| 1 |
Pacho Ramos <pacho@g.o> wrote: |
| 2 |
> |
| 3 |
> Is there any kind of information about performance penalty of |
| 4 |
> -fstack-protector? I have googled some time and there are various |
| 5 |
> estimations (from ~2 to ~8%), but I have no idea what have they checked |
| 6 |
> exactly. |
| 7 |
|
| 8 |
This depends extremely on the code: Most functions will be unchanged, |
| 9 |
and only a few functions with possibly dangerous code get a fixed |
| 10 |
impact whenever they are called. Whether they are called frequently |
| 11 |
or not cannot be said in general. |
| 12 |
|
| 13 |
> looks to me like it will have impact but "not too much" :/ |
| 14 |
|
| 15 |
Of course, it will have impact, but it is more theoretical than practical. |
| 16 |
|
| 17 |
> For some recent computers I would probably use it, |
| 18 |
> but for older ones, I am not sure if will be happy |
| 19 |
> with that additional impact :| |
| 20 |
|
| 21 |
On my old Pentium III laptop it makes no difference for my applications |
| 22 |
which can be felt. Of course, if you have a special time-critical |
| 23 |
application which just was on the limit, things might be different. |
| 24 |
However,in such cases you would also use things like -DNDEBUG which one |
| 25 |
would not choose as a default just because it gives some minimal |
| 26 |
performance enhancement. |
Archives