Please help confirm my faulty memory diagnosis

Memphis

Memphis

Memphis

Soldato
Joined
13 Jan 2004
Posts
12,197
Location
Inverness
Hi guys, I'm rather hoping you can help me confirm my diagnosis that I have a faulty RAM module :).

Firstly, the key specs:

Intel Core i7 920 D0
Gigabyte EX58-UD5
Corsair XMS3 DDR3 6GB - TR3X6G1600C9
PowerColor ATi HD 3870
OCZ StealthXStream 500W

I should state absolutely everything was at stock when fault testing was done.

Ok, so the rig was only built on the 26th as the bulk of the components were Christmas presents, including the memory. Over the last 4 days, I've mainly been building up the OS, getting things to my liking. All this time however, I can't say I've been entirely happy with the way things were running. This was just a gutt thing however, nothing scientific in any way. A couple of freezes, nothing too abnormal.

Last night however, I was attempting to unarchive a couple of files stored across multiple archive parts. I couldn't get the files to extract properly. I was getting consistent CRC errors. I fired up QuickPAR, thinking I had a damaged part, however it advised all parts were complete and undamaged. Confused, I copied the files to a laptop. The files then extracted without any problems what so ever.

Now worried about system stability, I fired up IntelBurnTest and LinX for the first time since building the rig. They both fail within a few minutes. With everything at stock and temperatures well within acceptable ranges, I Googled around and found suggestions of memory problems. I booted the latest version of MemTest86+ and let it run. Within half an hour, I was getting errors at one memory address. I ran the test a few times and every time got errors at this one address. Sometimes faster than others.

This morning I tested each stick individually. Letting the test run for 3 full passes, the first two sticks passed without a single fault. Within minutes of testing the third stick, I had errors and all at the same address. I replaced all three sticks and booted into MemTest86+ again. This time I got a full run before the same error address appeared again.

I should also add that I have also set the memory to run using the Intel Memory Profile, allowing the memory to run at full speed, timings and voltage. The errors still appear, seemingly at only the one address on the one stick.

I know that was a very long read, so to those who took the time you have my thanks :). Can we say the general concencious is I have a dead stick of RAM?

Thanks for any help!
 
Hi memphis,

It sounds faulty to me.

You could try and add a little to the voltage and see if that prevents the errors.

Otherwise its RMA time.
 
Thanks for your reply Jason :).

I had thought about trying to up the voltage, however I wondered if this was really necessary given errors are present when running at stock speeds and voltages but also when using the Intel Memory Profile which although increasing speeds also increases both the memory and QPI voltages?

I have to be honest, I'm very new to this setup so don't 'fully' know what each bit should be set to heh. I seemed to think that I should be careful with the memory voltages because of the memory controller being onboard the CPU?

Thanks again though!
 
You should be safe enough with the memory voltage at 1.64v to test it with, but as you say using the intel profile ups the ddr and qpi voltages and its still erroring in memtest, its looking likely that the ram is faulty unfortunately.
 
The Intel Profile sets the memory voltage to 1.65V and the QPI to 1.35V I think (whatever it is that Intel says the QPI should be). My understanding is you shouldn't really go higher? Though I could be very much wrong :), still learning heh.

I have actually already ordered a new kit that will be here tomorrow, and I pre-emptively requested an RMA with Corsair, which I now have and just need to send the memory off :(.

I really just wanted the confirmation that I haven't missed anything stupid and that it's likely I do have a dud chip.

Cheers guys.
 
Itll probably be set to 1.66 or 1.64v as unfortunately the ddr ram voltage on x58 boards can only rise in increments of 2, 1.66 is considered safe though allowing for vdroop, and 1.35 on qpi is max reccomended. Hope the rma goes smoothly and your new ram arrives quickly.:)
 
Well I've now got 3 new sticks of Corsair XMS3 installed and everything seems good :). It's certainly passed 3 full runs of MemTest86+ which I could never get before.

I do have a couple of quick questions now though.

1) Nothing is overclocked at all at the moment, so the memory is running around 1066MHz and only 1.5-ish volts. Should I just load the Intel XMP and leave it be?

2) What is the likelyhood, if any at all, that a weakness in the system (power supply?) caused the fault on the one stick before when I loaded the system with IntelBurnTest and LinX? This is more for my own sanity so I don't manage to kill anything else by running them again!

Thanks once again guys and gals.
 
Hi memphis glad its up and running again.

If I were you I would be tempted to leave it for now, just to make sure all is ok.

Then start again.

If it was the PSU I expect it to cause errors when the GFX card was maxed out and not on a CPU or RAM stress test. Of course I could be wrong.
 
Thanks again for your reply Jason.

I actually loaded the Intel XMP last night before heading to bed and left it running MemTest86+ overnight. It passed 7 full runs and was most of the way through the 8th, so I'm happy enough that it was a faulty stick causing me problems before.

I've actually got a new PSU coming sometime next week (Corsair TX850W) so I think for safety's sake I'll leave everything as it is for now.

Now to RMA the dud set :(...
 
You must log in or register to reply here.
Back
Top Bottom