Craig Lamparter

Using "-n 2" on the direwolf command line with no other options and no config file fixes the problem. cpu usage is normal (17%) related unresolved thread, which also uses...

replicated on Buster, direwolf 1.6 using "-n 2" or "ACHANNELS 2" is a fix in this case as well

``` PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 580 pi 20 0 139440 10744 7812 S 100.0 2.2 3:38.49 direwolf ``` MYCALL KM6LYW-2 IGSERVER noam.aprs2.net...

Using "-r 22000" also fixes the 100% cpu thread. ``` PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 1962 pi 20 0 132268 11372 7900 S...

Thanks David, this isn't a USB device (I2C device), and yah, it has something to do with the supported hardware rates for sure. 32K seems to be a hardware rate...

Understood. No Pulse Audio -- ever. libasound2 honors "ALSA_CARD", so direwolf does by extension. I get the same result when specifying an audio device in direwolf.conf (ADEVICE plughw:0,0) no dmix...

Increasing frames per period in audio.c resolves the issue. ``` fpp = buf_size_in_bytes / (pa->adev[a].num_channels * pa->adev[a].bits_per_sample / 8); //craiger - this normally comes out to 512 fpp = 1024;...

Thanks for the info. If we're dividing sample rate by 3 on ARM, is there a downside to simply specifying a rate of "14700"? That yields 10% cpu on a...

**Bookworm** observations. 32bit/lite, Direwolf 1.6, Linux 6.1.0 CPU usage is much better than bullseye, but significantly higher when using 44100 or 48000 44101 12% (10.0user, 2.3sys) 44100 20% ( 6.3user...

dbgsym version of alsa-libs compiled, this is where the direwolf process is spending time, still need to see which driver the "sys" time is being spent... 22.82% direwolf direwolf [.]...