Random Memtest errors. Can be faulty cpu memory controller?

[litwicki22]

Hello all. I have a CPU 4790k stock and rams Corsair Vengeance 2x4gb

http://www.morele.net/pamiec-corsai...-9-9-9-24-xmp-white-cml8gx3m2a1600c9w-419635/


I can run Prime 95 without errors, SupePi , games are stable. However yesterday memtest shows errors after 5 minutes of test. I repeated test and now i am in test 24 hours without errors. So errors dont back.
Today i am running memtest 24 hour without any errors. So it can be still memory or memory controller in cpu?


If memory errors in memtest are random ( not in the same places in test ), can be issue with cpu memory controller?


Change ram or cpu then? Please help

 

[Idontcare]

What you are describing is a type of failure that is one of the hardest to debug - a walking wounded hardware issue.

Walking wounded means it intermittently fails, possibly due to temperature at the time or a voltage fluctuation.

But then when you go to debug and isolate the error (typically a process that requires the error to be repeatable) you can't get the error to happen again because the boundary conditions and initial conditions are not identical anymore.

But to answer your question - it doesn't necessarily mean the error is in the mem controller, it can be the ram still. Tricky to isolate the trouble on this one. Any reason to suspect the mem controller? (was it running overvolted ram before?)

 

[litwicki22]

I am just asking becasue i had errors now i dont have any errors

I suspect the mem controller because errors was random ( not always in the same places in test ). Three days ago i get 34 errors after 5 minutes. But after that memtest not showing errors anymore. Now i can test 24 hours without single error. How is this possible?

 

[2is]

It's probably going to be impossible to narrow down what's at fault without swapping hardware. I'd try testing with one stick at a time.

 

[redzo]

Maybe is too early to blame the memory controller.

Have you tried testing the modules one by one? Since you've got 2 modules, there's a very low probability that both of them are malfunctioning if the memory is indeed the issue.

Malfunctioning PSU's and RAM memory corruption are probably the trickiest to identify. Many people do not get to this point and they just replace good working components like hdd's or do pointless OS reinstalls.

The devs of memtest themselves say not to bother interpreting the results too much. You get horizontal red lines(errors), that's it. You should stop here. I advise everyone to read the memtest FAQ as it is important to know for yourself what memtest can and cannnot do for you.

But now you are at least sure that memory corruption occurs and you can say with 100% certainty that a system experiencing memtest errors is an unstable system. So it's up to you to continue troubleshooting after this step and figure out what's causing this: the most logical thing here is to blame one of the modules. I hope this is the case, since really identifying a bad cpu or mobo is the nastiest part if you do not have an identical working component at your disposal.

 

[Tuna-Fish]

It could also be a fault in the motherboard. Even if there are no silicon components on the memory path on it, if some of the memory bus traces are a bit marginal, it's possible to get random memory errors if the voltage/temp is a little bit off.

It's very, very hard to locate the sources of these kind of errors.

 

[ClockHound]

Have you checked with your local cosmologist? You may have tested during an intense quark shower in your area three days ago.

Had a similar issue last year. The good news is, it's solved. The bad news is, it took 3 months to definitively track down the intermittent ram stick.

 

[litwicki22]

How did you explain this? 3 days ago i get 34 errors in memtest about 5 minutes after started test. Then i stopped test run again and after that i can run 24 hours of memtest without problem. So errors stopped of unknown reason. I dont change anything in bios. Its mobo, ram, psu? I want to change now memory but i am not sure.

 

[VirtualLarry]

How did you explain this? 3 days ago i get 34 errors in memtest about 5 minutes after started test. Then i stopped test run again and after that i can run 24 hours of memtest without problem. So errors stopped of unknown reason.

You're not going to get an exact answer. IDC gave you the best explanation.

I dont change anything in bios. Its mobo, ram, psu? I want to change now memory but i am not sure.

You're actually going to have to do work on this one. No-one from a forum will be able to give you an exact answer on what, if anything, needs to be replaced. Not without some exhaustive diagnostic work on your part.

Edit: For those that don't know, "litwicki22" is a known tech-support troll that likes to start multiple threads, gets advice, and repeatedly ignores the advice, and goes on to repeat the same question over and over again, and starts more threads.
http://forums.anandtech.com/showthread.php?t=2328724&highlight=litwicki22

Supposedly, his problem is fixed, according to his original thread in Computer Help.
http://forums.anandtech.com/showthread.php?t=2428588

But he'll just keep repeatedly asking "why", until the mods lock this thread, like so many others he's started.

 

[redzo]

How did you explain this? 3 days ago i get 34 errors in memtest about 5 minutes after started test. Then i stopped test run again and after that i can run 24 hours of memtest without problem. So errors stopped of unknown reason. I dont change anything in bios. Its mobo, ram, psu? I want to change now memory but i am not sure.

