|
Post by Art Hunkins on Jan 18, 2005 13:58:18 GMT
Michael,
Do you have any advice to offer as to how to avoid audio breakup with ASIO4ALL and older, less powerful computers (Pentium 1 vintage)?
As explained in a question under Problems & Solutions, doing all to maximize quality (latency is not an issue) still gives broken audio on these machines. Any chance of increasing the number or size of buffers (or whatever) in v2.3 - or would that not help?
Thanks much -
Art Hunkins abhunkin@uncg.edu
|
|
|
Post by Michael Tippach on Jan 19, 2005 0:21:59 GMT
The minimum configuration that I ever tested ASIO4ALL on is a PII 350Mhz with an on board ESS Solo sound chip under WinMe.
To me, this appeared like this was something close to the very minimum system spec to run ASIO4ALL reasonably on.
I do not think that further increasing the buffer size range beyond 2048 samples would improve things a lot. 2048 samples is nearly 50ms for stuff to sort itself out, which is like an eternity - even in Pentium I computing terms.
Chances are that you may get better results by avoiding things that would increasease the CPU load:
- If you do not need audio input, disable all audio inputs! ASIO4ALL will no longer choke on input/output syncronisation issues.
- Most of the time, using 48Khz sampling rate will lower CPU usage as opposed to 44.1Khz. The reason being that AC97 onboard stuff only natively runs at 48KHz and everything else is being resampled. If ASIO4ALL does not resample, the WDM driver will, both of which costs CPU power. This is counter-intuitive as you would expect CPU usage to be lower when only 44.1k vs. 48k samples need to be calculated per second.
|
|
|
Post by Art Hunkins on Jan 19, 2005 16:44:12 GMT
Thanks for the info, Michael.
My "older computer" is an AMD K-6 under WinME, with 288MB RAM. The soundcards I'm dealing with, besides VIA AC'97, are a Crystal (WDM) and an Avance ALS4000 (WDM). (Do they all work the same with regard to 48K and 44.1K sample rates?) I've already turned off the audio inputs in ASIO4ALL.
I tried going to 48000 SR instead of 44100, and I can tell no discernible difference. (In some cases of high CPU demand, my audio file even freezes up and refuses to render.)
Using MME drivers and internal settings available in Csound, I've been able to get fairly unbroken playback with very long latency (> 1 second). As I mentioned, this works fine for me. Unfortunately, with Csound5 coming out, ASIO is required.
Do you have any experimental versions of ASIO4ALL available that I might try, or should I simply give up on ASIO with such older computers? I'd be happy to *try* anything.
TIA,
Art Hunkins abhunkin@uncg.edu
|
|
rol1
New Member
Posts: 1
|
Post by rol1 on Mar 18, 2005 18:29:07 GMT
Litepc.com has a program that lets you uninstall all kinds of windows stuff that slows your machine down.
|
|