again bad module

[VGR]

Hello,

 

I bought a Part Number: TWINX1024-3200XLPT less than one year ago, counting on the "lifetime warranty" and the good reputation of Corsair memory.

 

That module runs DDR400 dual channel at 2-2-2-5

 

I already returned the set of modules in May 2006 (RMA Number: R126210) and now the replacement module again experiences errors. I get blue screens on my server. MemTest shows errors in first pass (high portion of memory, so I can run for one or two days before I get a blue screen ; yes, this is windows ;-).

 

I don't want to downgrade to 2-3-2-5. I would like to reduce the number of unattended reboots. I would like to have "lifetime" RAM modules that last longer than 6 months ;-)

 

This time, I'm buying an other brand's 1GB memory module (only 3 years warranty, but handled in my country) to enable my server to run in degraded mode during the shipment (expensive!) to Corsair, USA.

 

I would like to have a new RMA number, thanks.

[RAM GUY]

Can you tell me the exact bios settings you have set for both CPU and memory?

[VGR]

no overclocking at all. machine is stable and doesn't overheat (43°C at full thottle). Memory is set as 2-2-2-5 althought I wonder if it isn't set simply to "autodetect". I don't have full RAM verification at POST. I don't use ECC memoty (I should, probably ;-)

 

If you need "exact" settings, (including voltages?) I need to reboot and this is a server. Again, the machine is stable except for those RAM problems. The first set was RMA'ed and exchanged so I suppose the errors that MemTest (under DOS) found were real and discharge the mobo from any suspicion ;-) and I have the feeling it's the same case again now.

 

I could have tried to downgrade to an other setting than 2,2,2,5 but I doubt this is the correct solution for a guaranteed 2,2,2,5 RAM.

 

please advise

[RAM GUY]

First Set? How many modules do you have installed? And again what are the exact bios settings you have set and what do you have the memory voltage set to?

[VGR]

and again, it's set to auto.

 

I had one more blue screen, so I left the server down while copying BIOS settings on paper, just for you.

 

here they are :

 

CPU Type AMD Athlon 64FX 3200+

BIOS W7025NMS v 1.4 112204

chipset nForce3 250, ECC disabled

 

QuickBoot

Phoenix-Award BIOS settings :

PC Health status 34°C/93°F

CPU 40°C / 104°F

NBFan 5357rpm

CPU fan 1962rpm (Zalman full Copper)

VCore 1.37-1.39 V

others rails ok too

 

DRAM Config

Max Memlock : Auto (hence 200MHz)

1T/2T : 2T

CAS# : Auto ]

Trcd : Auto ]

Tras : auto ] hence 2,2,5,2 128bits Dual Channel DDR401

Trp : auto ]

 

High Performance : Optimized

Aggressive Timing : Disabled

Dynamic Overclocking : Disabled

Spread Spectrum : Disabled

HT Freq : 5x

Cool 'n Quiet : Disabled

Adjust CPU ratio : Startup

FSB : 200 MHz

AGP : 66 MHz

CPU VID : Startup

 

That's all I have.

 

regards

[RAM GUY]

But how many total modules do you have? And these modules have a minimum suggested Voltage of 2.75 Volts so with this MB I would set the memory Voltage to 2.8 Volts. If you have more than 2 modules you may have to relax the timings or run the memory at DDR333 depending on the core of the CPU!

[VGR]

ok, I was in the process of checking the acceptable voltage range of the modules I have now.

 

To answer your question, I used to have only one set of modules : the Corsair PC3200 XT something of 1GB which is a 6 month-old replacement (RMA) of a previous identical set purchased one year ago. That memory is guaranteed to run at 2,2,5,2 / 2T

 

Given that I had to shut down my main server to send the original set in an RMA, this time I purchased a second set of two 512 MB SDRAM PC3200 modules. Those, apparently, can also operate in dual channel DDR400 but with different CAS/RAS/etc settings : 3,4,8,4 / 2T

 

When I exchanged both sets, I hoped to regain my rock-solid stability but in fact it was worse than ever : one BSOD every two hours or so, with error code 1D and no trace of a failing driver, and every memory or kernel memory image was corrupt - this usually indicates bad RAM, if I'm not mistaken.

