Good memory works in some slots, but not others

x_jose_x

Distinguished
Feb 16, 2009
7
0
18,510
Hello,

Motherboard: Asus M3A79-T Deluxe
CPU: AMD Athlon 64 X2 6400+
RAM: G.Skill 2GB DDR2-800 PC26400 5-5-5-15
OS: Windows 7 64bit

I recently attempted to upgrade my system from 4GB of RAM to 8GB. The current two sticks work just fine. They pass memtest. However, when I installed the two new sticks the system started freezing. I ran memtest on the new sticks, and they fail. Through trial and error I ended up testing the two good sticks in the slots where I had put the new RAM. They failed memtest in those slots, however, they pass in their original slots. The new sticks also pass memtest in the other slots. At this point I am thinking perhaps it is the mobo and not the RAM.

To test this theory I placed two Corsair 1GB DDR2-533 PC24200 sticks in the slots that are failing memtest, and there was no errors. I booted the system and it ran fine. This seems to support that the problem is not the mobo.

Futhermore I tested the two new sticks in a different system, and they pass memtest and run the system just fine.

For some reason my mobo does not seem to like having the G.Skill sticks in those two slots.

Perhaps I can get some input as to why? Is there a BIOS setting I am missing? (I have updated the bios to the most recent version) The mobo supposedly supports that much RAM. This is a mystery!

Thanks,
Jose
 
One of the more common causes for system freezing is not enough voltage being supplied to the RAM. That's my guess for the following reasons:

1. The default voltage that most motherboards supply for DDR2 is 1.8V. That G.Skill memory looks like it's 1.9V or 2.0V depending on which model you have.

2. The Corsair RAM you mentioned is the standard 1.8V and works fine.

3. When you have all four DIMM slots filled with memory, often times it will require even MORE voltage than when you have just two sticks installed. So it's possible that with two sticks of the G.Skill RAM, the underpowering issue was not severe enough to lock up your system, but inserting four sticks made the problem worse.

3a. Or, it's possible that your board was autodetecting the right voltage of 1.9-2.0V with just the two sticks in, but with four sticks you needed an additional .1V or so bove that, so the autodetect settings were no good anymore.

4. When using only two sticks of RAM, you should insert them into the lowest pair of DIMM slots only (e.g. DIMM0 instead of DIMM1, or in the case of ASUS, I believe it's DIMM_A_1 and DIMM_A_2 instead of DIMM_B_1 and DIMM_B_2). Inserting them into the higher-paired slots can cause issues, as can inserting one stick of ram into one DIMM pairing and one stick into the other DIMM pairing (e.g. A_1 and B_1). If you are inserting the sticks into two slots that are right next to each other, odds are you've got it set up wrong. Most often there's an empty slot between the sticks if you're doing it right.


Fortunately, there's an easy fix in the BIOS that lets you set the memory voltage manually. Go in there and see if it's set to the correct amount for your RAM's specs. You may also need to manually set the timings, but that's pretty self-explanatory as well.
 

x_jose_x

Distinguished
Feb 16, 2009
7
0
18,510
Hello,

Thank you for the reply.

According to the Asus manual the slots are A1 A2 B1 B2 from left to right. I have just two sticks in A1 and B1, that is not next to each other. It is when I have sticks in A2 and B2 that I have issues.

I will increase the voltage and see if that helps.

-Jose
 


Ahh. Thanks for clearing that up ... couldn't remember whether ASUS boards went by letter instead of number, but it sounds like A1-B1 is the correct pairing if you only have two sticks, then. If you have sticks in ONLY A2-B2, there's definitely a chance that the system won't like that. So I hope that means some of your memtest failures were false positives and it was really a voltage issue. I'll keep my fingers crossed for you.

 

x_jose_x

Distinguished
Feb 16, 2009
7
0
18,510
It appears that it was a voltage issue. I had tought of that, but was too conservative in modifying the setting, from 1.8 to something like 1.84. I went for 2.0v and that seems to have solved that issue. However, the PC now reboots randomly at different times. I am guessing a heat issue. Next time I boot up I'll watch the temps and see things are over heating. I'll also set Windows to not reboot on BSOD to see if there is some other possible issue. Thanks again for pointing me in the right direction.
 

x_jose_x

Distinguished
Feb 16, 2009
7
0
18,510
Ok, the problem is back. Windows just freezes, and the memory is failing memtest again. I know the RAM is good, I know the mobo is good. Voltage is 2.0v, temps appear normal.

What else could it be?
 

x_jose_x

Distinguished
Feb 16, 2009
7
0
18,510
I think I have finally solved this problem.

After tinkering with the BIOS setting so much, I decided may be it would be a good idea to load the defaults. After loading the defaults the system runs fine, and there is no more memtest errors.

After restoring the BIOS defaults, the setting that changed in the BIOS is Clock Skew. The Clock Skew setting was automtically changed after enabling the OC profile in the BIOS. I enabled OC profile orginally to adjust the RAM voltage.