[Pilou42]

Ok. No problem. I just asked in case I missed something.
But sorry, it's a blocker in my case.

[Dynamite]

soo... which one is better? dmaHD or mitsubishi's one?

[SubJunk]

dmaHD audio is included in Mitsu's build along with other good bugfixes and enhancements so I recommend using Mitsu's build
also hey man good to see ya around here :)

This post has been edited by SubJunk: 26 January 2011 - 08:22 AM

[v01d]

Hi all, congratulations p5yc0runn3r, nice work you've done. I have a question about the cvar dmaHD_interpolation: what about the differences between setting it to 0, 1, 2 or 3 in terms of sound quality and performance?

[p5yc0runn3r]

v01d, on 13 February 2011 - 05:23 AM, said:

Hi all, congratulations p5yc0runn3r, nice work you've done. I have a question about the cvar dmaHD_interpolation: what about the differences between setting it to 0, 1, 2 or 3 in terms of sound quality and performance?

Urban Terror uses 22Khz sounds which by today's standards are low. dmaHD uses 44Khz mixing but using 22Khz sounds it is still not good so the dmaHD_interpolation CVAR is there. This CVAR tries to fill in the gaps and upscales the current sounds from 22Khz to 44Khz using interpolation techniques. As of performance, there is not much difference since sound is pre-allocated and pre-interpolated.
The best sound quality is to set the CVAR to 3 which uses 4-point Hermite spline interpolation while second best is 2 which uses 4-point Cubic spline interpolation. Normal quality comes from 1 which uses just linear interpolation which worst quality is 0 which does not interpolation at all.

I hope this answers your question.

This post has been edited by p5yc0runn3r: 14 February 2011 - 02:35 AM

[v01d]

Hi P5yc0runn3r, when set in_mouse "1" in q3config.cfg can not start the client, it fails, gives no error message, just not start. Options in_mouse "-1" and in_mouse "2" do not work properly on my system (in_mouse "-1" also works well with pure client)
I'm not using the Mitsubishi's builds, I have drop fps with the same. I am using windows 7 but found that too fails in Windows XP.

There are available a build of the client that includes the dmahd sound system but without the raw mouse support?

Thanks in advance.

[undead]

p5yc0runn3r, on 14 November 2010 - 08:37 AM, said:

No no I'm not but until it is finalized I'm not making it public. I'll send it to Mitsubishi now...

Is there a standalone diff for this? Something other than what's in mitsubishi's build? I keep in sync with the latest ioquake3 source and it would be a pain to maintain all of those changes that I don't care about. I want to make sure that I'm including only the dmaHD changes.

If I pull all the changes in the comment at the top of snd_dmahd.c, is that sufficient?

[undead]

p5yc0runn3r, on 09 November 2010 - 03:49 PM, said:

The following are some other CVARS that affect dmaHD. Please set them as specified:
/com_soundMegs
This has no effect anymore. This should be set to default "8"
(This option needs a total game restart after change)

Is this still the case? The only code I have to go from is mitsubishi's and it doesn't agree with the above. Here's part of the diff from mitsubishi's build:

diff -ruN -x '*svn*' -x '*gitignore*' -x build -x 'testb*' -x 'testcp*' ../vanilla-untouched//code/client/snd_mem.c ../ioq3-urt//code/client/snd_mem.c
--- ../vanilla-untouched//code/client/snd_mem.c 2010-07-22 06:21:11 +0300
+++ ../ioq3-urt//code/client/snd_mem.c  2010-11-20 18:54:26 +0200
@@ -32,7 +32,9 @@
 #include "snd_local.h"
 #include "snd_codec.h"

-#define DEF_COMSOUNDMEGS "8"
+extern cvar_t *com_quiet;
+#include "snd_dmahd.h"
+#define DEF_COMSOUNDMEGS "32"

 /*
 ===============================================================================
@@ -81,7 +83,9 @@

        cv = Cvar_Get( "com_soundMegs", DEF_COMSOUNDMEGS, CVAR_LATCH | CVAR_ARCHIVE );

-       scs = (cv->integer*1536);
+       if (cv->integer < 16) Cvar_Set("com_soundMegs", "16"); //dmaHD optimization
+
+       scs = (dmaHD_Enabled()) ? (2*1536) : (cv->integer*1536); // not needed by dmaHD

        buffer = malloc(scs*sizeof(sndBuffer) );
        // allocate the stack based hunk allocator

As you can see, it sets the default to 32, conditionally sets it to 16, and then ignores the setting and uses 2 to alloc the buffer. Why is it not using 8 like you recommend above? This is why I was interested in having a diff with only dmaHD. I can't tell whether your post or mitsubishi's code is out of date.

[p5yc0runn3r]

The difference between this and Mitsu's build is that this build is based off the original UrbanTerror executable code while Mitsu's build uses all the latest ioQuake source code with SDL and stuff. The dmaHD code is different but both are good. The CVAR definitions in this post is mostly for this build while the most obvious CVARS are for both this and Mitsu's build.

[flajaz]

This rocks.