I use W2K+SP3 and Fb2k shows 14% cpu usage in the task manager when playing mp3 and 1% when playing ogg, while winamp2 shows 1% for playing either one. Does that have to do with the mp3 decoder being used?

Probably the decoding, the limiting, and the dithering. But I could be wrong.

I found out why: the default output method for WinNT version is waveOut Shouldn't it be set to DirectSound instead?

output device detection added in current build (dont feel like uploading now)

~12% when playing MP3 (fb2k settings: 16bit/48000Hz - DirectSound, WinXP). Is this normal?

12% of what CPU ?
i believe there is strange conflict between fb2k and other running programs, fb2k stays at 0% here, but when i run mplayer2.exe it suddenly goes up to 4-8% (old good athlon700), i need to research it more.

Here stays in 0% and sometimes goes to 2%. Playing MPC'S only. All other dlls and extras deleted!


Microsoft Windows XP Workstation 5.1 Service Pack 1 (Build 2600)
Intel Pentium 4 2528 MHz MMX / Memory: Total - 523 MB, Free - 334 MB

Internet Explorer: 6.0.2800.1106 / ComCtl32.dll: 6.00.2800.1106 / Shlwapi.dll: 6.00.2800.1106
Shell32.dll: 6.00.2800.1145wnaspi32.dll: 4.71 (0001) , ASPI for Win32 (95/NT) DLL, Copyright © 1989-2002 Adaptec, Inc. / Aspi32.sys: 4.71 (0001)

B)

Such conflict with other applications definitely exists, and they shouldn't be necessarily related to music playback for that.

Athlon 1200MHz, 256MB Ram...

Programmes running:

- WinBar 1.2.95 (www.winbar.nl)
- Foxmail 4.2 (www.foxmail.com.cn)
- DU Meter 3.03 (www.dumeter.com)
- Trillian 1.0b Pro (www.trillian.cc)
- MyIE2 0.6.835 (http://myie2.yeah.net)
- F-Secure AntiVirus 5.40 (www.f-secure.com)

[edit]
Note that I resample (no fast mode) to 48000Hz!!

It seems like MSN Messenger is one of the conflicting programs.
Even with resampling on, it usually takes up between 0 and 1% on my computer (Athlon classic 750MHz). But then, as soon as I open MSN Messenger 5.0, the CPU usage of foobar goes up to somewhere between 17 and 21%. I close MSN Messenger, and the CPU usage settles back into somewhere between 0 and 1%.
Right now I'm thinkin, "What the hell? "
What could be the cause here?

then it's resampling what eats your cpu time, if you want to save cpu resources, don't use resampling.

yep, similar effect here with mplayer2.exe (win2k, athlon700)
i need to look into it when i don't have anything more important to do, sounds like something with OS-level sound playback interaction to me but i can't be sure now.

Guillaume> I don't think the resampler is this only thing causing your high CPU usage....
I've got a similar config to yours (just an Athlon 1800+ instead of Athlon 1200) and with resampler ON and fast mode OFF, I still got 0% CPU usage...
Try leaving Trilian... it works like MSN messenger for some things... so perhaps it has the same effects...

ahem,
resampling without fast mode IS slow. fast mode hardly eats any cpu time here either.

Yes, fast mode eats less CPU... but even without fast mode, I still got 0% CPU... (perhaps it's a bigger 0 compared to fast mode )

What's the difference with fast mode on? Is fast mode less quality than regular?

fb2k used to only use up about 1,600 k of my mem usage, w/o resampler DSP, but thne I added that to my dsp list to see what it was all about, and my mem usage shot up to 7 - 9 000 k!
I was like whoa, that's weird
So then I took it out of my seleted dsp's, and my mem usage is still up that high

This happen to anyone else?

Even when shutting down all these programmes and restarting fb2k, the cpu level remains the same...

way to read replies you get.

I've noticed this problem too. Usually it uses 0 to 1%, but as windows continues running longer and more applications are opened, closed, etc., both fb2k and winamp use up to 14% CPU (directsound, win2k, sp3).

