1 |
On 05/22/2010 07:59 PM, Canek Peláez Valdés wrote: |
2 |
> On Sat, May 22, 2010 at 1:21 AM, Nikos Chantziaras<realnc@×××××.de> wrote: |
3 |
>> Latency is the delay between giving the order to play a sound and the sound |
4 |
>> actually being played. It's usually around 30ms here with ALSA/dmix, and |
5 |
>> around 10ms with OSS/vmix. It's not funny trying to play something in a |
6 |
>> software synth with a keyboard when having a 30ms latency. |
7 |
> |
8 |
> As I said, you're doing it wrong. No "normal" (average desktop, media |
9 |
> center, laptop, linux-phone) user needs 10ms of latency in audio. |
10 |
> That's overkill. Yours is a special case, and you need special |
11 |
> software. Try Jack. |
12 |
|
13 |
I don't do professional audio. I have a normal PC. And just like I |
14 |
sometimes use a synth in Windows (I'm just a hobbyist), I'd like to do |
15 |
the same in Linux. |
16 |
|
17 |
Windows: I don't need Jack there. Audio latency is low even with |
18 |
non-ASIO drivers. |
19 |
|
20 |
Linux: I suddenly need "Jack" and specialty hacks and must do without a |
21 |
mixer! No thanks. OSSv4 allows me to use my machine in the same manner |
22 |
as Windows: It just works and does the right thing regardless of the |
23 |
application I'm running. |
24 |
|
25 |
ALSA/Pulse needing third-party stuff just to get basics right |
26 |
(acceptable latency; not *ultra* low latency, just acceptable one) is a |
27 |
sign that they're not designed right. |
28 |
|
29 |
And OSS4 dying because of kernel-mixing is a bit far-stretched. "No FP |
30 |
mixing in kernel" is Linux-specific. Other kernels don't have a problem |
31 |
with that. |
32 |
|
33 |
And in the end, you know what? Even if OSS4 had a broken design, it's |
34 |
still better, because it works better. At least it gets the basics |
35 |
right. Other operating systems are much more advanced in that manner. |
36 |
It's ALSA that holds Linux audio back. |