1 |
On Thursday 27 April 2006 17:53, Paolo Ripamonti wrote: |
2 |
> On 4/27/06, Sergio Polini <sp_rm_it@×××××.it> wrote: |
3 |
> > As far as I can understand, this means that AMD64 takes "a lot" of |
4 |
> > time to switch between two frequencies, so "the system freezes a |
5 |
> > little bit" when this happens. |
6 |
> |
7 |
> As far as I can say... this is not true for me. |
8 |
> I'm using "ondemand" on my AMD Athlon 3000 and I've never had such |
9 |
> problems. I'd rather take a look at compile options, use flag and kernel |
10 |
> configuration. Also, have you tried running memtest? |
11 |
|
12 |
stop using 'ondemand'. |
13 |
|
14 |
I have only 1/3 to 1/2 of the fps in ut2004 I get with 'performance'. |
15 |
Sometimes even lower. |
16 |
|
17 |
Switching frequencies IS time consuming. So, if you measured runtimes for any |
18 |
app with ondemand, you have very 'wrong' results. |
19 |
|
20 |
>From the help of the 'conservative' governor: |
21 |
|
22 |
If you have a desktop machine then you should really be considering │ |
23 |
│ the 'ondemand' governor instead, however if you are using a laptop, │ |
24 |
│ PDA or even an AMD64 based computer (due to the unacceptable │ |
25 |
│ step-by-step latency issues between the minimum and maximum frequency │ |
26 |
│ transitions in the CPU) you will probably want to use this governor. │ |
27 |
|
28 |
|
29 |
Again, everything you measured with ondemand, is not worth the time you spent |
30 |
on it. |
31 |
|
32 |
-- |
33 |
gentoo-amd64@g.o mailing list |