There was one time when even a loaded plugin in winamp affected fb2k....I unloaded it from winamp and CPU was back to normal (0%) for both WA and fb2k.

Bizarre.....

huh?

Me too, fb2k uses 0% (P4 Willamate 1400MHz) to play all kind of files B)

You seem to have missed the part where you disable the resampler or turn on fast mode.



From the SSRC source code used by foo_dsp_extra:...
And DF is a frequency, with a significance I cannot explain with my level of understanding.

another conflicting program: flash plugin for Internet Explorer 6.

Normal CPU usage shows up as 0% for me, but goes to around 4 or 6% when IE6 is showing flash.

1.4GHz P4, WinXP, Via8233 DirectSound.

Total usage, or FB2k itself?

I didn't miss any reply, but why doesn't resampling to 48000Hz have as much influence on slower PC's (like psy_jc's Athlon) as on mine? Still don't get that...

0-1% with resample dsp off...

[edit]
This is so odd: now I have a 17% CPU usage (v0.36b and v0.38 test)

minimize foobar to tray, it should go back down to 1.5Mb

When I used resampling in Winamp through a DirectSound ouput plug-in, it used up about 22%. I was surprised to find that foobar does the same thing while using just 1% CPU. Perhaps its another one of your programs that's conflicting...

Could this be a DirectSound issue? Flash, mplayer2.exe, and MSN Messenger all use DirectSound don't they?

But then, on the other hand, the CPU% still remains high even when I select wave output in foobar ...

Anyone got a clue??

HIGH CPU USAGE?

1) Optimize code generally.
2) Make optimized build for P4 and other processors. If you don`t know how to make these builds - ask somebody OR (if you are too lazy/don`t know) relase full sourcecode and let someone to make these builds.

Also somebody can make external plugins optimized versions.

Technical question: Is making of P4 optimized (ICL7 compiled) versions hard? What must i have and what must i know to make P4 optimized builds?

Aren't you making a lot of assumptions here?

1. Do you know that foobar2000 is unoptimized for a fact? Peter seems to be a pretty solid programmer, so I think that telling him to "optimize" his code is at best unnecessary, and at worse kind of foolish. We know for a fact that Peter is focused on minimalism and low memory usage in foobar (and low cpu usage if you consider the fact that this issue is regarded as abnormal behavior), so you already have evidence that he is "optimizing" his code.

2. If the problem is related to Windows or directsound, then P4 optimization won't do squat. I know you that you really want P4 optimization, but there's no evidence to suggest that it would make a single bit of difference. The fact that the cpu usage jumps up only under some circumstances, but usually remains extremely low (0-5% on most peoples system), suggests that there are no general speed problems with foobar itself.

thanks, Dibrom.
to all others - reminder: fb2k STAYS at 0% on my 2.5 year old athlon700 box, playing mp3 with no DSPs enabled. if you are having CPU usage "problems" on newer computers ("problems" as in: numbers in task manager scare you away), it's something with either your settings, or some interaction between fb2k and other running programs, NOT "unoptimized" code (whatever that means).

For Dibrom:
1) - it was posted generally, not (only) for FB2K. PP is (very) good programmer indeed - i use ONLY FB2K and my conservative brother (he hates OGG, MPC etc.) uses FB2K too



but there's no evidence to suggest that it would NOT make a single bit of difference

For PP:

AthlonXP 1600+ 256mb ram DDR 2600 - approx 4% of CPU usage (NO DSP, OGG file, WinME)
For some people this value (4%) is too large (e.g. renderers of 3dmax scenes).

For same AthlonXP system with some DSP effects activated CPU usage jumps to 13-14%

Same as above but with resampler to 96000 (no-fast mode) there is approx 33% CPU usage.

You can`t say "if you don`t want high cpu usage, disable DSP effects" - Foobar should be as fast as possible with all options enabled, some people are using resampler and other effects.

