Profiling and C++

Profiling and C++

[Bohdan Vlasyuk]

Is it true that I can't do any profiling with my C++ programs?

I've tried it, and it doesn't seem to work for me.

That's piece of makefile:

CC = g++
CFLAGS = -g -pg
LFLAGS = $(CFLAGS)
LIBS =
COMPILE = $(CC) $(CFLAGS) -c

LINK = $(CC) $(LFLAGS) $(LIBS)

$(exe): $(o_files)
        $(LINK) -o $(exe) $(o_files)

%.cpp.o:
        $(COMPILE) $< -o $@


that's gprof output: 


Each sample counts as 0.01 seconds.
 no time accumulated

  %   cumulative   self              self     total           
 time   seconds   seconds    calls  Ts/call  Ts/call  name    
  0.00      0.00     0.00    18568     0.00     0.00  data_start

....

index % time    self  children    called     name
                0.00    0.00   18568/18568       <hicore> (2)
[1]      0.0    0.00    0.00   18568         data_start [1]
-----------------------------------------------


-- 
There is no opinion so absurd that some philosopher will not express it.
		-- Marcus Tullius Cicero, "Ad familiares"

Re: Profiling and C++

[John Levon]

On Wed, Jul 18, 2001 at 11:52:18AM +0300, Bohdan Vlasyuk wrote:

> Is it true that I can't do any profiling with my C++ programs?

no

> I've tried it, and it doesn't seem to work for me.
> 
> That's piece of makefile:
> 
> CC = g++
> CFLAGS = -g -pg
> LFLAGS = $(CFLAGS)
> LIBS =
> COMPILE = $(CC) $(CFLAGS) -c
> 
> LINK = $(CC) $(LFLAGS) $(LIBS)

probably need -g -pg for the final link stage too.

john

-- 
"Voodoo Programming:  Things programmers do that they know shouldn't work but
 they try anyway, and which sometimes actually work, such as recompiling
 everything."
	- Karl Lehenbauer

Re: Profiling and C++

[Bohdan Vlasyuk]

On Wed, Jul 18, 2001 at 04:48:46PM +0100, John Levon wrote:

>> I've tried it, and it doesn't seem to work for me.
>> That's piece of makefile:
>> CFLAGS = -g -pg
>> COMPILE = $(CC) $(CFLAGS) -c
>> LINK = $(CC) $(LFLAGS) $(LIBS)
> probably need -g -pg for the final link stage too.
It is included. That's not the problem.

I've figured out that I and too old binutils, so I updated it, and it
works now. Sorry. :(.

[I still have questions, but I suppose that's wrong list for it]


-- 
Where are the calculations that go with a calculated risk?

Re: Profiling and C++

[John Levon]

On Wed, Jul 18, 2001 at 07:02:41PM +0300, Bohdan Vlasyuk wrote:

> I've figured out that I and too old binutils, so I updated it, and it
> works now. Sorry. :(.
> 
> [I still have questions, but I suppose that's wrong list for it]

linuxperf@nl.linux.org ?

what are your problems ?

regards
john

-- 
"Voodoo Programming:  Things programmers do that they know shouldn't work but
 they try anyway, and which sometimes actually work, such as recompiling
 everything."
	- Karl Lehenbauer

Re: Profiling and C++

[Janis Johnson]

> Is it true that I can't do any profiling with my C++ programs?
>
>I've tried it, and it doesn't seem to work for me.

I did quick tests with 2.95.3, 3.0.x, and the mainline and it works
everywhere.  Notice, though that your program didn't accumulate
any time; that's why everything shows up as zero.

>Each sample counts as 0.01 seconds.
> no time accumulated

Janis