You are not logged in.
Yes that's what I'm wondering. Is Pulseaudio using more energy than ALSA used alone? Maybe because it's more sophisticated designed or so? I tried to google the answer but couldn't find out much.. Just some informations that Pulseaudio could be configured to cache a lot and such reduce the number of interrupts.
Do you have more information on this topic?
Offline
power saving measures are not something of the sound infrastructure, but the kernel module for your sound hardware.
Offline
cache a lot
What matters is the size of the hardware buffer. And it's only *ALSA* that talks to the hardware. ALSA can cache too (e.g. buffer_size).
Speculating, I could only imagine Pulseaudio reducing power, if the app is ludicrously badly written. In reality, I've never heard of such an app.
Note that Pulseaudio, being a middleman, is yet another app for the CPU to run, thus taking more CPU power.
Offline
According to Lennart Poettering it could result in power saving
http://0pointer.de/blog/projects/pulse-glitch-free.html
http://0pointer.de/blog/projects/pa-and-power.html
Offline
He's the author, talking about what could possibly be the best scenario, assuming that pulseaudio worked perfectly - biased much
I'm biased towards believing it's the work of Satan, after seeing hundreds of threads by users wondering how to get their audio working, and having the additional layer of Pulseaudio just be an additional source of confusion for them
Offline
I see the only reason to use pa - if you need echo cancelation for your audio/video conferencing. At all other cases alsa is the only way (to my taste, of course). To be happy with alsa just digg in slightly how to use dmix alsa plugin to permit multiple applications simultaneous access to audio devices.
"I exist" is the best myth I know..
Offline
In fact, I used to be a firm believer that alsa was the way to go, and when installing pre-built distros one of the first things I did was remove pulse.
Unfortunately, I've started encountering the cracks in alsa recently, too, and have come to the conclusion that the only sound server worth my time is OSS.
Alsa fixed problems I had with pulse, and OSS has fixed problems I had with alsa. I'm hoping I don't start seeing cracks in OSS, because there's nowhere to go from there
Ryzen 5900X 12 core/24 thread - RTX 3090 FE 24 Gb, Asus Prime B450 Plus, 32Gb Corsair DDR4, Cooler Master N300 chassis, 5 HD (1 NvME PCI, 4SSD) + 1 x optical.
Linux user #545703
Offline
What I could say about....well. On my computer seems, I'll repeat, seems that ALSA makes the system smoothly. I don't know exactly, but, with pulse, it's more or less like a 5% heavyer than alsa....
Repeat.... it seems.....
Offline
At any case pa uses alsa. Pulse is an additional tie between apps and drivers.
"I exist" is the best myth I know..
Offline
Pulse is a touchy subject, you'll be hard pressed to find factual data on power consumption with and without pulseaudio without the poster being biased to pulse (favorable or unfavorable)
I suggest you do live testing on your box manually to find out what provides the best result for you, power consumption is going to depend on the alsa driver as well as the use of pulse and isn't generally the same for every system. Currently, it's the only way to get a solid useful answer.
Last edited by stefanwilkens (2012-01-28 18:35:32)
Arch i686 on Phenom X4 | GTX760
Offline
what annoys me is that pulseaudio (?) seemingly keeps the soundcard active - well at least powertop claims so
but after reading http://linux-tipps.blogspot.co.at/2011/ … -alsa.html I'm not so sure if that matters so much ...
Hardcore Linux user since 2004
Offline
Yes that's what I'm wondering. Is Pulseaudio using more energy than ALSA used alone? Maybe because it's more sophisticated designed or so? I tried to google the answer but couldn't find out much.. Just some informations that Pulseaudio could be configured to cache a lot and such reduce the number of interrupts.
Do you have more information on this topic?
PulseAudio, if working as intended, should be able to reduce your powerusage a lot. If your music player informs PA that it is playing music (rather than something which requires low latency such as gaming or chatting) then PA will tell ALSA to increase the kernel buffers and only wake up to re-fill them once per second.
When I last tested this the difference in wakeups/second want from roughly 200 when using pure ALSA to less than 5 using PA. This corresponds to a couple of watts difference.
Obviously your milage might vary, as optimizing your sound software won't make much of a difference if something else on your system keeps your CPU awake, there might be issues/bugs with your particular hardware/software, etc.
The best would be: use powertop to measure the before/after powerconsumption (you'll probably be best off if you use PA in combination with a gstreamer based app).
Offline