[LAU] Massive Xruns during recording
David Kastrup
dak at gnu.org
Thu Nov 30 18:30:49 UTC 2017
"Jeanette C." <julien-3g/GUrBLTNIb1SvskN2V4Q at public.gmane.org> writes:
> Nov 30 2017, Peter has written:
> Dear Peter,
> ...
>> this looks indeed pretty slow, is it an external usb-2 drive?
> No, it's an internal S-Ata drive, apparently a SAMSUNG HD154UI.
>> You may perform a simple check:
>>
>> sudo hdparm -tT /dev/sda
> For cached reads this averages around 1200 MB/Sec, for buffered reads
> around 108 MB/Sec. I didn't find an option to test the writing speed. As
> a write test, I tried this:
> dd if=/dev/zero of=out.file bs=1M count=100
> with a result of around 560MB/Sec
>> If your are just recording, can you increase the buffer length in jack?
> I could, but I already have it a period size of 256. At 512 I started to
> get a noticeable delay. With cards that can potentially go down to 64
> samples, I feel a - perhaps unreasonable - hesitation to set the period
> higher. Is this very unreasonable?
Well, the problem is not as much your cards but the disk I/O. Getting
more buffers will already help (-n 3 instead of the default -n 2). You
can try improving this problem by using rtirq to distribute realtime
priorities: do you already use threaded irqs?
--
David Kastrup
More information about the Linux-audio-user
mailing list