=Oprofile into debian

#install oprofile into debian
> apt-get install oprofile 

#do not get kernel profile. 
> opcontrol --no-vmlinux

>vi array.c

#include <stdio.h>
#include <stdlib.h>
#define ROWS 1000
#define COLS 1000
#define table_ref(table,row,col) ((table)[(row)*COLS+(col)])
typedef int* Table;
void touch_row_col(Table from, Table to) {
    int i, j;
    for(i=0;i<ROWS;++i) {
        for(j=0;j<COLS;++j) {
            table_ref(from, i, j) = table_ref(to, i, j);
        }
    }
}
void touch_col_row(Table from, Table to) {
    int i, j;
    for(i=0;i<COLS;++i) {
        for(j=0;j<ROWS;++j) {
            table_ref(from, j, i) = table_ref(to, j, i);
        }
    }
}
int main() {
    Table from, to;
    from = malloc(sizeof(int)*ROWS*COLS);
    to = malloc(sizeof(int)*ROWS*COLS);
    touch_row_col(from, to);
    touch_col_row(from, to);
    return 0;
}

#-g option is for arrange order result when you use opannotate
> gcc -g array.c


> opcontrol --reset
> opcontrol --start && ./a.out ; opcontrol --stop

#view of result

> opreport
CPU: AMD64 processors, speed 1795.54 MHz (estimated)
Counted CPU_CLK_UNHALTED events (Cycles outside of halt state) with a unit mask of 0x00 (No unit mask) count 100000
CPU_CLK_UNHALT...|
  samples|      %|
------------------
     1810 49.2249 no-vmlinux
     1344 36.5515 a.out
      213  5.7928 bash
      181  4.9225 libc-2.3.6.so
      107  2.9100 ld-2.3.6.so
       16  0.4351 oprofiled
        2  0.0544 grep
        2  0.0544 libpthread-2.3.6.so
        1  0.0272 mawk
        1  0.0272 tr


# specify the some program.
> opreport a.out /bin/grep
CPU: AMD64 processors, speed 1795.54 MHz (estimated)
Counted CPU_CLK_UNHALTED events (Cycles outside of halt state) with a unit mask of 0x00 (No unit mask) count 100000
CPU_CLK_UNHALT...|
  samples|      %|
------------------
     1344 99.8514 a.out
        2  0.1486 grep

#-l is for symbol information
> opreport -l a.out
CPU: AMD64 processors, speed 1795.54 MHz (estimated)
Counted CPU_CLK_UNHALTED events (Cycles outside of halt state) with a unit mask of 0x00 (No unit mask) count 100000
samples  %        symbol name
1157     86.0863  touch_col_row
187      13.9137  touch_row_col

#-c is for call graph
> opreport -c  a.out
CPU: AMD64 processors, speed 1795.54 MHz (estimated)
Counted CPU_CLK_UNHALTED events (Cycles outside of halt state) with a unit mask of 0x00 (No unit mask) count 100000
samples  %        symbol name
-------------------------------------------------------------------------------
1157     86.0863  touch_col_row
  1157     100.000  touch_col_row [self]
-------------------------------------------------------------------------------
187      13.9137  touch_row_col
  187      100.000  touch_row_col [self]
-------------------------------------------------------------------------------

#-s is for looking source code
> opannotate -s a.out
/*
 * Command line: opannotate -s a.out
 *
 * Interpretation of command line:
 * Output annotated source file with samples
 * Output all files
 *
 * CPU: AMD64 processors, speed 1795.54 MHz (estimated)
 * Counted CPU_CLK_UNHALTED events (Cycles outside of halt state) with a unit mask of 0x00 (No unit mask) count 100000
 */
/*
 * Total samples for file : "/root/misc/array.c"
 *
 *   1344 100.000
 */


               :#include <stdio.h>
               :#include <stdlib.h>
               :#define ROWS 1000
               :#define COLS 1000
               :#define table_ref(table,row,col) ((table)[(row)*COLS+(col)])
               :typedef int* Table;
               :void touch_row_col(Table from, Table to) { /* touch_row_col total:    187 13.9137 */
               :    int i, j;
               :    for(i=0;i<ROWS;++i) {
    51  3.7946 :        for(j=0;j<COLS;++j) {
   136 10.1190 :            table_ref(from, i, j) = table_ref(to, i, j);
               :        }
               :    }
               :}
               :void touch_col_row(Table from, Table to) { /* touch_col_row total:   1157 86.0863 */
               :    int i, j;
     3  0.2232 :    for(i=0;i<COLS;++i) {
   172 12.7976 :        for(j=0;j<ROWS;++j) {
   982 73.0655 :            table_ref(from, j, i) = table_ref(to, j, i);
               :        }
               :    }
               :}
               :int main() {
               :    Table from, to;
               :    from = malloc(sizeof(int)*ROWS*COLS);
               :    to = malloc(sizeof(int)*ROWS*COLS);
               :    touch_row_col(from, to);
               :    touch_col_row(from, to);
               :    return 0;
               :}
               :


