ASUS P5n32-e SLI PLUS / TWIN2X2048-6400 - Very Strange Problem!

Hi,

I have 2 packs (4x1GB sticks) of the TWIN2X2048-6400C4.

Summary: 3 Sticks work, 1 Won't. No matter the combination on my ASUS p5n32-e SLI plus. Yet on other computers the faulty stick works!

This is the strangest computer problem I have ever experienced!

Here are the things I've tried and the results...

Installed all the RAM on ASUS "p5n32-e SLI Plus" but only 3GB were recognised. (even with 4th stick installed, the system would boot and run as normal)

Located the stick that wouldn't post and tried it in ALL slot configurations with other sticks, but no luck.

Tried the faulty stick on its own on each slot and the system will not even post.

Tried the faulty stick with the each of the other three in dual channel (2GB total) and only 1Gb would recognise.

I concluded the stick was faulty and returned to the supplier. They tested both dimms and found no fault. Complexed I tried again, (in all configurations) but no fix and the same stick would still not recognise. Took the faulty stick to my brother's house and tried on his computer (ASUS P5B-VM) and it posted with no problems in singular mode and dual channel mode.

All the other 3 sticks work perfectly in 677mhz or 800mhz mode, also with different voltage settings but this one stick will still not recognise on my computer

I have tried every combination, every voltage and bios tweak suggested on these forums, updated the bios to each version with no luck. In short I have exhausted all options and cannot work out what can be wrong with this one stick as it works perfectly in other machines. As do the other 3 identical sticks in mine.

My supplier cannot return for warranty as their tests show the stick to be working (and it does also in my brothers computer).

What can I do? Please HELP!!!!!

I live in Australia.

Specs Are:

QX6700 Quad Core

TWIN2X2048-6400C4 (4GB Total - With 1GB not posting)

ASUS p5n32-e SLI PLUS

BFG 8800GTX

PALID 7600GS (3rd Monitor, tried RAM tests also with this uninstalled)

4x320GB WD SATA-II in RAID 0

LG DVD Burner

SilverStone ST60F (600w true powersupply)

can you list all the individual stick serial and version numbers?

not sure what number means what, so this is what is printed on them.

(These are the numbers of the pack with the faulty stick - I can get the other pack numbers if needed)

Faulty Stick...

Side 1

CM2X1024-6400C4

XMS26404v2.1

003658 07020830-0

4-4-4-12

800Mhz

1024MB

XMS2-6400

Side 2 (on a sticker)

n11360

acct70113048

Other Stick...

Side 1

CM2X1024-6400C4

XMS26404v2.1

001643 07020830-0

4-4-4-12

800Mhz

1024MB

XMS2-6400

Side 2 (on a sticker)

n11360

acct70113048

BTW, nice rig you got there.

Information like these would be helpful. give us all.

CM2X1024-6400C4

XMS26404v2.1

001643 07020830-0

Installed all the RAM on ASUS "p5n32-e SLI Plus" but only 3GB were recognised. (even with 4th stick installed, the system would boot and run as normal)

So it is only 3GB when in POST and under BIOS? Not relevant here but helpful later, under 32bit systems, only 3-3.5GB will be recognized. 64bit will work but have software compatibility issues.

What is the Vdimm you have worked with?

BTW, nice rig you got there.

 

Information like these would be helpful. give us all.

CM2X1024-6400C4

XMS26404v2.1

001643 07020830-0

 

 

 

So it is only 3GB when in POST and under BIOS? Not relevant here but helpful later, under 32bit systems, only 3-3.5GB will be recognized. 64bit will work but have software compatibility issues.

 

What is the Vdimm you have worked with?

What I'm saying is that this one particular 1GB stick....

CM2X1024-6400C4

XMS26404v2.1

003658 07020830-0

will not post at all in my machine, it will post in other machines but not mine. I have tried it on it's own in each slot and it wont post at all... If i try it with a second stick only the second stick posts and the BIOS shows that only 1GB is installed. I have tried it in every slot and combination with the other sticks but nothing, it never detects it as installed.

It also not a case of a loose connection or faulty pins, as in every other computer i've tried it in, it installs with no problems and posts every time, first time.