Windows, even set to "reboot automatically", wouldn't reboot automatically either.

 

When I plugged both sets in (thus filling the 4 banks with modules, in an appropriate manner I think ;-) I got my 2GB but only at DDR 333 or 266 which is normal. Instability was the same.

 

In the meantime, I was searching which "drivers" could be a problem, given I had done no modifications at all before the BSoD begun to appear. I looked at the VIA 4-in-1, the VGA driver etc but they are all either the last available or the latest necessary release for being stable enough for me. I don't usually play with unstable beta releases or upgrades except in case of a documented problem of mine being solved. For instance, I can not use my Matrox G550 dual Head on this mobo since the start, and as the problem is not certainly solved in BIOS R 1.8+, I didn't upgrade to more than 1.5 (probably)

 

I found traces of the "memory voltage trap", but it seems to be related to overclocking which I don't do, and my settings were on "auto" with a new and sufficient PSU, so I was a bit lost. Moreover, I don't see how nor why I should set memory voltage to 2.75 given I don't even see the memory voltage set in my BIOS. It has to be automatic.

 

Both memory sets are supposedly compatible with the mobo/chipset.

 

Then I finally gave the Corsair set one last chance on its own, and since I un-plugged/re-plugged the memory modules (leaving only the Corsair 2x512 in the original slots & combination) I have no BSoD any more. It's a bit too early to affirm the problem is gone, but if it is the case I will post a follow-up in here.

 

It ***seems*** I have this problem :

- no modifications in drivers nor hardware

- sudden BSoD due to an unexplicable bad contact of the memory plugs

- solved by re-plugging memory modules

 

Confirmation or infirmation will follow in a couple of days

 

Thanks for your attention

[RAM GUY]

You cannot mix memory with this platform all of the modules would need to be exactly matched both in part# and revision. In addition, if you use more than 2 double sided modules it will by default down clock the memory to DDR333 that is normal.

[VGR]

Hello,

 

First to answer your message above, yes I know. The corsair twinset are double-sided, and the second set is single sided. I am not expecting more than I can have ;-) I know I will have dual channel but with inferior DDR speed.

 

Rest assured I never mixed the modules. Both pairs are perfectly matched (especially the Corsair TwinX , by purpose ;-)

 

 

Now let me come back to the situation.

 

I have nly the corsair set of two mudles set in slots 3 and 4, so called "channel B" and I get DDR400 dual which is nominal. The mùachine ran like that for months. I think I have sufficient PSU power, or else Tagan will hear from me :D

(I must confess that - like 99.99% of the people - I've no direct tools to measure the actual throughput on the rails ; all I have are the software/bios tools and everything seems ok)

 

Now here's the situation :

- I still have BSoDs and no driver is mentioned

= I also got one windows "freeze", which I don't consider abnormal but not that common either.

- I rebooted, and now I've resource starvation on windows after 2 minutes of use. That's pretty abnormal.

- I tested the RAM using memtest -> BSoD

- I stopped the machine, unplugged, cleaned and replugged the two memory modules. I started the machine. Same story about resource starvation (I can't launch anything, I have "unable to create socket XXX", no more bitmaps on screen, the fonts disappeared... like on an old overloaded Win 3.1 without RamDoubler ;-)

- I rebooted and envisaged to "relax" the memory settings from 2,2,5,2/1T to a "random" combination using 2T but first I forced the memory voltage to 1.75V (despite the red flashing message on my BIOS settings screen "not recommended, system may be unstable") and tried to boot.

- same story : resources starvation after having launched 6 applications essential to the server and representing less than 100MB memory. Now I have memtest running. It passed 100% without any error but this is just the start and doesn't prove a lot since I can't use the machine anyway (no more resource, like a full GDI segment).

 

I fear I won't recover my machine, at some point very soon.

 

Those problems occured all of a sudden and no hardware modifications had been done recently, except for some USB devices use here and there.

 

UPDATE : memtest found a memory error "error copying between word pairs #... and "..."

 

What is your best advice ? Relax memory settings ? Accuse the mobo ? Accuse Windows ?