#-a is for acemble code.
> opannotate -a a.out
/*
 * Command line: opannotate -a a.out
 *
 * Interpretation of command line:
 * Output annotated assembly listing with samples
 *
 * CPU: AMD64 processors, speed 1795.54 MHz (estimated)
 * Counted CPU_CLK_UNHALTED events (Cycles outside of halt state) with a unit mask of 0x00 (No unit mask) count 100000
 */
               :
               :/root/misc/a.out:     file format elf64-x86-64
               :
               :Disassembly of section .init:
               :Disassembly of section .plt:
               :Disassembly of section .text:
               :
00000000004004e3 <touch_col_row>: /* touch_col_row total:   1157 86.0863 */
               :  4004e3:       push   %rbp
               :  4004e4:       mov    %rsp,%rbp
               :  4004e7:       mov    %rdi,0xffffffffffffffe8(%rbp)
               :  4004eb:       mov    %rsi,0xffffffffffffffe0(%rbp)
               :  4004ef:       movl   $0x0,0xfffffffffffffff8(%rbp)
               :  4004f6:       jmp    400543 <touch_col_row+0x60>
               :  4004f8:       movl   $0x0,0xfffffffffffffffc(%rbp)
               :  4004ff:       jmp    400537 <touch_col_row+0x54>
    45  3.3482 :  400501:       mov    0xfffffffffffffffc(%rbp),%eax
               :  400504:       imul   $0x3e8,%eax,%eax
   370 27.5298 :  40050a:       add    0xfffffffffffffff8(%rbp),%eax
               :  40050d:       cltq
               :  40050f:       shl    $0x2,%rax
   415 30.8780 :  400513:       mov    %rax,%rdx
               :  400516:       add    0xffffffffffffffe8(%rbp),%rdx
               :  40051a:       mov    0xfffffffffffffffc(%rbp),%eax
    91  6.7708 :  40051d:       imul   $0x3e8,%eax,%eax
    33  2.4554 :  400523:       add    0xfffffffffffffff8(%rbp),%eax
               :  400526:       cltq
               :  400528:       shl    $0x2,%rax
    28  2.0833 :  40052c:       add    0xffffffffffffffe0(%rbp),%rax
               :  400530:       mov    (%rax),%eax
               :  400532:       mov    %eax,(%rdx)
   172 12.7976 :  400534:       incl   0xfffffffffffffffc(%rbp)
               :  400537:       cmpl   $0x3e7,0xfffffffffffffffc(%rbp)
               :  40053e:       jle    400501 <touch_col_row+0x1e>
               :  400540:       incl   0xfffffffffffffff8(%rbp)
     3  0.2232 :  400543:       cmpl   $0x3e7,0xfffffffffffffff8(%rbp)
               :  40054a:       jle    4004f8 <touch_col_row+0x15>
               :  40054c:       leaveq
               :  40054d:       retq
               :Disassembly of section .fini:
               :
               :/root/misc/a.out:     file format elf64-x86-64
               :
               :Disassembly of section .init:
               :Disassembly of section .plt:
               :Disassembly of section .text:
               :
0000000000400478 <touch_row_col>: /* touch_row_col total:    187 13.9137 */
               :  400478:       push   %rbp
               :  400479:       mov    %rsp,%rbp
               :  40047c:       mov    %rdi,0xffffffffffffffe8(%rbp)
               :  400480:       mov    %rsi,0xffffffffffffffe0(%rbp)
               :  400484:       movl   $0x0,0xfffffffffffffff8(%rbp)
               :  40048b:       jmp    4004d8 <touch_row_col+0x60>
               :  40048d:       movl   $0x0,0xfffffffffffffffc(%rbp)
               :  400494:       jmp    4004cc <touch_row_col+0x54>
    15  1.1161 :  400496:       mov    0xfffffffffffffff8(%rbp),%eax
               :  400499:       imul   $0x3e8,%eax,%eax
    28  2.0833 :  40049f:       add    0xfffffffffffffffc(%rbp),%eax
               :  4004a2:       cltq
               :  4004a4:       shl    $0x2,%rax
    21  1.5625 :  4004a8:       mov    %rax,%rdx
               :  4004ab:       add    0xffffffffffffffe8(%rbp),%rdx
               :  4004af:       mov    0xfffffffffffffff8(%rbp),%eax
    41  3.0506 :  4004b2:       imul   $0x3e8,%eax,%eax
    10  0.7440 :  4004b8:       add    0xfffffffffffffffc(%rbp),%eax
               :  4004bb:       cltq
               :  4004bd:       shl    $0x2,%rax
    18  1.3393 :  4004c1:       add    0xffffffffffffffe0(%rbp),%rax
     1  0.0744 :  4004c5:       mov    (%rax),%eax
     2  0.1488 :  4004c7:       mov    %eax,(%rdx)
    48  3.5714 :  4004c9:       incl   0xfffffffffffffffc(%rbp)
     3  0.2232 :  4004cc:       cmpl   $0x3e7,0xfffffffffffffffc(%rbp)
               :  4004d3:       jle    400496 <touch_row_col+0x1e>
               :  4004d5:       incl   0xfffffffffffffff8(%rbp)
               :  4004d8:       cmpl   $0x3e7,0xfffffffffffffff8(%rbp)
               :  4004df:       jle    40048d <touch_row_col+0x15>
               :  4004e1:       leaveq
               :  4004e2:       retq
               :Disassembly of section .fini:


#opcontrol --evnet=  is for event. Look data cache miss.
> opcontrol --shutdown
> opcontrol --reset
> opcontrol --event=DATA_CACHE_MISSES:500
> opcontrol --start && ./a.out ; opcontrol --stop

#you are able to see that cache miss happen in the touch_col_row.
> opreport -l ./a.out
CPU: AMD64 processors, speed 1795.54 MHz (estimated)
Counted DATA_CACHE_MISSES events (Data cache misses) with a unit mask of 0x00 (No unit mask) count 500
samples  %        symbol name
3329     92.8850  touch_col_row
255       7.1150  touch_row_col




#manual
http://oprofile.sourceforge.net/doc/index.html