A7N8X Deluxe -- False Memory Failure Reports?

[jim evans]

I initially configured with a 512 stick of Kingston memory. As soon
as I fired up my new A7N8X Deluxe motherboard the woman's voice
started reporting memory had failed the self-test. She gives this
failure report only once about every 5-10 restarts.

I have seen no behavior to suggest the memory is bad, so I ran
MemTest-86 version 3 off-and-on for about 25 hours. It reported no
errors. I bought a second 512 Kingston stick and replaced the first
one. The woman continues to report bad memory the same way. Again,
about 10 continuous hours of MemTest reported no errors.

Does this motherboard give false memory failure reports?

jim

 

[alan buckley]

I was going to psot the same message today, mine does the identical problem
as you describe

 

[Tocapet]

I was getting that message too until I reset it back to AUTO in bios. I had
it set at 200 and 120% and would get that message. So far no more msg after
putting it back on the default.

Why don't you try those 2 sticks in dual-channel mode
put 1 in slot 1 and the other in slot 3

You can run Winbond voice editor and delete that message if you're sure
everything else is OK. You can find it on your CD that came with the
motherboard. You find the msg, delete it, then burn it to the chip.

My memory is Samsung PC3200 DDR running in dual-channel mode.

Tocapet

 

[alan buckley]

I had mine in auto and it reported failed mem test, and now in user select
it still does the same, just every now and then

 

[Pete Zafian]

Jim: Just turn off the the voice. I found it rather annoying and it
did indeed give some erroneous reports once
in a while. Besides, you shouldn't always believe what a woman tells
you anyway...(oh, here comes the flood).

Pete

 

[Dale Cohen]

I have exactly the same problem. 512Mb Kingston PC3200 memory and the voice
reporter would give me an occassional memory error at startup. I switched on
the full memory test at POST in the BIOS and have never had an error
reported during that test. The machine is stable (enough) in day to day
operation to rule out a bad stick so I turned the voice reporter off. I'd
love to know if there's a bios revision planned to fix this!

 

[Jim]

Its been happening for a long time now. My rev 1.06 has the same problem.
To me it seemed to happen when I changed the configuration in some way that
effected the boot timing. Now I just use STR and never have to reboot.

 

[Paul]

Its been happening for a long time now. My rev 1.06 has the same problem.
To me it seemed to happen when I changed the configuration in some way that
effected the boot timing. Now I just use STR and never have to reboot.

That is an astute observation and points to the source of the problem.

The Voice POST Reporter is an autonomous subsystem. It doesn't need
a processor or memory to run. As long as it has voltage to run on,
and an event to kick it off, it is happy.

The Reporter consists of two chips. One is the Winbond monitoring chip
and the other chip is a socketed 8 pin DIP EEPROM. The EEPROM is what
gets programmed by the Voice Editor software.

Inside the EEPROM are stored two things (that I know of). One is
compressed voice samples (i.e. the quality is variable, so you can
have long, crappy quality messages or short, higher quality messages).
The other thing stored in there is the script for the monitoring chip.
This is a program the chip executes, and it links which voice sample
to play, in response to some external stimulus or error condition.

The monitoring chip has a timer in it, and for some of the error messages,
the message is triggered if the timer runs down, before the CPU gets
to the chip and clears the timer. A lack of response from the CPU would
be a good reason for issuing a "No CPU present" voice message, for example.

So, if you install that Voice Editor software and can find that script
file I looked at a while back, the answer to your problem might lie in
there. I don't know if the script file is editable by the Voice Editor
software, or even if it can be extracted from the EEPROM, but that is
where I would start, to fix it. Presumably, simply increasing a time
constant would fix it.

About the only thing I don't understand, is how it is possible for
the BIOS to enable or disable this feature. I mean, if the CPU is
dead, how can the BIOS "gate off" the Voice POST? Maybe the BIOS
actually writes something into the Voice EEPROM when the setting is
changed, but somehow I doubt it.

HTH,
Paul

 

[Gino Zantafio]

It appears you have a good knowledge of the voice reporter mechanisms.
May be you'll be able to give help.

Winbond voice editor doesn't work on my system (A7N8X Dlx/Athlon 2400+) and
98SE albeit installed in two partitions. It sends the following message when
I attempt to run it: "Load Driver failed!! ".
I don't use Asus Probe and MBM5 wasn't yet installed when I got the first
message.
Any idea ? Thanks

 

[Paul]

It appears you have a good knowledge of the voice reporter mechanisms.
May be you'll be able to give help.

Winbond voice editor doesn't work on my system (A7N8X Dlx/Athlon 2400+) and
98SE albeit installed in two partitions. It sends the following message when
I attempt to run it: "Load Driver failed!! ".
I don't use Asus Probe and MBM5 wasn't yet installed when I got the first
message.
Any idea ? Thanks

I have two versions on disk here. They are:

http://www.asus.com/pub/ASUS/misc/utils/Voice_10.zip
ftp://ftp.asus.com.tw/pub/ASUS/misc/utils/winbond/865875_veditor_M316.zip

After installing both versions, and using a hex editor, I can see
the main executable "speecher" contains the "Load Driver failed" message.
When the uninstaller runs, there are two files that were installed in the
system32 directory. They are WBHWDOCT.sys and Wbhwdoct.vxd. The .vxd
has a text section listing the chipsets supported, and the M316 version
has support for the A7N8X and P4C/P4P800 boards. Here is the text section
from M316's .vxd

