gprof reporting all 0.0 seconds

[Yang Zhang]

michael wrote:
> Yang Zhang wrote:
>> michael wrote:
>>> Yang Zhang wrote:
>>>> Retrying.
>>>>
>>>> Yang Zhang wrote:
>>>>> The output of gprof for my C++ program has no entries that contain 
>>>>> non-0 time:
>>>>>
>>>>>   %   cumulative   self              self     total
>>>>>  time   seconds   seconds    calls  Ts/call  Ts/call  name
>>>>>   0.00      0.00     0.00 20347515     0.00     0.00  
>>>>> __gnu_cxx::__nor..
>>>>>   0.00      0.00     0.00 18191192     0.00     0.00  
>>>>> __gnu_cxx::__nor..
>>>>>   0.00      0.00     0.00 15330884     0.00     0.00  
>>>>> std::strchr(char..
>>>>>   0.00      0.00     0.00 11405520     0.00     0.00  
>>>>> __gnu_cxx::__nor..
>>>>>   0.00      0.00     0.00 10668361     0.00     0.00  
>>>>> __gnu_cxx::__stl..
>>>>>   0.00      0.00     0.00 10173756     0.00     0.00  
>>>>> __gnu_cxx::__nor..
>>>>> ...
>>>>>
>>>>> What's going wrong? The source (it's a simple 1-file program) and 
>>>>> Makefile are available at:
>>>>>
>>>>> http://assorted.svn.sourceforge.net/viewvc/assorted/hash-join/trunk/src/ 
>>>>>
>>>>>
>>>>> In particular, the build command is:
>>>>>
>>>>> g++ -Wall -lprofiler -lpthread -o hashjoin-pg hashjoin.cc -g -pg
>>>>>
>>>>> Thanks in advance for any help.
>>>>
>>> Compile it without any optimization and retry.
>>>
>>> Michael
>>>
>>
>> I tried just now. That didn't change anything, sadly.
>>
> gprof if I remember is broken for multithread. I have some trouble to 
> compile your test.
> 
> Regards Michael
> 

Whoops, I completely disregarded your first sentence. Yeah, you're right 
- turns out gprof can't do threads. Pity!

I seem to have trouble finding any mention of this detail, except for:

http://sam.zoy.org/writings/programming/gprof.html

Yang