glitch with dsound+gfx

Paul Morel PaulM at
Tue Nov 3 18:31:40 EST 1998


I have seen a couple of software synths that are  bundled with sound
cards which use F/P.   These synths had F/P state problems with other
floating point applications.

Microsoft fixed this issue in DirectX 3.X,   so F/P state is saved
between processes.  I do not know if this also saved the MMX state as

	From:  Richard Dobson [SMTP:RWD at]
	Sent:  Tuesday, November 03, 1998 3:05 PM
	To:  Multiple recipients of list
	Subject:  Re: glitch with dsound+gfx

	I have a Mystique, on a humble Pentium133, and have not had any
	problems as such (though I have not been doing anything with
	I am however still highly suspicious about MMX, as on a Pentium
	uses the same registers as are used for floating-point. If the
audio is
	itself using f/p a lot, perhaps there is a clash there. Has
	done/know of any tests on this aspect of MMX Pentiums?  

	Richard Dobson

	Douglas Wright wrote:
	> I have heard of problems with Matrox drivers hogging the bus
	> Windows.  I have no direct experience with it though.
	> dug
	> >
	> >I noticed a disturbing glitch in my DirectSound output when I
	> interacted
	> >graphically with my Java programs.  I made my buffers really
huge, and
	> >reduced the Java program down to just drawing big rectangles
but still
	> >got glitches. It appears that a burst of graphics activity is
	> >interfering with the actual DirectSound playback, not my
	> code.
	> >Has anyone else noticed this behavior? Could a DMA or IRQ
	> cause
	> >this?
	> >
	> >java = JDK1.1.6 and JView (recent)
	> >dsound.dll = V4.05.00.0155
	> >gfx = Matrox Mystique
	> >CPU = 233 Mhz MMX
	> >OS = Win95 B
	> >
	> >Thanks,
	> >--
	> >Phil Burk
	> >
	> >mailto:philburk at
	> >
	> >

More information about the music-dsp mailing list