*********** Begin Sludge Version 316 ***********
write sucess
CX=%X
CAH=%x offset=>%x data => %x
VendorID=%x DeviceID=%x CAH=>%x
get smbus address Address =%8x FunNo=%4x ,DevNo=%4x, BusNo=%4x
DeviceVendor=%8x
Error:getSMBusBaseAddr() fail: %d
Check nVIDIA MCP2 Result = %d
Check nVIDIA MCP2 Chipsets
Check ALI_1535D Result = %d
Check ALI_1535D ChipSets
Check AMD_756 Result = %d
Check AMD_756 ChipSets
Check SIS_961a2 Result = %d
Check SIS_961a2 Chipsets
Check SIS_961 Result = %d
Check SIS_961 Chipsets
Check SIS_730 Result = %d
Check SIS_730 Chipsets
Check SIS_630 Result = %d
Check SIS_630 Chipsets
Check VIA_VT8233A Result = %d
Check VIA_VT8233A Chipsets
Check VIA_VT8233 Result = %d
Check VIA_VT8233 Chipsets
Check VIA_VT8235 Result = %d
Check VIA_VT8235 Chipsets
Check VIA_VT82C596 A B Result = %d
Check VIA_VT82C596 A B Chipsets
Check VIA_VT82C596B Old Result = %d
Check VIA_VT82C596B Old Chipsets
Check VIA_VT82C686A Result = %d
Check VIA_VT82C686A Chipsets
Check Intel_ICH3 Chipsets
Check Intel_ICH3 Result = %d address = %x
this is test for lyc
Check Intel_ICH2 Chipsets
Check Intel_ICH Result = %d
Check Intel_ICH Chipsets
Check Intel_ICH5 Chipsets
Check Intel_ICH4 Result = %d
Check Intel_ICH4 Chipsets
Check Intel_ICH0 Result = %d
Check Intel_ICH0 Chipsets
Check Severworks CSB5 = %d
Check Severworks CSB5 Chipsets
Check Severworks ROSB4 Result = %d
Check Severworks ROSB4 Chipsets
Check Intel_440MX Result = %d
Check Intel_440MX Chipsets
Check Intel_PIIX4 Result = %d
Check Intel_PIIX4 Chipsets
<bad format character><float format not supported>0X0x0 +
*********** End Sludge ***********

There are two possibilities. Either the SMBUS path to the Winbond
chip is more twisted than the driver can understand, or you need to
run the Voice Editor with Administrator permissions. Since both
versions of the editor have some strange .flh files, it looks like
Winbond at least considered different localizations, so
localization probably isn't the problem. Make sure you are using the
M316 version, as that knows what an MCP is - uninstall whatever other
version you are running first.

HTH,
Paul

 

[Vince]

I have exactly the same problem. 512Mb Kingston PC3200 memory and the voice
reporter would give me an occassional memory error at startup. I switched on
the full memory test at POST in the BIOS and have never had an error
reported during that test. The machine is stable (enough) in day to day
operation to rule out a bad stick so I turned the voice reporter off. I'd
love to know if there's a bios revision planned to fix this!

Exact same experience here with OCZ PC3200 512MB sticks(2 sticks).
Every 3rd boot or so, the voice would say memory test failed and then
it would bootup XP anyways. I've had a stable system right from Day 1
so there is obviously something wrong with the POST memory test OR
whatever it is that is causing the failure is insignificant enough to
ignore.

Vince

 

[Gino Zantafio]

You were right ! My old drivers weren't addressing the Nvidia chipset.
I followed your recommendation by removing the old program and installing
the "865875_veditor_M316" version.
Nothing landed in Windoxs\system32 directory. WBHWDOCT.sys was copied to
windows\system\drivers and Wbhwdoct.vxd to windows\system.

And the error message remains "load driver failed!".

By the way, what OS did you use when you tried the installation ?
What did you mean by "to run the Voice Editor with Administrator
permissions" ?
Win NT or XP ?

I suspect something else is missing in Win98SE.
On another hand, in Win98SE, the installation and uninstallation procedures
don't give any clue on what was put or removed.
So I can only suspect !

 

[Paul]

You were right ! My old drivers weren't addressing the Nvidia chipset.
I followed your recommendation by removing the old program and installing
the "865875_veditor_M316" version.
Nothing landed in Windoxs\system32 directory. WBHWDOCT.sys was copied to
windows\system\drivers and Wbhwdoct.vxd to windows\system.

And the error message remains "load driver failed!".

By the way, what OS did you use when you tried the installation ?
What did you mean by "to run the Voice Editor with Administrator
permissions" ?
Win NT or XP ?

I suspect something else is missing in Win98SE.
On another hand, in Win98SE, the installation and uninstallation procedures
don't give any clue on what was put or removed.
So I can only suspect !

I was using WinXP Pro when I did the test install. I did it just so I
could see what files were in the installed directory or elsewhere.
The problem with Installshield, is I don't have access to whatever
other files might be installed for a different OS, so it is hard
to guess what differences there might be. (Since the computer I tried
this on doesn't have Voice POST, it wasn't possible to get
the program to actually run.)

I've had the Voice Editor installed and running on my P4B motherboard.
And that machine was running Win2K. My Win98 machine doesn't have a
Voice POST on it, so isn't a good candidate.

It would be interesting to see what is in the temporary directory
that the Installshield creates, but I haven't had any luck getting
to it. I would have thought the driver system was different enough
between Win98SE and WinXP, that different driver files would
be needed.

Good luck getting it to work
Paul

 

[Gino Zantafio]

Thanks a lot for your answer.
It's quite a confirmation.
I feel I've to upgrade Windows
I need to !
And I'll take benefit of the upgrade to install the Voice editor again.

Have a merry Christmas