Clearly you have no clue what I said....
Originally Posted by seeker010
And uumm how is that comparison fair? You take the one of the most expensive C-Media boards and compare it with one of the cheapest X-fi boards....
Not to mention that the C-media board has --way-- better drivers and --way-- better op amps. Not to mention an entire of team of open source developers backing it.
How about this for a better comparison...
A fairer comparison though the Auzentech will sound better due to it's better op amps, and will work better due to its bettter drivers.. And will cost less due to it's lack of hardware mixer.... And to top it all off it even has better connectivity and is capable of decoding far more formats....
Last edited by duby229; 04-19-2008 at 07:15 PM.
Thank you, and I guess I should retract my claim then. It was based on anecdotal evidence.
Originally Posted by deanjo
I'll look into it. Maybe I misconfigured something.
One big question I don't see anyone asking:
Does the X-Fi even do hardware mixing under Linux? I had heard that it didn't, and hadn't heard otherwise from anyone else.
You probably dont understand what is hardware mixer. Its not responsible for volume control, or sth like this. Its mixing sound channels(not 5.1 to stereo, but from different applications). And its not only simple mixing i.e. 5 applications(like system sounds, winamp, game, communicator and sth else), but threre are specialized OpenAL implementations that use hardware mixer(or some emu10k1 sound processor) that mix in hardware many voices/sound/music channels in realtime(damny usable in fps games) saving alot of CPU power(including energy. If you are not running games, remember that most softfare mixer implementations(OSS/ALSA) are using fixed number of channels(channel != 5.1, stereo, in simplicity its number of applications using soundcard), and because its fixed number(in OSS you can change it manually, but theres no dynamic automats), it must big enough (i.e. 32 multiplied by number of speakers, so it will result in 64 channels, for 5.1 it will be 192 mixed in realtime !), otherwise next opened applications will fail to open soundcard. And remember about possibility of logging of multiple user in one time.
Originally Posted by duby229
Its definitely better to have hardware mixer, because it saves CPU cycles, power(in laptops battery), it prevents from mixing buffer overrun/latencies in games under heavy CPU usage. And if hardware mixer doesnt have enough channels, we can always mix in software and use only one virtual device(or partially use software mixing in some implementations like ALSA that allow us to create virtual devices and configure them, like only we want). Sot theres no any advantage not having hardware mixer, because we always can not use them. And I dont undertand anything about "upgrading mixer, even on old hardware", because mixer has only one specified task, and you cant do it better, than old sound blasters(emu10k1) to audigy4(emu10k2) doest it.
mixing doesent really take much cpu time.. its more resampling that does if it has to be super high quality.. and well.. most people wont care, and why should they? its not as if you need super high quality of TWO sound tracks playing same time.. or super high quality (maybe) resampling of some annoying IM buzz sound..
all in all, hardware mixing isnt really much help today, except for really low latency requirements, which isnt what gamers or audiophiles are gonna be demanding. everyone has plenty of cpu power either way, hell, most people could even dedicate an entire cpu core to the thing, if they need really low latency resampling and remixing.. fortunately this isnt necessary at all.
what hardware mixing has as a potential disadvantage though, is that you have no idea if they are still running it through some resampler or weird stuff in hardware, or if the dsp alters the bitstream, potentially degrading performance.
No, using opensource driver on linux, we definitely have idea what it does with stream, and it doesnt affect sound at all, and most integrated cards today, accept stream mostly in one format, so its resampled in software. Even worse on Vista binary drivers affects very much sound, to be worser to avoid recording (fantastic DRM).
Maybe today CPUS are powerfull enough, but all arguments about disadvantages of hardware mixing are a bullshit (epsecially that only better cards has hwmixer, avoiding of hwmixer is cost-cutting, so cards w/o hwmixer are usually lowe-end). And price of old(but damny good) sblive with hwmixer is not so high.
Actually the on the creative emu10k series the DSP resamples everything to 16/48. So if your playing say a 24/96 stream it gets thrown into a funnel. 24/96 -->resampled in DSP to 16/48--->upsampled with padding to whatever output set at. Even native 16/48k signals get resampled and because of the src bug found in the emu10k all the audio that goes through the dsp contains a partially wrong 16th bit.
Originally Posted by evil_core