Win2k3 server

Toggle sidebar Toggle sidebar
Z

zerocool1

Diamond Member
Jun 7, 2002
4,486
1
81
femaven.blogspot.com
I posted this in TS yesterday afternoon and haven't gotten much response as expected.

Ok, here are my system specs.

a8nsli on 1008 bios
venice 3000+
1 gig dual channel geil memory DDR400
Leadtek TV tuner card
ATI/Sapphire x800XL
Ultra PSU(the modular ones)

I'm running Server 03 on this box. Not sure if that'd make a difference.


I keep getting a BSOD, and the error given is machine_check_exception and it doesn't give me any more information. Before it would just give me an error when watching a DVD over Daemon tools but it errors on just about everything now.

I'm not overclocking at all. I've checked if it were something I did with my memory timings. umm...
Thanks in advance.

Amit

EDIT: Using SATA Raid
Some error info:

machine_check_exception Stop: 0x0000009C
just did memtest for 13+ hours. No errors at all. I'm very confused/
 
V

Vegito

Diamond Member
Oct 16, 1999
8,329
0
0
could be driver issue... just fyi let us know what the event says.. im on the a8n sli and ran 2003 server till yesterday.. just upgraded to server 2003 64bit. So i might be able to try to help..
 
Z

zerocool1

Diamond Member
Jun 7, 2002
4,486
1
81
femaven.blogspot.com
I called AMD and they said that it could be a memory issue because Win2k3 server can be picky with memory. I can't install or uninstall anything because it blue screens then.
 
D

drag

Elite Member
Jul 4, 2002
8,708
0
0
You don't want to memory test from inside the OS anyways. There are to many limitations..

Win2003 is sensitive my ass. AMD 64 proccessors are the sensitive part, with their built-in memory controller. Some motherboard and memory combinations are simply unusable sometimes.

Check out memtest86 or memtest86+ (not sure which is better, both are fairly up to date). You can burn it to a cdrom or they have floppy images aviable. Boot up with that, run the test overnight and you can pretty easily determine memory problems.

If you have constantly the same addresses screwing up then a memory module is bad. If you have memory problems, but only in a certain range, but only every once in a while (sometimes it pasts the tests) it's almost certainly bad ram module. Figure out which module or if all are bad, return them to manufacturer and get a replacement.

If it's a large amount of problems over a large range of memory, then it's probably motherboard/controller problem working with the RAM. Use conservitive timings, underclock to 333mhz. Reseat the modules, make sure that there isn't much dust. Wiggle them around, move them to different slots. Certain setups will only work with memory at 400mhz in slots 1 and 2, only 333mhz will work in slot 3. The ultimate fix then is to buy memory that has been tested and certified by the motherboard manufacturer to work with with your model and cpu.
 
Z

zerocool1

Diamond Member
Jun 7, 2002
4,486
1
81
femaven.blogspot.com
Originally posted by: drag
You don't want to memory test from inside the OS anyways. There are to many limitations..

Win2003 is sensitive my ass. AMD 64 proccessors are the sensitive part, with their built-in memory controller. Some motherboard and memory combinations are simply unusable sometimes.

Check out memtest86 or memtest86+ (not sure which is better, both are fairly up to date). You can burn it to a cdrom or they have floppy images aviable. Boot up with that, run the test overnight and you can pretty easily determine memory problems.

If you have constantly the same addresses screwing up then a memory module is bad. If you have memory problems, but only in a certain range, but only every once in a while (sometimes it pasts the tests) it's almost certainly bad ram module. Figure out which module or if all are bad, return them to manufacturer and get a replacement.

If it's a large amount of problems over a large range of memory, then it's probably motherboard/controller problem working with the RAM. Use conservitive timings, underclock to 333mhz. Reseat the modules, make sure that there isn't much dust. Wiggle them around, move them to different slots. Certain setups will only work with memory at 400mhz in slots 1 and 2, only 333mhz will work in slot 3. The ultimate fix then is to buy memory that has been tested and certified by the motherboard manufacturer to work with with your model and cpu.
Click to expand...

I was running memtest on boot for like 1.5 hours but no errors. I've reseated the ram. I'm going to double check the timings, lower them and see if that's the problem. Its set up in dual channel. I will also check out if they are in the right slots like you'd mentioned.
This is my my memory
 
D

drag

Elite Member
Jul 4, 2002
8,708
0
0
heh, I just bought the same memory myself for a machine that I am building. Nice sale on newegg. ;) Now I hope it's not the ram!