It's so strange that 3 identical sticks would work but this particular stick will not work, on its own, in dual, three or 4 stick configurations. All slots on my motherboard work fine with the other sticks.

Just seems to be some bizaar incompatability with this ONE stick and my Motherboard. I can't fault the motherboard however, cause it works perfectly with the other ram and is stable as can be.

(ps. I'm running vista 64bit ultimate and hope to make use of the full 4gb)

Can you please post the information like below for all 4 sticks?

XMS26404v2.1

003658 07020830-0

I understand how you feel, as i had 2 XMS2 DDR2 6400 that won't POST on one motherboard but will work 100% for 2 months on another.

Can you please post the information like below for all 4 sticks?

 

XMS26404v2.1

003658 07020830-0

 

I understand how you feel, as i had 2 XMS2 DDR2 6400 that won't POST on one motherboard but will work 100% for 2 months on another.

Yes it sucks ;)

OK they were all the same except the first 6 digits of the code...

XMS26404v2.1 < Pack 1, Bad stick

003658 07020830-0

XMS26404v2.1 < Pack 1, stick 2

001643 07020830-0

XMS26404v2.1 < Pack 2, stick 1

001105 07020830-0

XMS26404v2.1 < Pack 2, stick 2

002152 07020830-0

Only noticable difference is the bad stick is a much higher number on the serial?

As for the XMS2 Twin2x modules i got, their serial numbers are in sequence.

I will read your motherboard manual to see if there's anything that comes to mind. I am researching this problem.

There are certainly problems with that stick, even though it might be a tiny thing. I had similar difficulties with RMA as well, until i brought the CPU-Motherboard-RAM-HDD to the retailer.

Some of my modules stress tested perfectly for 24hr+ on another test motherboard, but just failed to POST (Blank screen with POST beeps) on one and failing Prime95 & Memtests on the 3rd motherboard.

This is a hellish problem to troubleshoot.

http://www.houseofhelp.com/forums/showthread.php?t=59571

I was strangely surprised recently at the amount of Corsair XMS2 6400 i have RMA. 3 sets of Twin2x. They have caused me to RMA a CPU & 2 motherboards.

After reading your motherboard manual... i seriously think it is a sweet motherboard.

I have been researching this strange issue for almost 2-3 weeks now. It is easier with 6 PCs in my office, plus spare components. :bigeyes: So, i understand your frustration and helplessness. I also had long RMA 'discussions'.

In your motherboard manual at page 4-24, there is a section that talks about:

[1] DDR2 Controller Reference Voltage

[2] DDR2 Channel A Reference Voltage / DDR2 Channel B Reference Voltage

Here's how you should proceed with testing.

[0] Reset your BIOS to default values.

[1] Set your DRam voltage (Vdimm) to 2.1v

[2] Change 3 settings:

- "DDR2 Controller Ref Voltage"

- "DDR2 Ch A Ref V"

- "DDR2 Ch B Ref V"

Test-01 - See if POST will correctly identify 4GB.

- "DDR2 Controller Ref Voltage" = "DDR2_Vol/2+10mv"

- "DDR2 Ch A Ref V" = "DDR2_Vol/2+10mv"

- "DDR2 Ch B Ref V" = "DDR2_Vol/2+10mv"

Test-02 - See if POST will correctly identify 4GB.

- "DDR2 Controller Ref Voltage" = "DDR2_Vol/2+20mv"

- "DDR2 Ch A Ref V" = "DDR2_Vol/2+20mv"

- "DDR2 Ch B Ref V" = "DDR2_Vol/2+20mv"

Test-03 - See if POST will correctly identify 4GB.

- "DDR2 Controller Ref Voltage" = "DDR2_Vol/2+30mv"

- "DDR2 Ch A Ref V" = "DDR2_Vol/2+30mv"

- "DDR2 Ch B Ref V" = "DDR2_Vol/2+30mv"

Once you got to what you consider stable baseline, reduce the Vdimm gradually and re-test. That will make those pretty dimms run cooler.