 

From what I read on MSI, Corsair and hardware forums, this problem is either a bad PSU, or bad RAM. My RAM is only the Corsair Twin3200 set.

 

thanks for support.

[VGR]

ok, I moved memory voltage up to 2.8 V as recommended, I lowered the memory settings to 2.5,3,8,3 as recommended and now Windows (NT5) won't boot any more. Cool day.

 

I hope this hasn't to do with the RAM... I suspect a bad module since the beginning. Last time, a defective Corsair twin set has made me reformat the machine from scratch. It was 6 months ago. I don't want this again. I really hope that this time, it's not the RAM again... :mad:

[SLIM-19]

Sorry to read about all your issues. Just wanted to say your not the only one and I sympathise with your situation.

 

Seems to me that Corsair isn't the reliable manufacturer that it once was, I've always had Corsair But I fear its now time to jump ship to a different manufacturer.

 

I've had to cut a loss with my ram and move on (i'm spitting feathers too)

I had £230 of TwinX 2048 4400Pro which you'd expect to be rock solid, After Blaming everything else first I reluctantly turned to the Ram and eventually confirmed that it was at fault.

 

After shipping it back to Corsair in the USA from my native UK (not too cheap) They eventually sent a replacement 408Hrs later which is a far cry from the acclaimed 72Hr turn around time they stated.

 

I was delighted to recieve it UNTIL I OPENED THE BOX only to find two seperate 512Mb sticks not even a matched pair! when i had sent a matched pair of 1Gb sticks, Moving on slightly its now mid November and this was the beginning of October and i've had countless emails ignored:mad:

 

Looks like i'll be getting some ******** Ram instead:laughing:

[RAM GUY]

SLIM-19

I am sorry that you are not happy with us and that you have had such a hard time, but this post is on the verge of getting you banned from the forum.

Please send an email to me at Ramguy@corsairmemory.com and put ATT Ram Guy in the Subject line, please include the time zone and when you can be reached and we will do our best to help you get your issue resolved.

In the future please just start your own thread and I will be happy to help you!

 

 

VGR,

I am still not clear on exactly what modules you have in your system

Do you have Twinx1024-3200XL or do you have Twinx1024-3200?

You can call our tech support at 510-657-8747 or send me an email with the time zone you are in and put ATT Ram Guy in the subject line and I will be happy to call you.

[VGR]

thank you Slim-19 for your compassion and sympathy ;)

I also had to pay a not-that-cheap shipping to RMA to Corsair.

 

RamGuy, I have a Twinx1024-3200XL and am at GMT+1 but not in front of this server during day hours. It is supposed to run unattended and smoothly :/

If you want to contact me, use this place.

 

I know it's not 100% sure, but I ran memtestx86 for 4 hours on that RAM and found no errors. Now I ***hope*** to be able to recover my windows installation - right now I'm struggling with the crappy recovery console -

 

More news later if I find a RAM-related cause or otherwise solve my problem.

 

Last time (6 months ago) it was the same story because of the defective first Twinx1024-3200XL : slight memory problems degrading the machine slowly step by step, until even the BIOS loses settings and of course the OS does not start any more (ntldr missing, etc). All because of the RAM.

 

I will perhaps move to Ubuntu, hardware permitting. At least a *nix will always boot, one way or the other. Except if RAM is bad, of course ;-)

[SLIM-19]

VGR

Sorry to hi-jack your thread the other day but it appears this was the only way I was able to get a response from Corsair.

 

Hope you get your issues resolved soon.

 

RAM GUY

I apologise for previous comments made however relevant they may have been, But i'm sure you understand my concerns regarding the problems I have faced trying to converse with Corsair unsuccessfully many times.

 

You will be receiving an email from me shortly entitled; "ATT Ram Guy", as you intructed.

 

Regards, SLIM

[RAM GUY]

Again,

I am not sure what the problem is or was but I will do my best to help you get it resolved.

And again you can call us at 888-222-4346 if you are in the USA other wise 510-657-8747 is the direct number.

 

And there is no email from you as of yet, so if you sent it please send it again!