To get a good idea you realy need to let it run for quite a long time. For memory to be healthy you can expect it to run weeks with no errors. For bad ram it may take a few hours for it to show the problem. 8 hours is about the right amount of time, I figure, for a good test.


edit: a nice thing about having a seperate bootup into the cdrom or floppy vs running memory tests in-OS is that there are always going to be certain parts of the RAM that are reserved... like were the kernel resides, which is the most likely place for bad ram to cause BSODs and crashes.
 
M

Mark R

Diamond Member
Oct 9, 1999
8,513
16
81
Are the PSU voltages OK?

One thing I don't like about 'modular' PSUs, is the extra connections - connections are the weak points of PSUs, and add another place that can cause bad connections and voltage sags. Ideally, check the voltages in the BIOS and also with a voltmeter (probes inserted into the big Molex connector).

Note - that if there is a bad connection, then the voltmeter may give an unreliable reading - it will read the voltage produced by the PSU, but not the voltage that the motherboard is actually getting.

I also note that memtest has gone 13 hours without an error - this generally means that your memory is fine. However, I had several faulty systems (Nforce 2 motherboards and incompatible RAM) that would crash all the time - but, memtest would only very occasionally find errors (sometimes a 1 hour run would find hundreds of errors, most of the time 12 hours would find nothing) - very odd. It's a bit of a long shot, but you could try MS's Windows Memory Diagnostic (WMD). When I tried it, it would always find thousands of errors on the above systems, within 1 hour.

I've no idea what WMD does differently, but I tested this over and over on several machines.

You may also want to get hold of The ultimate boot CD and try the CPU burn in tests - this will test your system, without loading windows. If your system works perfectly in linux (which these tests use) then you have to consider the possibility that this is a bad driver (or malfunctioning graphics card) - although Machine_check_exception usually means a CPU/RAM prob.

 
Z

zerocool1

Diamond Member
Jun 7, 2002
4,486
1
81
femaven.blogspot.com
Originally posted by: Mark R
Are the PSU voltages OK?
....

I also note that memtest has gone 13 hours without an error - this generally means that your memory is fine. However, I had several faulty systems (Nforce 2 motherboards and incompatible RAM) that would crash all the time - but, memtest would only very occasionally find errors (sometimes a 1 hour run would find hundreds of errors, most of the time 12 hours would find nothing) - very odd. It's a bit of a long shot, but you could try MS's Windows Memory Diagnostic (WMD). When I tried it, it would always find thousands of errors on the above systems, within 1 hour.

I've no idea what WMD does differently, but I tested this over and over on several machines.

You may also want to get hold of The ultimate boot CD and try the CPU burn in tests - this will test your system, without loading windows. If your system works perfectly in linux (which these tests use) then you have to consider the possibility that this is a bad driver (or malfunctioning graphics card) - although Machine_check_exception usually means a CPU/RAM prob.
Click to expand...
yea the psu voltages are alright according to the motherboard. how long do I need to run the cpu burn in test?
 
C

Chris S

Senior member
Feb 8, 2000
447
0
0
I had the same code on a similar machine(NF3 board), tried everything...PS, video card, reinstalled several times. I had my memory in slots 1&2, moved to 1&3...It's been up for 4 days, I was lucky to get 20 min before.
 
Z

zerocool1

Diamond Member
Jun 7, 2002
4,486
1
81
femaven.blogspot.com
Originally posted by: Chris S
I had the same code on a similar machine(NF3 board), tried everything...PS, video card, reinstalled several times. I had my memory in slots 1&2, moved to 1&3...It's been up for 4 days, I was lucky to get 20 min before.
Click to expand...

I haven't tried that yet. let me know if it stays that way.
 
N

NogginBoink

Diamond Member
Feb 17, 2002
5,322
0
0
Replace the CPU already!

If you get the details of the machine check exception from the event logs, a high enough engineer at AMD can tell you what they mean.

But ultimately, for you, it probably means "replace the CPU"
 
Z

zerocool1

Diamond Member
Jun 7, 2002
4,486
1
81
femaven.blogspot.com
Originally posted by: NogginBoink
Replace the CPU already!

If you get the details of the machine check exception from the event logs, a high enough engineer at AMD can tell you what they mean.

But ultimately, for you, it probably means "replace the CPU"
Click to expand...

I'm gonna do that tomorrow.
 
S

Smilin

Diamond Member
Mar 4, 2002
7,357
0
0
Originally posted by: NogginBoink
Replace the CPU already!

If you get the details of the machine check exception from the event logs, a high enough engineer at AMD can tell you what they mean.

But ultimately, for you, it probably means "replace the CPU"
Click to expand...

:thumbsup:
 
You must log in or register to reply here.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom