1 |
Mark Knecht schreef: |
2 |
|
3 |
>> Can you record audio from the command line? Or do the X-based |
4 |
>> programs you use run under DirectFB? What I'm getting at is getting |
5 |
>> rid of all the obstructions that could possibly interfere with the |
6 |
>> kernel and introduce even more latency issues than what it already |
7 |
>> has |
8 |
> <snip> |
9 |
> |
10 |
> Good questions. I didn't say this earlier. I probably should have. If |
11 |
> I boot this machine into a console mode (i.e. - no xdm/gdm) and run |
12 |
> Jack in one console I can log in as root in another console, do |
13 |
> emerges all day long and I get no xruns, at least with the small |
14 |
> amount of testing I've done so far. This is using 2.6.14-rc2-mm1 so |
15 |
> it has some new code but not all of Ingo's stuff. |
16 |
|
17 |
OK, so X really is a problem, then. Now I really want to find a way to |
18 |
get you rid of it. |
19 |
|
20 |
> |
21 |
> |
22 |
>> I mean, X is a horrible hog, heaven only knows what effect your |
23 |
>> nVidia or ATI kernel modules may be having on the ability of the |
24 |
>> kernel to behave properly, since they also make demands on the |
25 |
>> kernel that 'distract' it, as it were. And if Jack is a daemon |
26 |
>> (which I know it is), it's not like it needs X for itself. |
27 |
> |
28 |
> |
29 |
> Right, but as I say, much slower PCs are able to use the standard |
30 |
> Gentoo kernel and run Gnome with no xruns. It's only this 3GHz 64-bit |
31 |
> machine that has the problem. The sound card has been used in an |
32 |
> Athlon XP 1600+ machine and it works fine so I trust its drivers at |
33 |
> least in 32-bit mode. |
34 |
|
35 |
Honestly, we don't care what much slower PCs can do, because this isn't |
36 |
one of them, and we don't think there's something wrong with the sound |
37 |
card. The issue is that this particular machine is a 64 bit one that |
38 |
apparently needs special handling in order to minimize the pre-existing |
39 |
latency issues with 64-bit kernels/drivers/environments so that you can |
40 |
use it for what you intend to use it for. Other conditions are |
41 |
irrelevant, imo. |
42 |
|
43 |
> |
44 |
>> It's of course quite possible that I'm talking out of my butt, |
45 |
> |
46 |
> |
47 |
> Not the least bit possible. Your thought are clear and very coorect |
48 |
> IMO. |
49 |
|
50 |
:-) |
51 |
|
52 |
|
53 |
> |
54 |
>> since I am not a member of the Linux audio community, but I do know |
55 |
>> that the first step in troubleshooting is to simplify the |
56 |
>> environment as much as possible, and then slowly increase the |
57 |
>> complexity to see when and where things break down. |
58 |
> |
59 |
> |
60 |
> Absolutely. Hopefully with the additional info above you'll see that |
61 |
> is what I've been doing, within my limited abilities to patch |
62 |
> kernels, etc. |
63 |
|
64 |
Patching the kernel isn't simplifying the environment if you're piling |
65 |
possibly unnecessary additional demands on the kernel. The X server runs |
66 |
on top of the kernel. The window manager runs on top of the X server, |
67 |
which runs on top of the kernel. The whole thing is rather like a head |
68 |
wound (the premise being that even non-serious head wounds tend to heavy |
69 |
bleeding, obscuring the nature and severity of the wound itself). The |
70 |
use of the X server, and anything but the lightest possible WM puts |
71 |
additional stress on the system, which may be the straw that breaks the |
72 |
camel's back in this case. |
73 |
> |
74 |
> |
75 |
>> Were I you, I would consider: |
76 |
>> |
77 |
>> - If keeping X, switching to the absolute most minimal wm possible |
78 |
>> (twm, ratpoison, ion), to see what effect that had. |
79 |
> |
80 |
> |
81 |
> Right. FVWM, fluxbox, etc. These can just be tested. |
82 |
|
83 |
No, I really mean twm, ratpoison, ion and the like. FVWM can be |
84 |
configured to be absolutely minimal, but learning to do that is an |
85 |
unreasonable distraction. Fluxbox uses too much X (has to draw toolbars |
86 |
and tabs and decorative windows). Even openbox might, and I don't know |
87 |
enough about pekwm or kahakai to know if they would be appropriate. |
88 |
|
89 |
If you must use X (which I will accept for the moment) for the GUI |
90 |
applications, well, fine, but what I'm suggesting is a window manager |
91 |
that uses the absolute minimum of X resources possible. |
92 |
|
93 |
> |
94 |
> |
95 |
>> - If downstepping from X, investigating what programs run under |
96 |
>> DirectFB and seeing what effect that had. - If going cold-turkey |
97 |
>> off X, seeing how far you get with the command-line and ncurses |
98 |
>> programs. |
99 |
> |
100 |
> |
101 |
> Neither are really acceptable as far as I know today. |
102 |
> |
103 |
> |
104 |
>> Am I, in fact, talking out of my butt (since it seems that the |
105 |
