b6 45 a7 movzbl -0x59(%rbp),%eax
28b: 89 c6 mov %eax,%esi
28d: 48 8b 45 98 mov -0x68(%rbp),%rax
291: 48 8b 4d 90 mov -0x70(%rbp),%rcx
295: 48 8b 55 88 mov -0x78(%rbp),%rdx
299: 40 80 c6 ff add $0xff,%sil
29d: 48 11 c8 adc %rcx,%rax
2a0: 0f 92 c1 setb %cl
2a3: 48 89 02 mov %rax,(%rdx)
2a6: 88 4d e7 mov %cl,-0x19(%rbp)
2a9: 48 8b 45 e8 mov -0x18(%rbp),%rax
2ad: 48 8d 70 08 lea 0x8(%rax),%rsi
2b1: 48 8b 45 f0 mov -0x10(%rbp),%rax
2b5: 48 83 c0 08 add $0x8,%rax
2b9: 48 8b 00 mov (%rax),%rax
2bc: 48 8b 55 f8 mov -0x8(%rbp),%rdx
2c0: 48 83 c2 08 add $0x8,%rdx
2c4: 48 8b 12 mov (%rdx),%rdx
2c7: 0f b6 4d e7 movzbl -0x19(%rbp),%ecx
2cb: 88 4d c7 mov %cl,-0x39(%rbp)
2ce: 48 89 55 b8 mov %rdx,-0x48(%rbp)
2d2: 48 89 45 b0 mov %rax,-0x50(%rbp)
2d6: 48 89 75 a8 mov %rsi,-0x58(%rbp)
2da: 0f b6 45 c7 movzbl -0x39(%rbp),%eax
2de: 89 c6 mov %eax,%esi
2e0: 48 8b 45 b8 mov -0x48(%rbp),%rax
2e4: 48 8b 4d b0 mov -0x50(%rbp),%rcx
2e8: 48 8b 55 a8 mov -0x58(%rbp),%rdx
2ec: 40 80 c6 ff add $0xff,%sil
2f0: 48 11 c8 adc %rcx,%rax
2f3: 0f 92 c1 setb %cl
2f6: 48 89 02 mov %rax,(%rdx)
2f9: 88 4d e7 mov %cl,-0x19(%rbp)
2fc: 48 8b 45 f0 mov -0x10(%rbp),%rax
300: 48 8b 00 mov (%rax),%rax
303: 48 8b 55 f8 mov -0x8(%rbp),%rdx
307: 48 8b 12 mov (%rdx),%rdx
30a: 0f b6 4d e7 movzbl -0x19(%rbp),%ecx
30e: 88 4d e6 mov %cl,-0x1a(%rbp)
311: 48 89 55 d8 mov %rdx,-0x28(%rbp)
315: 48 89 45 d0 mov %rax,-0x30(%rbp)
319: 48 8b 45 e8 mov -0x18(%rbp),%rax
31d: 48 89 45 c8 mov %rax,-0x38(%rbp)
321: 0f b6 45 e6 movzbl -0x1a(%rbp),%eax
325: 89 c6 mov %eax,%esi
327: 48 8b 45 d8 mov -0x28(%rbp),%rax
32b: 48 8b 4d d0 mov -0x30(%rbp),%rcx
32f: 48 8b 55 c8 mov -0x38(%rbp),%rdx
333: 40 80 c6 ff add $0xff,%sil
337: 48 11 c8 adc %rcx,%rax
33a: 0f 92 c1 setb %cl
33d: 48 89 02 mov %rax,(%rdx)
340: 88 4d e7 mov %cl,-0x19(%rbp)
343: 0f b6 45 e7 movzbl -0x19(%rbp),%eax
347: c9 leaveq
348: c3 retq
0000000000000349 <main>:
349: 55 push %rbp
34a: 48 89 e5 mov %rsp,%rbp
34d: 53 push %rbx
34e: 48 81 ec f8 02 00 00 sub $0x2f8,%rsp
355: 89 bd 0c fd ff ff mov %edi,-0x2f4(%rbp)
35b: 48 89 b5 00 fd ff ff mov %rsi,-0x300(%rbp)
362: 48 8d 85 10 fd ff ff lea -0x2f0(%rbp),%rax
369: 48 89 c7 mov %rax,%rdi
36c: e8 00 00 00 00 callq 371 <main+0x28>
371: 48 8d 85 10 fd ff ff lea -0x2f0(%rbp),%rax
378: ba 08 00 00 00 mov $0x8,%edx
37d: be 00 00 00 00 mov $0x0,%esi
382: 48 89 c7 mov %rax,%rdi
385: e8 00 00 00 00 callq 38a <main+0x41>
38a: 48 8d 8d 60 ff ff ff lea -0xa0(%rbp),%rcx
391: 48 8d 85 10 fd ff ff lea -0x2f0(%rbp),%rax
398: ba 40 00 00 00 mov $0x40,%edx
39d: 48 89 ce mov %rcx,%rsi
3a0: 48 89 c7 mov %rax,%rdi
3a3: e8 00 00 00 00 callq 3a8 <main+0x5f>
3a8: 48 8d 8d 20 ff ff ff lea -0xe0(%rbp),%rcx
