Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 26 Nov 2005 12:20:20 +0800
From:      Ariff Abdullah <skywizard@MyBSD.org.my>
To:        pyunyh@gmail.com
Cc:        freebsd-multimedia@freebsd.org
Subject:   Re: Sound card recommendations?
Message-ID:  <20051126122020.5b15de14.skywizard@MyBSD.org.my>
In-Reply-To: <20051126034958.GB22479@rndsoft.co.kr>
References:  <20051124215447.GA790@trimind.de> <20051125131548.sx9hpr45dvoc0gss@netchild.homeip.net> <20051125153348.GA6748@trimind.de> <20051125235318.06250835.skywizard@MyBSD.org.my> <20051126034958.GB22479@rndsoft.co.kr>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, 26 Nov 2005 12:49:58 +0900
Pyun YongHyeon <pyunyh@gmail.com> wrote:
> 
> In the days of FreeBSD 2.2.x, I also used AWE64 with sound font. It
> was well made card and I'm satisfied with its quality of sound. Now
> I had to use SB Live due to lack of ISA slots and I lost
> bass/treble/MIDI support and suffered from stuttering issues. :-(
> 
> Since I noticed some audio cards does not show stuttering issue(even
> with Giant lock and without PREEMTION option), I guess the real
> issue is in driver itself. ATM it seems that increasing pcm
> buffersize is the only way to remedy the issue. As you know, it has
> side effect of lagging audio playback and DMA memory extravagance.
> I'm really like to make emu10k1(4) to use its hardware MMU. This
> would be the prerequisite condition to support sound font. However
> due to lack of documentation, it's not easy to add this capability,
> I think.
>
Another technique to eliminate stuttering issue without increasing
buffersize is to increase DMA block count, at the cost of increasing
interrupt rate. Unfortunately, most driver does not adhere with this
property changes an assume static block allocation early during device
attach and DMA segment allocation. This can be changed in future.

>  > >  Without lapsing into an audiophile discussion, the Live
>  > > card sounds just ?flat? in comparison to the AWE64 with
>  > > properly adjusted bass and treble mixers.
>  > > 
>  > This is a known deficiency of sblive especially among
>  > audiophiles. Its internal sample rate converter (convertor?) and
>  > the behaviour of sampling everything out as 48k are pretty much
>  > unsatisfactory.
>  > 
>  > Now, if you're talking about crackling or stuttering sound, you
>  > should give those patchsets a try. If you think something had to
>  > be done about the pci latency timer, you can use pciconf(8) to
>  > adjust/increase its value or perhaps using this wrapper:
>  > 	http://people.freebsd.org/~ariff/utils/pcilattimer
>  > .. since pci latency_timer is available only for snd_es137x.
>  > 
> 
> Probably not, as maestro(4) also provides R/W access to PCI latency
> timer in PCI configuration space. I guess all sane PCI interface
> hardwares provides R/W access to the register. It seems that cheap
> PCI audio hardware are exception.
> 
I was talking about sysctl hw.snd.pcm<unit>."latency_timer", which is
currently specific only for snd_es137x. Few motherboards indeed have
option to tune general PCI latency timer settings, or if not, use
pciconf(8).


--
Ariff Abdullah
MyBSD

http://www.MyBSD.org.my (IPv6/IPv4)
http://staff.MyBSD.org.my (IPv6/IPv4)
http://tomoyo.MyBSD.org.my (IPv6/IPv4)



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20051126122020.5b15de14.skywizard>