>> 'real' audio community would have tried at least some of this)? Or |
106 |
>> are there reasons that this simplification process is not possible |
107 |
>> for professional audio recording? |
108 |
> |
109 |
> |
110 |
> As above - see Ardour, Jamin, Muse, Rosegarden, etc. |
111 |
|
112 |
I'm not completely convinced that Ardour, Jamin, Muse and Rosegarden |
113 |
won't run under DirectFB, but I'm not so experienced with DirectFB that |
114 |
I can say definitively one way or the other. |
115 |
|
116 |
I see that at least Muse does have an ncurses interface (or at least an |
117 |
ncurses USE flag which would suggest that it has an ncurses interface). |
118 |
|
119 |
And looking at the DirectFB site, it seems possible that there could be |
120 |
a place for it to help work around the issue: |
121 |
|
122 |
FusionSound |
123 |
Audio sub system for multiple applications |
124 |
FusionSound is a very powerful audio sub system in the manner of |
125 |
DirectFB and a technical demonstration of Fusion. |
126 |
FusionSound supports multiple applications using Fusion IPC. It |
127 |
provides streams, static sound buffers and control over any number of |
128 |
concurrent playbacks. Sample data is always stored in shared memory, |
129 |
starting a playback simply adds an entry to the playlist of the mixer |
130 |
thread in the master application. |
131 |
FusionSound currently is a module of DirectFB. The current API is fully |
132 |
implemented and the complete code is smaller than just the header file |
133 |
of DirectFB! There's a complete API Reference to learn more about |
134 |
FusionSound. |
135 |
FusionSound future plans include interfaces for hardware/software |
136 |
codecs, hardware gain control, device configuration and different device |
137 |
backends (currently OSS only). |
138 |
|
139 |
XDirectFB |
140 |
Rootless X Server based on DirectFB |
141 |
XDirectFB Screenshot |
142 |
XDirectFB is a rootless X Server using DirectFB windows for X11 |
143 |
toplevel windows. This way you can adjust the opacity of every |
144 |
application with your mouse wheel (while holding Meta down over a |
145 |
window). More details about these shortcuts can be found in the DirectFB |
146 |
README. |
147 |
Window movements are initiated by the applications or the window |
148 |
manager. The graphical movement is done by DirectFB using available |
149 |
hardware acceleration. |
150 |
Overlapping toplevel windows do not cause expose events, i.e. redrawing |
151 |
of the window contents, as they are DirectFB windows and therefore have |
152 |
an own surface, a.k.a. backing store. |
153 |
You should use the CVS version, because the last release is too old. |
154 |
|
155 |
Qt on DirectFB |
156 |
qt-directfb is a GPLed DirectFB port of Qt based on Qt-X11 3.2. |
157 |
|
158 |
Qingy (I've used this, it's nice enough. I mean, it's just a DM, after |
159 |
all, so not much to say) |
160 |
Qingy is a replacement of getty. Written in C, it uses DirectFB to |
161 |
provide a fast, nice GUI without the overhead of the X Windows System. |
162 |
It allows the user to log in and start the session of his choice (text |
163 |
console, gnome, kde, wmaker, ...). |
164 |
|
165 |
wyoDesktop |
166 |
wyoDesktop is an effort to create a graphical desktop environment where |
167 |
an ordinary user immediately feels comfortable through the use of a well |
168 |
designed and consistent look and feel. |
169 |
|
170 |
I know I'm very hardheaded (I have even been so evaluated by |
171 |
professionals in vocational assessment testing, so it's really true), |
172 |
but the upside is that I tend to believe that there is a solution or |
173 |
workaround to most problems, and I tend to go a long way towards |
174 |
twisting myself into a pretzel to find it (and the upside of that is |
175 |
that if I tie myself into a pretzel and find nothing, then there's |
176 |
really likely nothing to find). |
177 |
|
178 |
I understand that serious audio recording has serious problems under |
179 |
Linux, but I also know that many people are very interested in finding |
180 |
solutions. I just wonder if many of those solutions may not lie 'outside |
181 |
the box' at this time. It's within the realm of possibility that this |
182 |
could all work effectively (if not elegantly), if radical changes were |
183 |
made to the PC's setup, with the sole purpose of getting it working. |
184 |
|
185 |
Oh and by the way-- I noticed that both directfb and ardour (I think it |
186 |
was) seem to prefer Matrox cards. I've got two in the house (though only |
187 |
one is mine), neither in use-- a G400 and a G400Max (but the fan on the |
188 |
Max keeps falling off :) as I'm missing a screw I think, otherwise it's |
189 |
fine). |
190 |
|
191 |
Wanna discuss a trade for any old nVidia card you might have laying |
192 |
around? Seriously. |
193 |
|
194 |
Holly |
195 |
-- |
196 |
gentoo-user@g.o mailing list |