3af: 48 8d 85 10 fd ff ff lea -0x2f0(%rbp),%rax
3b6: ba 40 00 00 00 mov $0x40,%edx
3bb: 48 89 ce mov %rcx,%rsi
3be: 48 89 c7 mov %rax,%rdi
3c1: e8 00 00 00 00 callq 3c6 <main+0x7d>
3c6: 48 8d 95 20 ff ff ff lea -0xe0(%rbp),%rdx
3cd: 48 8d 8d 60 ff ff ff lea -0xa0(%rbp),%rcx
3d4: 48 8d 45 a0 lea -0x60(%rbp),%rax
3d8: 48 89 ce mov %rcx,%rsi
3db: 48 89 c7 mov %rax,%rdi
3de: e8 00 00 00 00 callq 3e3 <main+0x9a>
3e3: 88 45 ef mov %al,-0x11(%rbp)
3e6: be 00 00 00 00 mov $0x0,%esi
3eb: bf 00 00 00 00 mov $0x0,%edi
3f0: e8 00 00 00 00 callq 3f5 <main+0xac>
3f5: 48 8d 85 60 ff ff ff lea -0xa0(%rbp),%rax
3fc: 48 89 c7 mov %rax,%rdi
3ff: e8 00 00 00 00 callq 404 <main+0xbb>
404: be 00 00 00 00 mov $0x0,%esi
409: bf 00 00 00 00 mov $0x0,%edi
40e: e8 00 00 00 00 callq 413 <main+0xca>
413: 48 8d 85 20 ff ff ff lea -0xe0(%rbp),%rax
41a: 48 89 c7 mov %rax,%rdi
41d: e8 00 00 00 00 callq 422 <main+0xd9>
422: be 00 00 00 00 mov $0x0,%esi
427: bf 00 00 00 00 mov $0x0,%edi
42c: e8 00 00 00 00 callq 431 <main+0xe8>
431: 80 7d ef 00 cmpb $0x0,-0x11(%rbp)
435: 74 0a je 441 <main+0xf8>
437: bf 31 00 00 00 mov $0x31,%edi
43c: e8 00 00 00 00 callq 441 <main+0xf8>
441: 48 8d 45 a0 lea -0x60(%rbp),%rax
445: 48 89 c7 mov %rax,%rdi
448: e8 00 00 00 00 callq 44d <main+0x104>
44d: bb 00 00 00 00 mov $0x0,%ebx
452: 48 8d 85 10 fd ff ff lea -0x2f0(%rbp),%rax
459: 48 89 c7 mov %rax,%rdi
45c: e8 00 00 00 00 callq 461 <main+0x118>
461: 89 d8 mov %ebx,%eax
463: eb 1d jmp 482 <main+0x139>
465: 48 89 c3 mov %rax,%rbx
468: 48 8d 85 10 fd ff ff lea -0x2f0(%rbp),%rax
46f: 48 89 c7 mov %rax,%rdi
472: e8 00 00 00 00 callq 477 <main+0x12e>
477: 48 89 d8 mov %rbx,%rax
47a: 48 89 c7 mov %rax,%rdi
47d: e8 00 00 00 00 callq 482 <main+0x139>
482: 48 81 c4 f8 02 00 00 add $0x2f8,%rsp
489: 5b pop %rbx
48a: 5d pop %rbp
48b: c3 retq
****************************************
$ cat -n test.cxx
1 #include <iostream>
2 #include <string>
3 #include <fstream>
4 #include <cstdio>
5 #include <cstdint>
6 #include <cstring>
7
8 #include <immintrin.h>
9 #if defined(__GNUC__)
10 # include <adxintrin.h>
11 #endif
12
13 #if !defined(__ADX__)
14 # error ADX is not available
15 #endif
16
17 template<unsigned int N>
18 void Print(uint64_t vals[])
19 {
20 for (size_t i=0; i<N; ++i)
21 {
22 printf("%08X", static_cast<uint32_t>(vals[i] >> 32));
23 printf("%08X", static_cast<uint32_t>(vals[i]));
24 }
25 printf("\n");
26 }
27
28 template<unsigned int N>
29 uint8_t Add(uint64_t C[], const uint64_t A[], const uint64_t B[])
30 {
31 const ssize_t NN = static_cast<ssize_t>(N);
32 const long long unsigned int* AA = reinterpret_cast<const
long long unsigned int*>(A);
33 const long long unsigned int* BB =
reinterpret_cast<const long long unsigned int*>(B);
