What is the current situation of low-latency audio driver
support in SDL? Will there be such thing in 1.3, or do I
need to wait for 2.x? Or have miracles happened and is
it already available?
When I used Amiga, audio latency was not an issue. Now
that I have otherwise superior setup, it seems that audio
drivers have made progress backwards 
– Timo Suoranta – @Timo_K_Suoranta –
Hi All!
I’ve discovered that probably the RH RPMs are not working properly on some
machines (I face two from 5 machines I’ve met until now). I’ve been
playing with the CVS SDL12/configure file and after I commented out the
USE_ASMBLIT directive definition I finaly got SDL working.
I’ve the suspicion that it is caused by RH gcc 2.96, but I don’t have it
confirmed. Is it know issue or not? Do you compile the www.libsdl.org/RPMs
with gcc 2.96? Would the USE_ASMBLIT commenting out noticable performance
slowdown? Do you know anybody about this bug?
regards
STan
---------------------- Cut ------------------------
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 1026 (LWP 1026)]
0x4005efd3 in _Hermes_X86_CPU () at eval.c:41
41 eval.c: No such file or directory.
in eval.c
(gdb) bt
#0 0x4005efd3 in _Hermes_X86_CPU () at eval.c:41
#1 0x00000084 in __strtol_internal (nptr=0x8216fa0 “”, endptr=0x0,
base=136408312, group=1073934622) at eval.c:73
#2 0x4002f2bd in SDL_CalculateBlit () at eval.c:41
#3 0x400416ad in SDL_MapSurface () at eval.c:41
#4 0x40041d63 in SDL_LowerBlit () at eval.c:41
#5 0x400442dd in SDL_UpdateRects () at eval.c:41
#6 0x4004416d in SDL_UpdateRect () at eval.c:41
(gdb)
Standa Opichal wrote:
I’ve discovered that probably the RH RPMs are not working properly on some
machines (I face two from 5 machines I’ve met until now). I’ve been
playing with the CVS SDL12/configure file and after I commented out the
USE_ASMBLIT directive definition I finaly got SDL working.
This is likely the problem I had with SDL built with NASM 0.98
(the redhat-packaged NASM). Either build without the asm routines
or get a new nasm (0.98.08 seems to work)
(I wrote about the bug on this list a couple of days ago, so this should
not come as a surprise)
I’ve no idea, but I think SDL has been using shared buffer style I/O on
some targets for a good while - which means that most of the low level
work is already done.
However, an API extension would be required, as “zero latency mixing” is
done in a pretty different way to the normal method. Something like:
/* Get info on the circular hardware audio buffer */
SDL_AudioSpec *SDL_GetAudioBuffer();
/* Get current audio output positionOn Monday 15 October 2001 13:18, Timo K Suoranta wrote:
What is the current situation of low-latency audio driver
support in SDL? Will there be such thing in 1.3, or do I
need to wait for 2.x? Or have miracles happened and is
it already available?
*
* Start mixing low latency sound events slightly
* ahead of this position, and it will be played
* with practically no latency. Don't start right
* *at* this position, as that will cause the first
* few samples to be lost - how many depends on
* driver, OS, hardware and CPU speed.
*/
int SDL_GetAudioPlayPosition();
/* Get current audio clalback mixing position
*
* The old style callback writes it's data in
* front of this position, and then bumps it
* forward. Don't mix ahead beyond this position,
* or you data will be overwritten!
*/
int SDL_GetAudioMixPosition();
Obviously, using this API, you lose automatic format and rate conversion,
as you’re dealing directly with the hardware DMA buffer. As to systems
without this kind of audio API, the SDL_GetAudioBuffer() call should
return NULL, IMHO - it could be emulated, but that would result in
worse performance than the current API when done on this level.
When I used Amiga, audio latency was not an issue. Now
that I have otherwise superior setup, it seems that audio
drivers have made progress backwards
Hey, AmigaDOS was a microkernel, almost an RTOS, and you could even
bypass it entirely! 
Not a fair comparison - although I do agree that
some things are much worse these days, despite hundreds of times more
power. (In fact, some things are so bad you can hardly tell that a P-III
933 really is faster than an Amiga… What good is 800 fps rendering
power without retrace sync, for example? Still looks like crap - even a
C64 can do better. heh)
//David Olofson — Programmer, Reologica Instruments AB
.- M A I A -------------------------------------------------.
| Multimedia Application Integration Architecture |
| A Free/Open Source Plugin API for Professional Multimedia |
----------------------------> http://www.linuxdj.com/maia -' .- David Olofson -------------------------------------------. | Audio Hacker - Open Source Advocate - Singer - Songwriter |-------------------------------------> http://olofson.net -’
Hey, AmigaDOS was a microkernel, almost an RTOS, and you could even
bypass it entirely!
that some things are much worse these days, despite hundreds of
times more power. (In fact, some things are so bad you can hardly
tell that a P-III 933 really is faster than an Amiga… What good
is 800 fps rendering power without retrace sync, for example? Still
looks like crap - even a C64 can do better. heh)
Yeah =D–
Trick
Linux User #229006 * http://counter.li.org