GNU Info

(gmp.info)Profiling

---

Next: Autoconf Prev: Debugging Up: GMP Basics

---

Enter node , (file) or (file)node

Profiling
=========

   Running a program under a profiler is a good way to find where it's
spending most time and where improvements can be best sought.

   Depending on the system, it may be possible to get a flat profile,
meaning simple timer sampling of the program counter, with no special
GMP build options, just a `-p' when compiling the mainline.  This is a
good way to ensure minimum interference with normal operation.  The
necessary symbol type and size information exists in most of the GMP
assembler code.

   The `--enable-profiling' build option can be used to add suitable
compiler flags, either for `prof' (`-p') or `gprof' (`-pg'), see Note:
Build Options.  Which of the two is available and what they do will
depend on the system, and possibly on support available in `libc'.  For
some systems appropriate corresponding `mcount' calls are added to the
assembler code too.

   On x86 systems `prof' gives call counting, so that average time spent
in a function can be determined.  `gprof', where supported, adds call
graph construction, so for instance calls to `mpn_add_n' from `mpz_add'
and from `mpz_mul' can be differentiated.

   On x86 and 68k systems `-pg' and `-fomit-frame-pointer' are
incompatible, so the latter is not used when `gprof' profiling is
selected, which may result in poorer code generation.  If `prof'
profiling is selected instead it should still be possible to use
`gprof', but only the `gprof -p' flat profile and call counts can be
expected to be valid, not the `gprof -q' call graph.