34 long long unsigned int* CC =
reinterpret_cast<long long unsigned int*>(C);
35
36 uint8_t carry = 0;
37 for (ssize_t i=NN-1; i>=0; --i)
38 carry = _addcarryx_u64(carry, AA[i], BB[i], &CC[i]);
39
40 return carry;
41 }
42
43 template<>
44 uint8_t Add<8>(uint64_t C[], const uint64_t A[], const uint64_t B[])
45 {
46 const long long unsigned int* AA = reinterpret_cast<const
long long unsigned int*>(A);
47 const long long unsigned int* BB =
reinterpret_cast<const long long unsigned int*>(B);
48 long long unsigned int* CC =
reinterpret_cast<long long unsigned int*>(C);
49
50 uint8_t carry = 0;
51
52 carry = _addcarryx_u64(carry, AA[7], BB[7], &CC[7]);
53 carry = _addcarryx_u64(carry, AA[6], BB[6], &CC[6]);
54 carry = _addcarryx_u64(carry, AA[5], BB[5], &CC[5]);
55 carry = _addcarryx_u64(carry, AA[4], BB[4], &CC[4]);
56 carry = _addcarryx_u64(carry, AA[3], BB[3], &CC[3]);
57 carry = _addcarryx_u64(carry, AA[2], BB[2], &CC[2]);
58 carry = _addcarryx_u64(carry, AA[1], BB[1], &CC[1]);
59 carry = _addcarryx_u64(carry, AA[0], BB[0], &CC[0]);
60
61 return carry;
62 }
63
64 int main(int argc, char* argv[])
65 {
66 enum {COUNT=8};
67 uint64_t C[COUNT], A[COUNT], B[COUNT];
68
69 std::ifstream in;
70 in.open("/dev/urandom");
71 in.read(reinterpret_cast<char*>(A), sizeof(A));
72 in.read(reinterpret_cast<char*>(B), sizeof(B));
73
74 uint8_t carry = Add<COUNT>(C, A, B);
75
76 std::cout << "A:\n ";
77 Print<COUNT>(A);
78
79 std::cout << "B:\n ";
80 Print<COUNT>(B);
81
82 std::cout << "C:\n";
83 if (carry) printf("1");
84 Print<COUNT>(C);
85
86 return 0;
87 }
****************************************
$ cat /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 94
model name : Intel(R) Core(TM) i5-6400 CPU @ 2.70GHz
stepping : 3
microcode : 0xba
cpu MHz : 799.914
cache size : 6144 KB
physical id : 0
siblings : 4
core id : 0
cpu cores : 4
apicid : 0
initial apicid : 0
fpu : yes
fpu_exception : yes
cpuid level : 22
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge
mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe
syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts
rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq
pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 sdbg fma cx16
xtpr pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer
aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault
intel_pt tpr_shadow vnmi flexpriority ept vpid fsgsbase tsc_adjust
bmi1 avx2 smep bmi2 erms invpcid mpx rdseed adx smap clflushopt
xsaveopt xsavec xgetbv1 xsaves dtherm ida arat pln pts hwp hwp_notify
hwp_act_window hwp_epp
bugs :
bogomips : 5424.00
clflush size : 64
cache_alignment : 64
address sizes : 39 bits physical, 48 bits virtual
