| 1 |
On 8/21/06, Stefan G. Weichinger <lists@×××××.at> wrote: |
| 2 |
> Acer TM 634 |
| 3 |
> P4-M 1.8GHz (cpu family : 15, model : 2) |
| 4 |
> 512 MB RAM |
| 5 |
> 30 GB 5200 rpm HDD |
| 6 |
|
| 7 |
It's all relative. I have a 2.1Ghz Core Duo with 2G of RAM and a |
| 8 |
160Gb HD, so *I* would consider your laptop, um, "underpowered". :-) |
| 9 |
|
| 10 |
But I also run with -Os. The fact is that some things will run |
| 11 |
slightly faster at -Os than -O2, and some things will be slightly |
| 12 |
slower. The same applies comparing -O3 to -O2, or -O3 to -Os; it all |
| 13 |
depends on what you are doing at the moment. |
| 14 |
|
| 15 |
So you should not assume that -O3 is faster for some random task just |
| 16 |
because it is "more optimized". It simply makes different trade-offs |
| 17 |
than -O2 or -Os, and because of the way CPUs and caches work these |
| 18 |
days, those trade-offs may help or hurt a particular segment of code. |
| 19 |
|
| 20 |
FYI, in all of the tests I did, the performance was within 10% of the |
| 21 |
median. The real deciding factor for me now is that -Os seems to take |
| 22 |
much less time and memory to compile than -O2 or -O3. And being a |
| 23 |
~arch user, time-to-compile is a nice thing to reduce. |
| 24 |
|
| 25 |
-Richard |
| 26 |
-- |
| 27 |
gentoo-user@g.o mailing list |
Archives