1 |
On Samstag 05 September 2009, Mark Knecht wrote: |
2 |
> On Fri, Sep 4, 2009 at 4:43 PM, Volker Armin |
3 |
> |
4 |
> Hemmann<volkerarmin@××××××××××.com> wrote: |
5 |
> > On Samstag 05 September 2009, Nikos Chantziaras wrote: |
6 |
> >> I recently stumbled upon an LWN article that mentioned Con Kolivas is |
7 |
> >> working on a new kernel scheduler for Desktop/Multimedia/Gaming PCs |
8 |
> >> called "BFS": |
9 |
> >> |
10 |
> >> http://lwn.net/Articles/350100 |
11 |
> >> |
12 |
> >> Well, I've tried it. I wrote my experiences with it here: |
13 |
> >> |
14 |
> >> http://lwn.net/Articles/350820 |
15 |
> >> |
16 |
> >> If you're feeling adventurous, you might want to give that one a try. In |
17 |
> >> my case, it helped immensely, especially with sound latency and skips |
18 |
> >> and other artifacts during real-time playback (I was not using an RT |
19 |
> >> kernel before that though). Note that BFS has been updated to 0.206 |
20 |
> >> since I wrote that. |
21 |
> >> |
22 |
> >> The patch to kernel 2.6.30 and docs can be found at: |
23 |
> >> |
24 |
> >> http://ck.kolivas.org/patches/bfs |
25 |
> > |
26 |
> > and what is with people like me - who for some magical reasons don't have |
27 |
> > problems with skips or latency? Without using rt-kernels of course.- |
28 |
> |
29 |
> Fire up Ardour and record 32 channels of audio at the same time set to |
30 |
> <5mS latency using Jack and see if whatever version of the mainline |
31 |
> kernel you are running doesn't have. I've recorded as many as 48 |
32 |
> channels @ 48KHz across three hard drives at less than 2mS on my main |
33 |
> recording platform, but that requires rt-sources. I doubt I could do |
34 |
> better than about 25mS with vanilla-sources. |
35 |
> |
36 |
> Just my experience, |
37 |
> Mark |
38 |
> |
39 |
|
40 |
well, and your workload asks for rt. But rt also means overhead, reduction of |
41 |
performance. So why cater for 1 in a thousand system and punish the other 999? |
42 |
|
43 |
if you like the 'bfs' scheduler, that is great. |