It is 0-2% here without Resampler thing.
AthlonXP 1600+ 512 DDR266 WinXP SP1

for fucks sake, i repeat: if you get more than 0% CPU usage with no active DSPs on a machine faster than athlon700, its NOT caused by fb2k's audio processing.
and if you care about CPU usage, you should stay away from resampling first (wonder if you could actually abx the difference between resampling being active or not on your hardware), resampling will ALWAYS eat noticeably more CPU.

Yes, but P4 optimized version will (maybe/I think) consume less CPU (for P4 systems of course)

Maybe.

It also might make the executable 4x bigger and introduce new bugs. ICL isn't the most stable compiler...

At any rate, I think you could test some of this out yourself by downloading the SDK and compiling P4 versions of some of the plugins/etc. AFAIK, Peter doesn't have ICL.

standard DSPs are in the SDK. what are you waiting for ? go compile them.

i guess i wouldn't have much trouble getting ICL, but i've used it before and i got sick of problems caused by it.
another thing, athlon700 is the faster box i have, not gonna upgrade anytime soon because it does it's job fine, so i couldn't even test / debug those "P4-optimized builds".

So what - you think that you are big ass because you have P$ (sorry, 4)?
If it is THAT powerful, than special builds aren't required. And if it is that powerful, how it comes that in_spc crashes it?

Cobra, I must say that you beat me in annoyance thing - I think peter can definitly confirm this.

OMG. LMAO.
Lamers fighting over who is more annoying...

There you go.

So WinME is so bad ? Maybe on FB2K someone haveto wrote "Under WinME and WIN2K there is higher CPU usage"

Please do not be pairing those two together for any reason.

Use foobar2k without any programs running, and you will get 0% CPU usage......use it with either MSN Messenger or Kazaalite....and you get 2-7% CPU usage......
Blame it on MSN..........

Yup... blame it on Bill Gates and his crew...
If a small program like MSN Messenger is so full of sh*t, I wonder how Windows actually manages to run on my PC .
But damn it, a lot of people use MSN Messenger and there's not that many compatible alternatives out there. What the hell's MS thinkin'? Leave his crappy work the way it is and change a program that's perfectly fine?
I say it ain't nobody's fault except Microsoft.

Disable DSP~
It will be better,but.....

Yes, the
OGGDropXP is encoder, which speed is important and makes very big difference if used with that compiler.
But I agree with Dibrom on this point: "It also might make the executable 4x bigger and introduce new bugs. ICL isn't the most stable compiler..."
Yes, there are P4 users, but I just think it isn't worth it.
And also, if fb2k is good optimized that uses 0% on peter's Athlon 700, I don't really get the point of making special build for much faster CPUs.
Maybe I am wrong, but this is what I think.

For me the killer app is not Kaaza (dunno about MSN Messenger) but... kX Mixer. When the mixer sits in the tray, foobar2k takes 20-40% of my K6-2 300MHz processor time. If kX Mixer disabled, foobar only needs 0-6%. It's a shame because I'm so used to those kX Drivers (I've got a SB Live! Value). Actually this same happens to Winamp but it only needs up to 20% when kX Mixer started.
I'm dumb. Can anyone explain why is this happening? Is the kX Mixer somewhat broken?

When I shake my mouse around on the screen, my CPU usage (on a dual XP1600+ machine) goes up to 20%, according to WinXP Task Manager. Whoop de doo.

Are any of you CPU whiners actually doing benchmarks while Foobar is running, to see if it makes the slightest difference in your performance? If you do a 3DSMax render with and without Foobar, and the render takes 61 minutes when Foobar is playing instead of 60 when Foobar is off, does it really matter? If so, stop typing on this forum and get back to work.

Or, if it's a matter of principle that Foobar be as fast as possible, and you have the programming knowledge necessary, take your best shot at optimizing stuff yourself. It would be interesting to hear if any significant speedups can be made.

Also if cpu usage stays constantly above 10%, just stop foobar, close it, and then open it.

I found another "killer" app to add to the list... stupid fricking Quicktime tray icon that I never use.