Confused with memory/time measurements in C++/Linux

[std10093]

Ok, so I have an application, where I first build a tree and then I search it. The tree has 130.000 elements and I perform 1400 queries.

How should I measure time/memory? The second one is actually what I do not know. I am using the terminal in Linux.

Take a look at this

samaras@samaras-A15:~/code/dD_spatial_drarch_NN$ /usr/bin/time -f "\n%E elapsed,\n%U user,\n%S system,\n%M memory\n%x status" ./nn
Search for 1 NN of 1400 queries in the tree,with approximation 0.0 took 1.456464554seconds wall clock time.

34:01.32 elapsed,
2031.99 user,
0.44 system,
694304 memory
0 status

The Search for...clock time comes from my code (where I use a method named Nomimal’s animal’s (which once was a member of this forum :/) approach from here).
However, I measure only the queries in my code.
What is the time command actually says me? Googling confused me very much!

Every element of the tree contains 17 doubles.

Is the memory that's there (694304 memory) telling me how much memory does the application takes? I remember that valgrind can help, but I do not remember/know how!

I also thought of sizeof(), but I think that's not good, since

myDS bla;
bla.insert5Elements;
sizeof(bla);
myDs foo;
foo.insert654735463Elements;
sizeof(foo);

will yield the same results.

EDIT: I also found this # valgrind --tool=massif program

So what is the output that time gives me? Is it enough?

[tabstop]

You should type "man time" into your system to find out how your specific version of time operates (it's probably the same as time(1) - Linux man page, but you never know).

Valgrind can check your memory accesses and your allocations/deallocations, so you can see how much memory you were using over time. Only you know what that number should be (ie the log will show that you created 130000 elements in your tree, but only you (and your algorithm) know whether they needed to all be there at the same time or not).

EDIT: You may want to look at some of the other tools of Valgrind as well, such as DHAT.

[std10093]

Thanks for the info.