Memory and data corruption.

voltaic

Distinguished
Oct 7, 2002
5
0
18,510
I'm running XMS3200C2 memory (2-3-3-T1) on a SF51 Shuttle mobo. The drive is a WD1200JB which is the special edition one.

My initial problem was that Windows XP kept writing corrupt data on the drive, and at every restart Checkdisk would run and complain about a bunch of things. After some troubleshooting I found out that lowering the memory's frenquency and timings apparently fixes the problem. I'm currently running it at 133mhz with CAS 2.0 and T1. The latencies are set to "Normal" in the bios (there are 6 options Normal being the second lowest one).

The memory I'm sure can run a lot faster, at least it is supposed to. So either the RAM is bad or there is something else that I suspect may be causing this.

I built this system recently and I still don't have a video card (still deciding), so I'm just using the onboard video which is Integrated SiS651 VGA. So I'm sharing the system memory with the integrated video. I already found out that this is degrading the memory's performance a lot (Sandra reports some low numbers), but I'm not sure if this would affect the memory and things like data integrity. Does anyone know if sharing stresses the memory a lot? I'm not sure what's causing data corruption but this looks like a possiblity.

If anyone can help or have a suggestion I'll appreciate it.
 

slvr_phoenix

Splendid
Dec 31, 2007
6,223
1
25,780
I'm not sure about the Integrated SiS651 VGA causing memory corruption problems or not. I mean it is possible, but how to know? I know Matrox had a problem with a shared memory AGP bug that caused memory corruptions that locked up a PC on a Via chipset in their Win2K drivers, so it is entirely possible that SiS has a bug too.

Shared memory between AGP and the CPU does hinder your CPU's available memory bandwidth badly, which kills performance. (That's why most people call onboard 3D accelerators 'decelerators', because sharing the bandwidth like that just to provide for the graphics 'accelerator' chokes the CPU.) To my knowledge though it has been pretty rare that this actually causes memory corruption problems.

Then again, SiS isn't exactly a what I'd call a quality motherboard vendor in the first place. I still put them back with ALi and Via, so who knows what they're capable of screwing up.

The problem really is that the possabilities of the cause are nearly endless. It could be bad firmware/drivers. It could be a fault in the southbridge. It could be a bad stick of RAM. It could be a power supply problem. It could just be that the motherboard wasn't manufactured to spec. **big shrug**

The only real way to diagnose the problem is to try other parts (and newer drivers) and see what replacement fixes the problem. That, or just live with the memory timings that work.

I do had to admit though, if lowering the memory timings fixes the problem, it really does sound to me like a bad stick of RAM.

<pre><A HREF="http://www.nuklearpower.com/comic/186.htm" target="_new"><font color=red>It's all relative...</font color=red></A></pre><p>
 

voltaic

Distinguished
Oct 7, 2002
5
0
18,510
Thanks slvr_phoenix

I'll see if a video card fixes the problem. Otherwise it seems like it's very hard to tell what is causing the problem. If that doesn't work I think I'll see if I can replace the memory.
 

lhgpoobaa

Illustrious
Dec 31, 2007
14,462
1
40,780
1. Video card.
2. Latest video card drivers.
3. Latest chipset drivers.
4. Run MEMTEST86 to evaluate your ram's stability.

<b> :cool: I :smile: think :frown: I'm :redface: addicted :wink: to :eek: these :mad: ever :eek: so :tongue: cute :cool: smiley :lol: faces! :wink: </b>
 

voltaic

Distinguished
Oct 7, 2002
5
0
18,510
LHGPooBaa,

I did try to run memtest86 couple of weeks ago but it counted a lot of errors. It didn't fail though it ran for 11 hours, then I shut it down.

I searched for memtest86 articles but I couldn't really find anything on what to expect, and how to interpret the results. I mean I was getting literally thousands of errors but it just kept running. Maybe that shows that the data is corrupt or something but I'm not sure. Then after I noticed those corruption problems I had to lower everything from frequencies to latencies.

I have not searched this forum for memtest related issues though, I'll do that now.

If you have any additional suggestions or comments please post them.

Thanks for all the help.
 

EvilHammie

Distinguished
Jul 29, 2002
128
0
18,680
ah ha!
errors = bad
thats the good thing about memtest, it keeps going even when it finds errors.

Usually when you get errors that means 1 of 2 things.
A. That your memory stick is just bad
<b>OR[/b}
B. The memory speeds, & timings are too harsh for your memory.

Example: I have nice PC3200 ram running at 166mhz with hardest memory timings. When i first ran at such speeds and timings memtest gave me a few errors. From there i could do 2 things, run at more conservative timings or increase the ram voltage. As i didnt want to sacrifice performance i did the latter. after an extra +0.1v it runs perfectly.

so if you can, bump up the ram voltage by +0.1 or +0.2, run memtest and see if there is an improvment. If not, you will have to run at more relaxed timings (or get better ram).

If errors persist no matter what you do, then its probably a bad stick.

<b>Small.Cute.Fluffy.Evil. But looks so cute attached to your leg smiling up at you! :smile: </b>
 

voltaic

Distinguished
Oct 7, 2002
5
0
18,510
Thanks EvilHammie,

I'll try that as soon as I have time :)

The last day I was testing some higher frenquencies to see where it ran stable, unfortunately it decided to destroy a lot of files so I don't have an OS now. My OS partition is pretty much gone, but I don't keep data in there anyways, and I have backups. So if I have some time I'll have to reinstall the OS and waste couple of hours that way. But I think I have a bootable Memetest86 CD so I guess I won't have to install the OS..

In another forum someone was suggesting that this may have been caused by a bad CPU, is that credible or can any component be a possible suspect?

Oh by the way this machine is a little mini tower thing (Shuttle SS51) so they decided that modifying the voltages was too dangerous for some people. I won't be able to play with voltages in any way, I can't change the core either (unless I do it physically) so I'm stuck with the defaults. I bought this RAM for future use but it seems like I won't be able to use it in this box.

Again thanks for all the info and if you can help I'll appreciate it.
 

lhgpoobaa

Illustrious
Dec 31, 2007
14,462
1
40,780
Such a helpful little hamster aint he? :smile:

I find its unusual that such high speed ram is unstable in your system.

I too am using PC3200 Cas2.5 ran and it runs very well up to 166mhz on absolute hardest timings. It is possible the board and ram are incompatible.

Bad cpu's are actually extreemly rare. 99% of problems are cauzed elsewhere. e.g. insufficient cooling, poor PSU, bad ram, bad mobo. etc


<b>Plaid will NEVER go out of style! :cool: </b>
 

voltaic

Distinguished
Oct 7, 2002
5
0
18,510
Then maybe it's the mobo after all. It doesn't "officially" support DDR333 but so many people are running high speed memory on the board I find it surprising that it's giving me so much trouble.

I'll do some testing, if nothing works I'll start to RMA the components, mobo being the first one to go.