My research points me in this direction after testing various hypothesis. Due to some unknown reasons for nVidia chipsets, the voltage signal noise level might be high enough to interfere with some memory modules with IC that is more sensitive.

My contacts in various contract manufacturing facilities in east asia mentioned that sometimes batches of tiny 'cheap' components that they used to build motherboards with, has small variation in quality and tolerance which might cause higher degree of fluctuations in voltage (eg).

A senior quality assurance engineer in a plant is pressured by management staffers to sign-off with less than the normal level of 'quality' due to production schedule. BAM! Consumers eat dirt in a full load even when the component cost only 5 cents. Technically they are not broken parts, but not perfect either. The problem can be cascading along until something big enough and break compatibility.

This is not exactly a Corsair problem and is something that easily falls outside of standard engineering consideration.

I suspect changes to Vref might make overclocking characteristics a bit different.

Good luck and godspeed.

After reading your motherboard manual... i seriously think it is a sweet motherboard.

 

I have been researching this strange issue for almost 2-3 weeks now. It is easier with 6 PCs in my office, plus spare components. :bigeyes: So, i understand your frustration and helplessness. I also had long RMA 'discussions'.

 

In your motherboard manual at page 4-24, there is a section that talks about:

 

[1] DDR2 Controller Reference Voltage

[2] DDR2 Channel A Reference Voltage / DDR2 Channel B Reference Voltage

 

Here's how you should proceed with testing.

[0] Reset your BIOS to default values.

[1] Set your DRam voltage (Vdimm) to 2.1v

[2] Change 3 settings:

- "DDR2 Controller Ref Voltage"

- "DDR2 Ch A Ref V"

- "DDR2 Ch B Ref V"

 

Test-01 - See if POST will correctly identify 4GB.

- "DDR2 Controller Ref Voltage" = "DDR2_Vol/2+10mv"

- "DDR2 Ch A Ref V" = "DDR2_Vol/2+10mv"

- "DDR2 Ch B Ref V" = "DDR2_Vol/2+10mv"

 

Test-02 - See if POST will correctly identify 4GB.

- "DDR2 Controller Ref Voltage" = "DDR2_Vol/2+20mv"

- "DDR2 Ch A Ref V" = "DDR2_Vol/2+20mv"

- "DDR2 Ch B Ref V" = "DDR2_Vol/2+20mv"

 

Test-03 - See if POST will correctly identify 4GB.

- "DDR2 Controller Ref Voltage" = "DDR2_Vol/2+30mv"

- "DDR2 Ch A Ref V" = "DDR2_Vol/2+30mv"

- "DDR2 Ch B Ref V" = "DDR2_Vol/2+30mv"

 

Once you got to what you consider stable baseline, reduce the Vdimm gradually and re-test. That will make those pretty dimms run cooler.

 

My research points me in this direction after testing various hypothesis. Due to some unknown reasons for nVidia chipsets, the voltage signal noise level might be high enough to interfere with some memory modules with IC that is more sensitive.

 

My contacts in various contract manufacturing facilities in east asia mentioned that sometimes batches of tiny 'cheap' components that they used to build motherboards with, has small variation in quality and tolerance which might cause higher degree of fluctuations in voltage (eg).

 

A senior quality assurance engineer in a plant is pressured by management staffers to sign-off with less than the normal level of 'quality' due to production schedule. BAM! Consumers eat dirt in a full load even when the component cost only 5 cents. Technically they are not broken parts, but not perfect either. The problem can be cascading along until something big enough and break compatibility.

 

This is not exactly a Corsair problem and is something that easily falls outside of standard engineering consideration.

 

I suspect changes to Vref might make overclocking characteristics a bit different.

 

Good luck and godspeed.

Really appreciate your help on this matter, you're a champ. I'll test these settings tonight and let you know how I go.

We have tested 4 1.0 Gig modules of XM2S-6400C4D on Striker Extreme, also tested on P5N-E and P5N32-E SLI.

