Partial explanation to 3538m driver problems and a way to solution

Estarriol · Feb 14, 2007, 12:04 PM

This text relates only to Audigy 2 ZS notebook and its audio input problems

Well, guys, the problems with 3538m driver are slowly moving to resolution. Here's a detailed example how I made my Audigy2ZS notebook constantly work.

Yesterday I restored an image of a WinXP system (I had Windows for Legacy PCs installed before). This WinXP had kXdrv installed (I dont remember the exact version). So I just replaced the

Code:

C:\%WINDOWS%\System32\drivers\kx.sys

driver with original 3538m.

A2ZSnb works fine when plugged in a already loaded system. Mic is noisy and reacts to almost anything, including knocking a card with a fingernail. But I still could use a mic well, and did a funny Skype call using DSP processing of audio-in stream.

I switched the computer off per night, and turned it on at morning with a card still in. The system halted on boot, and after its normal boot-up without a card in I tried to plug A2ZSnb - the system halted constantly.

After a small thinking I concluded: all the problems of inconsistency of my prior tests come from caching. And this caching in WinXP is called PREFETCH.

First thing you have to do is to delete all the files from

Code:

C:\%WINDOWS%\Prefetch

, and turn off ALL the prefetching by setting HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Contro l\Session Manager\Memory Management\PrefetchParameters\EnablePrefetcher to zero.

The option by default, "3", means prefetching of boot files and applications. I had no intent to dig in deeper, so I just set it to "0" = no prefetch at all.

WinXP loaded fine, I plugged the card in... and it started working!

The underground of the problem is the following: 3538m original driver has a bug, leading to system halt during system load. This halt somehow corrupts prefetch file corresponding to A2ZSnb. Is it this .pf incompleteness or some incorrect bytes order, it does not matter.

After we had corrupted this .pf file, the next load without a A2ZSnb goes fine coz' WinXP does not try to load this .pf. But when A2ZSnb is plugged in, this prefetch file tries to load, halting the system again.

I dunno what these .pf contents is, but here are the experimental facts. So the solution to fix the driver is to fix a boot-bug. The card works fine when plugged in after boot-up.

thomasabarnes · Feb 14, 2007, 01:33 PM

OK, How about making an Audigy 2 ZS Platinum Pro work?

Estarriol · Feb 14, 2007, 01:47 PM

This topic applies only to Audigy 2 ZS notebook

silvertones · Feb 14, 2007, 03:25 PM

OK what abt win2Ksp4. I guess the real answer is to fix the bug ib 3538m?

thomasabarnes · Feb 14, 2007, 09:35 PM

Quote:

Originally Posted by Estarriol

This topic applies only to Audigy 2 ZS notebook

LMAO

Man, I just thought if you could solve that problem with the notebook, maybe you could solve the problem with the Audigy 2 ZS Platinum Pro input noise.

Anyway, was just kidding around with you.

Congratulations on your success.

Max M. · Feb 14, 2007, 10:57 PM

Leoromanelli · Feb 19, 2007, 10:37 PM

has anyone knows how to make the beeping dissapear of the audigy 2 zs notebook when i play a song in winamp? I've noticed that if you press play several times the music appear, and when you press again it dissapears, and you have a beeping instead of music. I've also noticed that some color in the graphic display of kx moves everytime y press play.
I'm a newbie on kx, so excuse my "indian like" description.

**Eugene Gavrilov** · Feb 24, 2007, 03:20 PM

leoromanelli: currently, there's no solution
you need to do one of the following things:
1. unload kxmixer; go to devman; disable the driver; re-enable the driver; re-start kxmixer (with --startup option)
OR:
2. close any audio applications; in kx console type rv ('reset voices')
OR:
3. playback any MIDI file (open kX Mixer's analyzer and watch until all hardware voices are allocated at least once)
OR:
4. keep pressing pause/play/pause/play in Winamp (watch kXMixer's analyzer - voices should shift to the right) -- after ~16-32 attempts this usually fixes the issue

E.