RAM issue - need a utility if there is one

I have an old laptop with a RAM issue. The issue is unfortunately not in the memory modules themselves. I get a "bad" address in MemTest at a fixed location, regardless of what modules are in the machine. I've tried completely different SODIMMs but the faulty address remains.

I don't really understand how that's possible... but that's the result. Could it be a fault in the CPU rather than RAM or Mobo? Would a new CPU fix it?
The CPU is a T7700, so maybe could be had cheaply on eBay.

Aside from this, the laptop works fine... other than not having a screen so I have it plugged into an external monitor. I'm using it now to type this.

When Windows uses that faulty address, I get a BSOD with MEMORY_MANAGEMENT error.


My thinking is... could a utility load up into a fixed location, occupying the bad memory sector, and preventing any other program from using it? Does such a utility exist?

 

The utility would need to address the sector which would cause a BSOD anyway.

 

Try one slot at a time and see which one gives you an error. Possibility of a bad ram slot - it can happen

cheers ...

 

Thanks for the replies.

Any SODIMM, either slot or both, same address at fault every time.

If it's a bad slot, do you mean it may just need a clean or a physical fault beyond that?

 

The issue could also be with a motherboard trace. Some things only pop in the most isolated of ways.

 

Have you checked and making sure that the current ram modules are for this model?

When one has a bad slot, one can try cleaning it up, but most of the time it is the main board issue

cheers ...

 

What was the memory address?

Memtest DOES test CPU L1 and L2 cache so it could pick up an error there. A memory address (and what module it was doing), might help pinpoint the location....

 

This is what I get (attached)... this happens regardless of what I have in the slots. At the moment that's the 2 x 2Gb Crucial RAM in there, but I have tried 2 x Kingston Value sticks at lower speed, and also the original 1 x 2Gb stick supplied with the machine. All work fine in other laptops, all give the same error in the same location in the 'at fault' laptop.

 

I'd reflash the BIOS of our computer, and then clean up the pins that the ram modules connect to.

if you're still getting that error then may as well start looking for another computer. Either that or install less memory than where the bad address occurs .. i.e. 1.5GB RAM

 

I've got my eye on a 17" Clevo laptop... just that the laptop is 95% compatible with MacOS-X so I wanted to keep it going.

Refresh the BIOS - do you mean re-flash, or just wipe settings? I've tried clearing the settings already, but not re-flashing it.

 

Reflash, by the way, it is against the forum rules to discuss a certain OS on non fruit computers, as long as the discussion isn't about that it's fine, this is just to prevent from getting in anydirection related to that particular topic.

I don't like to be teh bearer of bad news, but it looks more and more like a motherboard problem.

 

If you're using win7 then could remove RAM so run with 2GB physical then limit the memory use by Win7 to be 1.7GB: Memory - Set Maximum Amount Used by Windows 7 - Windows 7 Forums .

Other options I would consider would be to decrease the CAS rating on a module (spdtool or Thaiphoon Burner) in case the RAM timing is a problem. Could also use memset to slow things down. Finally, could use Throttlestop to set SLFM mode on your CPU (100Mhz FSB), halving the internal CPU cache speed to see if it's the problem.

 

at 5-5-5-15 I dont think ram timing is the problem. Besides, if he's getting the same error address with different modules then obviously the problem is somewhere else.

- by "reflash" I really mean reflash the BIOS with the newest you can find from the manufacturers website. Thought it's not directly connected with the RAM modules it might solve the issue. But I'm more leaning towards bad connection between the memory slot and the memory module itlself, and if it's not that then it will most likely be not worth dealing with.

 

Pins cleaned (switch contact cleaner and a cotton swab).
BIOS flashed, and all settings redone.
...problem remains.

Do you think it's worth buying a different CPU to try, or is that unlikely to make any difference?

 

No, don't think the CPU is the issue. Give the address, the memory error is occuring higher. Since it only occurs with two ram sticks in, the problem has to be in the dual channel implementation.

At first I didn't get where he was going with the BIOS flash, but now I get it.

How does this manifest itself in day to day operations? Are you get BSODs? Or, did you just run memtest and it is bothering you? Memtest is not perfect...might be a bad read.

Did you try Memtest+ or an earlier version of Memtest?

 

Tried a few different versions of MemTest and Memtest+. Aside from one that flips out completely (I think it doesn't run properly with 4Gb RAM Installed) they all gave the same address.

The fault is there with only one stick in either slot, same address, not only with both slots occupied.

I get BSODs... usually right in the middle of something important. Of course, if I'm just messing about on the web, it rarely happens. Sometimes I get BSOD on boot. It's always the same type of memory-related error, either MEMORY_MANAGEMENT or PAGE_FAULT_NONPAGED_AREA - mostly management. I get it a few times a day at least - if it's in a bad mood sometimes it'll take me a few attempts to log on and get into software.

 

try blowing some air around the RAM module connectors. There might be some particle that is making contact between two pins under the board where the connectors are soldered.

 

How do you know windows is using that particular address, an assumption?

You might be able to disable Windows from using a bad memory area via E820 but it's not easy to do. You would possibly have to hook the interrupt via a specially coded MBR loader and include the bad memory as reserved in the SMAP table.

 

Could try grub2 bootload's BADRAM config entry. badram - GNU GRUB Manual 1.99 .

 

I sent the dump file from the crash to a friend that knows how to analyse them, and they said the address matched the one in MemTest.

If I knew what that meant, I'd give it a go! I very much doubt I could code that in Visual Basic though, lol!

That looks like it might be absolutely ideal! Thank you, I'll give it a try!

 

and a different crash analysis will show a different program in that location.

Look up ASLR. It's been years since windows had a static memory map.

 

Yes, it's not a simple solution.

I'm not sure what OS your using. If it's vista or later you could try

Code:

bcdedit /set {badmemory} badmemorylist 0x70497

providing that is the correct PFN and there is only one bad place. If there are more bad spots you can add to it, each PFN representing 4k of RAM.

 

True, but they said everything pointed to the location being constant.

Yes, it's Win 7 64Bit Ultimate. I will try that before messing with the GRUB bootloader etc.

Thank you!

 

I second QHN and state it may be a slot and to check the memory with 1 slot empty and the other occupied.

 

If you read back on the thread you'll see this error occurs in the same memory location regardless of configuration.

I have tried:

Slot1: 2Gb DDR
Slot2: 2Gb DDR

Slot1: 2Gb DDR
Slot2: Empty

Slot1: Empty
Slot2: 2Gb DDR

Slot1: 2Gb DDR (borrowed RAM)
Slot2: 2Gb DDR (borrowed RAM)

Slot1: 2Gb DDR (borrowed RAM)
Slot2: Empty

Slot1: Empty
Slot2: 2Gb DDR (borrowed RAM)

...same error all the time.

Plus, memory sticks all check out in another system.