Motherboard giving Memtest errors

[DanGent]

Hi there,
Got a laptop here thats been dropped round mine for being "poorly" ;^) So thought I'd do the diagnosis.
Memtest86 gives loads of errors on the 2 x 512MB sticks that are in. Taking out either of these sticks and swapping slots and etc gives loads of errors (almost instantly). Got 3 256MB sticks of RAM (old ones from when I've upgraded laptops), and when I stick any of these in in any slot (done many tests) get lots of errors.
I haven't tested any of these sticks in another laptop, I've got a DDR2 laptop in pieces that I'm going to setup later today to verify all these sticks are OK, but I'm quite certain I'd tested these 256MB sticks already before I took them out the previous laptop.

Had a quick google for my problem but all I could find was "begginer to Memtest" kind of posts. Any ideas on whats wrong? On Speedfan the laptop didn't look very hot, almost too cool really! Left it running Prime95 for about 8 hours and it barely got above 40c! (very odd for a laptop, its an AMD dual core so it should be hiting much hotter than that). And the laptop did feel warm.

Is the mobo just screwed or is there anything else I should be looking at?
What kind of performance problem will I notice from many, many Memtest errors. It still seemed to run Prime95 OK.

Much thanks,
Dan Gent

[DanGent]

To clarify this post. the laptop originally had 2 x 512MB sticks in. When running memtest I got multiple errors, only then did i take those sticks out and add others.

To add to this post, I have put one of the original 512MB sticks in another laptop and its memtested for 20 hours with no errors. I am now doing this with the other 512 MB stick. It really does look like a motherboard problem!

[Hawker1986]

Agreed, looks like a dead motherboard, new laptop time unless its under warranty still?

Hawker

[DanGent]

cheers hawker for confirming my suspicion
Yeh its well past warranty. For a laptop this old not sure if its even worth getting a new motherboard. I'll see how much they are and what type of warranty they come with.

What exactly will happen with this laptop when its being used? I ran Prime95 on it and Speedfan to see if it was a heat issue and they both ran fine for hours. Previous laptops I've had with RAM errors (not mobo) have done odd things like freezing when you put the mouse to the bottom right of the screen. really wierd stuff. Will this laptop just be very unstable?

Much thanks,
Dan Gent

[Zak33]

if it were a normal PC motherboard I'd ask you to take off the heatsink from the chipset and reapply some new thermal compound.... but I don't even know where to start with a laptop motherboard.

[DanGent]

Zak33 - do you think it might be a heat problem? Although the laptop felt quite warm around the keyboard. Speedfan (temperature monitoring software) shows the laptop as being very cool. Maybe a little too cool I was thinking!

I disassemble laptops all the time. Its not hassle to reapply thermal paste. Have you experience of this kind of thing working before? I suppose its got to be worth a try!

[scaryjim]

If it's a recent AMD (i.e. Athlon 64 based) it'll have a built in memory controller, so it could CPU on the way out rather than Motherboard. Did you run CPU-Z at the same time as Prime and Speedfan to check what clock speed the CPU was running at? If it had a problem it might potentially run permanently at low power states which would account for the temps remaining very low...?

[DanGent]

scaryjim - I ran CPU-Z but didnt even check what the clock speed was. Thats a very good idea. I never thought that one of the advantages with having the memory controller on the CPU chip was that you could replace it.
I'll find out the details and post back

Much much thanks,
Dan Gent

[DanGent]

scaryjim - here are some of the details from CPU-Z: please let me know if you want me to get more.
AMD Turion X2 64bit Mobile TL-50
Core Speed: 798Mhz with a multiplier of 4.0. The bus speed is 199Mhz.
At the top the Core VID is: 0.8V

Just realised I should have taken a screenshot and uploaded it really ;^)

I ran CPU-Z on an Intel laptop. Its clock speed was 1196Mhz with a x6 multiplier. thats quite a lot faster isnt it. but the Core VID is only 0.975V

Do you think it could possibly be a dying CPU then?

Much thanks,
Dan Gent

[scaryjim]

At idle 800MHz with a 4x mutliplier is right - that's the lowest power state for the processor. You need to run something CPU intensive (like Prime95) and watch CPU-Z - the multiplier should jump to x8 and the core speed to ~ 1600MHz. If it doesn't, first make sure you've got your power management settings set to highest performance, then try again. If it still doesn't work, *something* is wrong, because the CPU shouldn't be sticking on lowest power settings under heavy load. It could be the temperature spiking very quickly due to a poorly seated heatsink which could be fixed be reseating it with fresh thermal compound; it could be the processor dying; or it could be some other part of the motherboard chipset causing problems. What laptop / chipset is it? I had some real issues with an Turion in a GeForce 6200 GO based chipset...

[DanGent]

of course, with mobile CPUs the multipliers change really well for power consumption. In the Acer Epower management thing the CPU speed was only on medium even though the setting was "AC power - maximum performance".
Bit riduclous, maybe someone had messed around with the settings.
Now with the CPU performance set to maximum running Prime95 in stresstest mode (the top option for CPU testing) it goes x8 multiplier and 1596Mhz base clock and 1.075V. Does that sound about right?
Speedfan is now measuring 60c and still climbing! I'll leave it on for a few hours (and also try the Prime95 setting for maximum heat) and see what happens.
could it be that Memtest was causing it to overheat yet any testing in Windows was severely curtailed by the Acer EPower things?

Learned a few new things today anyway scaryjim. Is it manchester UK you're in? thats pretty much where I am as well (just in Lancashire really).

[scaryjim]

... could it be that Memtest was causing it to overheat yet any testing in Windows was severely curtailed by the Acer EPower things? ...

I guess that's possible, although I don't know a huge amount about the way memtest does its testing. If that was the case though, I'd expect the first few passes of memtest to go through OK and the number of errors to increase as the test went on. Certainly it sounds like the reason it passed Prime95 @ 40 degrees was that the CPU wasn't stepping up above 800MHz, which would keep the temps much lower!

And yes, I'm in Manchester, UK - as are quite a large number of Hexusites, I believe...

[DanGent]

"I'd expect the first few passes of memtest to go through OK and the number of errors to increase as the test went on"
hmm. that does make sense. The errors come up almost straight away.
The temperature of the CPU seems to have stabilized at 60c running prime95 at full steam. I guess this only narrows it down to a motherboard or CPU problem!

[Zak33]

Zak33 - do you think it might be a heat problem? Although the laptop felt quite warm around the keyboard. Speedfan (temperature monitoring software) shows the laptop as being very cool. Maybe a little too cool I was thinking!

I disassemble laptops all the time. Its not hassle to reapply thermal paste. Have you experience of this kind of thing working before? I suppose its got to be worth a try!

yes. Memtest can throw wobblers if a chipset is not dealing with the test properly.

And lets face it... what's it gonna cost to check? Nothing.

[DanGent]

To add to this post, I took an AMD Sempron out of an old broken laptop I've got (same socket S1) and the same thing happened. Memtest errors after a few minutes.....
Looks like a mobo problem then!