Re: [SOLVED] Crackles in audio, drifting intermittent noise etc.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Mon, 31 Oct 2016, termtech wrote:

Yeah, I know. Very strange. Single card. No other audio enabled.
Onboard audio disabled.

OK, I have a similar audio interface (delta 66) with the same chip (ice1712) and do not have this problem.

No trouble there. It's on internal 44KHz clock, not spdif or external clock.

I use 48k for everything. I think the analog filtering works better with it. But filtering is not likely to make this difference.

IIRC Someone told me that a quad core is two chips each with two cores ??

Depends on the chip. i3 is dual core with hyperthread, looks like 4 cores.
i5 should be 4 single cores (mine is... I made sure when I bought it).
i7 should be 4 cores with HT to look like 8 (why I didn't bother with the extra expence). _But_, almost all i3, i5 and i7 in portable (laptop and nuc) computers are two cores with HT. This has to do with heat handling. I consider this to be a way of selling an i3 as an i7 (though there are some other advantages to the i7 besides core number).

[With four CPU cores:]
========================
cat /proc/cpuinfo
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 6
model           : 30
model name      : Intel(R) Core(TM) i5 CPU         750  @ 2.67GHz

looked this up:
http://ark.intel.com/products/42915/Intel-Core-i5-750-Processor-8M-Cache-2_66-GHz
this chip does not have hyper threading so that is not the problem for sure. It does have "boost" technology which I turn off for the same reason I set speed steady state (Performance mode). Boost means that the OS can tell the cpu to run at up to 3.2Ghz and it will run as fast as core temperature will allow. I have personally found that for anything below 64/2 the cpu speed must be steady state I can set the governor to userspace and set the cpu speed to 800Mhz and get better performance (latency wise) than with Ondemand up to 3.2Ghz. I also monitor my CPU temperature for two reasons: One to make sure that in performance mode I don't cook things/the cpu doesn't change speed to reduce over heat. It is also a good indication of time to clean the dust out. I run the cpus at 100% load quite often, not while doing audio, but compiling Ardour. If that doesn't over temp things... audio won't either.

So clock sync issues... As I have said I run basically the same setup (though I do have more than 1 audio IF going) and have no problems. I can run the D66 at 16/2 with no hickups. I am not sure, but I think the D66, like the LT has the converters on the card and the D66 box is just a breakout box unlike the D1010 which has convertors in the BOB.

So what is different between running two cores or four? Cpu load is higher with two. That would be a Performance/Ondemand governor difference. However, the size of the cycle does not make sense with that. nor the idea that the distortion fades in and out. Speed changes give xrun right at downshift (for me anyway).

Buffer access: buffer access should be interupt driven... new MB maybe only one PCI slot? I purposely chose a new MB with three PCI slots. I use two as I use an old ensoniq AudioPCI for MIDI ports and I wanted to be able to use any PCI slot that shared irqs (probably irq 16). I don't use that slot. So check /proc/interupts to see if you are sharing irq for ice1712. A usb port is common on irq16 too. How is your /etc/default/rtirq setup?

Try running 48k instead of 44k1 and set jack up as */3 and see if that makes any difference. (48k with whatever/3 should line up with just about anything/sec such as USB) I don't think 44k1 can line with anything... which should be an advantage... but making things worse tells you things too.

Finally, Caps. Because the board is inside the case... try (with the case open) bringing your finger close to different parts of the audio interface card while it is making noise. does it make any difference? Your finger is a portable capacitor for testing :) Making a grounded metal case to go around the card may help too. (it may bring the noise floor in general down too)

--
Len Ovens
www.ovenwerks.net

_______________________________________________
Linux-audio-user mailing list
Linux-audio-user@xxxxxxxxxxxxxxxxxxxx
http://lists.linuxaudio.org/listinfo/linux-audio-user



[Index of Archives]     [Linux Sound]     [ALSA Users]     [Pulse Audio]     [ALSA Devel]     [Sox Users]     [Linux Media]     [Kernel]     [Photo Sharing]     [Gimp]     [Yosemite News]     [Linux Media]

  Powered by Linux