Load Setup Defaults and then set these settings:

SLI ready: Disabled

CPU Clock: Auto

Mem Clock: Auto

Memory Voltage: 2.0v

NB Core Voltage: 1.4v

TCL: 4

TRCD: 4

TRP: 4

TRAS: 12

Command Per Clock: 2

TRC: 24

Thanks,

I already tried these settings but no luck. This is not a 4GB issue but rather one particular stick of RAM that refuses to post. These settings work for the 3 other sticks just fine.

Please use the On Line RMA Request Form and we will be happy to replace them or it.

Is this ok even though it posts on other machines with no errors? Also I live outside the USA... is that ok?

 

Test-01 - See if POST will correctly identify 4GB.

- "DDR2 Controller Ref Voltage" = "DDR2_Vol/2+10mv"

- "DDR2 Ch A Ref V" = "DDR2_Vol/2+10mv"

- "DDR2 Ch B Ref V" = "DDR2_Vol/2+10mv"

 

Test-02 - See if POST will correctly identify 4GB.

- "DDR2 Controller Ref Voltage" = "DDR2_Vol/2+20mv"

- "DDR2 Ch A Ref V" = "DDR2_Vol/2+20mv"

- "DDR2 Ch B Ref V" = "DDR2_Vol/2+20mv"

 

Test-03 - See if POST will correctly identify 4GB.

- "DDR2 Controller Ref Voltage" = "DDR2_Vol/2+30mv"

- "DDR2 Ch A Ref V" = "DDR2_Vol/2+30mv"

- "DDR2 Ch B Ref V" = "DDR2_Vol/2+30mv"

 

.

Hey, thanks for the help but unfortunately none of these settings have improved the situation. Any other ideas?

Hey, thanks for the help but unfortunately none of these settings have improved the situation. Any other ideas?

Sorry it did not work for you. Try RMA the Twin2x pair.

I'll take another look at your board manual after some coffee.

hmm.. coffee didn't help. See nothing else that might help in your manual that is in my research area of signal noise and EMR.

RMA is the best option now.

-----------------------------------

If you still got a bit of time... So this is something to try.

Use only the 'bad' ram in slot 1 run at 500MHz or whatever is stable. That is your baseline.

When working with a baseline, minor improvements or worse-off results is a good indication. Different ways of crashing are good, at least for troubleshooting.

With the changes, does things get worse or better?

Try Vref -10, -20 and -30mV with 800MHz and VDimm at 2.1v.

If you got it working, just RMA anyway. There's something funny with that dimm.

Good luck.

TAGI

There are a few things I am not clear on, and every one else trying to help please refrain from comments till I can work with TAGI for a few posts.

 

If you can please call me at 800-205-7657 and ask for Ram Guy would be best to do the rest on the phone.

 

But to be clear the system will post with all 4 modules using i or 2 modules but it will not post when you have all 4 installed, is that correct?

Ram Guy,

An International call will cost me a fortune, so i'll stick to here if ok.

To be clear, one of my 4 sticks is faulty. I have identified which one this is and have tried many combinations of installation but this specifc stick will NEVER show up as installed in the bios.

So that's a simple conclusion right? the stick is faulty and needs to be replaced, but it gets weird cause the faulty stick works, perfectly, in other computers.

Some examples...

i.e if I have faulty stick installed on its own, the system won't post.

I.e. if I have faulty + another... posts but only 1GB is shown.

i.e. if I have faulty + another + another... posts but only 2GB is shown.

i.e. if I have all 4 installed... posts but only 3GB is shown

i.e if install only the 3 working sticks... it posts and 3GB is shown.

I have tried each of the above combinations in EVERY slot configuration also.

In all situations where the faulty stick is physically installed (besides on it's own) the computer will still boot and operate as normal... its as if the stick is not even installed.

So very strange.

How come it posts on other computers but not mine when identical sticks do work on mine and are listed as compatible?

That is odd especially if it will post in another system, can you tell me the make and model of MB you tried it in that it will post?

 

And let's get them all replaced, please use the On Line RMA Request Form and we will be happy to replace them or it.

The other system Motherboard is: (ASUS P5B-VM)

Also the reseller had it tested in several motherboards that are not identified... All work perfectly.

Will the RMA be ok as I am in australia, I am also conerned that if the stick works in almost every other mobo they will deny the return.

How long does RMA take internationally?