Subject: RE: [linux-audio-user] 24bit 96khz cards that work well
From: Mark Knecht (mknecht_AT_controlnet.com)
Date: Tue Dec 31 2002 - 01:03:40 EET
It could be a huge number of things, but like you I doubt it's just
interrupts. As you say, no card is going to generate more interrupts than it
needs to. Some cards could require a few more interrupts to get the job
done, but I doubt they'll generate that many more interrupts.
More likely is the quality of the DMA controller in the sound chip, and how
well it gets the data across the bus. If a card has to deliver 1024 bytes,
and can do that in one burst of 256 cycles, while another card delivers the
same 1024 bytes in 8 bursts of 128 bytes, the first card will be much more
efficient and better behaved in the system, and probably use less PCI time
This is also HEAVILY driven by what the chipset will allow a PCI device to
so, hence my RME and your RME cards, while identical, might give different
performance results because our PC chipsets give it different access to
system resources like memory.
To answer Brian's original question, I don't think Jack yet supports 96K,
but the RME cards do, and they work pretty nicely. Unfortunately they are
One thing I hope Brian understand is that not every xrun is caused by the
sound card. Many (IMO most!) xruns are potentially caused by the disk
subsystem. Study that machine carefully Brian!
With best regards,
[mailto:linux-audio-user-admin_AT_music.columbia.edu]On Behalf Of Anthony
Sent: Monday, December 30, 2002 2:15 PM
Subject: Re: [linux-audio-user] 24bit 96khz cards that work well
* Brian Redfern <bredfern_AT_calarts.edu> [Dec 30 02 14:03]:
> Well, I figured out that the culprit to my xrun problem wasn't my 533mghz
> processor, its my envy24 chipset based card, from the alsa wiki:
> "Many but not all users complain about sound glitches. This may be due to
> the cards being IRQ hungry. (VU meter interface?) -- Tobiah"
> Given that the evil envy24 chip is in a lot of cards out there, I'm
> wondering what cards people have had better results with?
Also, out of interest, what does IRQ hungry mean? Isn't the frequency
of interrupts determined by HW parameters you choose? How can there be
any 'extra' interrupts. You mentioned in an earlier thread that your
SB card didnt seem so interrupt hungry, but are you sure the device is
opened with the same params? I'm no expert. Can anyone explain this?
This archive was generated by hypermail 2b28 : Tue Dec 31 2002 - 01:03:45 EET