Better check Idontcare's post once again!
You want a precise straight to the point answer which no one can deliver. It's not unusual for malfunctioning hardware to behave like this. Of course that it's a pain in the a*s to troubleshoot, but sh*t like this just happens.

Why don't you just test/use the system with one module and then with the other? It's plain simple. The more tests / the more you know, the better in your quest of identifying the faulty hardware component.

Edit: For those that don't know, "litwicki22" is a known tech-support troll that likes to start multiple threads, gets advice, and repeatedly ignores the advice, and goes on to repeat the same question over and over again, and starts more threads.
http://forums.anandtech.com/showthread.php?t=2328724&highlight=litwicki22

Supposedly, his problem is fixed, according to his original thread in Computer Help.
http://forums.anandtech.com/showthread.php?t=2428588

But he'll just keep repeatedly asking "why", until the mods lock this thread, like so many others he's started.

That's quite rude litwicki22 ... Burning neurons for no good, that's not nice ... not nice at all.

 

[Idontcare]

Edit: For those that don't know, "litwicki22" is a known tech-support troll that likes to start multiple threads, gets advice, and repeatedly ignores the advice, and goes on to repeat the same question over and over again, and starts more threads.
http://forums.anandtech.com/showthread.php?t=2328724&highlight=litwicki22

Supposedly, his problem is fixed, according to his original thread in Computer Help.
http://forums.anandtech.com/showthread.php?t=2428588

But he'll just keep repeatedly asking "why", until the mods lock this thread, like so many others he's started.

I had absolutely no idea D:, thank you VirtualLarry :thumbsup:

You went to extra effort to make sure we forum peeps knew about this, I appreciate that. If we had the ability to bump up a posters rep or cred, I'd +1 yours right now

 

[litwicki22]

ok i am buying new memory HyperX 2x4GB 1600MHz DDR3 Non-ECC CL9 DIMM Plug n Play (KHX1600C9D3P1K2/8G)

 

[sm625]

How did you explain this? 3 days ago i get 34 errors in memtest about 5 minutes after started test. Then i stopped test run again and after that i can run 24 hours of memtest without problem. So errors stopped of unknown reason. I dont change anything in bios. Its mobo, ram, psu? I want to change now memory but i am not sure.

It happens. I have an FPGA board that has 1GB of DDR3 onboard and it fails at 12 addresses intermittently. The addresses that fail are always the same but it doesnt always fail. Usually it only fails the first time I run the test after power up. Luckily today it is failing repeatedly.

 

[Corsair Joseph]

Thanks for the heads up VL

ok i am buying new memory HyperX 2x4GB 1600MHz DDR3 Non-ECC CL9 DIMM Plug n Play (KHX1600C9D3P1K2/8G)

^This or you could always take advantage of your lifetime warranty and get that RAM replaced via RMA :biggrin:

 

[ikachu]

I write memory initialization firmware for my current job, and I'd say the chances of this being a memory controller failure are very slim.

Chances are that replacing the RAM will fix the issue, but if it doesn't, my next guess would probably be that it's temperature related. The BIOS does memory training during boot, and if that's done when the DRAM is cold, and then the case/DRAM heats up a lot, then the signal delays chosen by the BIOS could be suboptimal. If there is another issue on top of that like a bad board layout leading to a lot of crosstalk, I could see it causing intermittent issues.

 

[Dave3000]

I want to share my experience in which I thought my system had defective memory / memory controller issue but it was not. I used to have an i7-3930k then I switched to an i7-4820k (wanted better ST performance) and then I started to have problems with Intel Burn Test making my system give me an error message saying that my PC has encountered a problem and must shutdown and then my PC shutdown on me once in a while even at stock settings. I also had a problem where my system would lockup in Memtest but the cursor was still blinking. At first I though it was a motherboard issue so I bought another Asus P9X79 Pro motherboard and still had the same issue with Intel Burn Test. Then I upgraded to an i7-4930k and 32GB of a different brand memory and my problems stopped with Intel Burn Test. I RMA'd my 4820k to Intel because I thought it could also been my CPU that was the issue and I also RMA'd my Corsair DDR3-1600 memory while I was RMA'ing my 4820k because I thought it was the memory as well and I already bought new memory. When my replacement CPU and memory arrived I tested again and same problem with Intel Burn Test. I removed the Corsair memory and reinstalled the Crucial memory and did not experience that issue while running IBT. I didn't not experience this issue with my Corsair memory with my 3930k or 3820 that I used to have. I also tested my replacement Corsair memory with my 4930k and it gave me the same issue with IBT as it did with the 4820k. Maybe the combination of my Corsair memory and IBE CPUs don't get along with each other since that same memory I did not have any issues with my 3820 and 3930k on the same motherboard?

 

[tynopik]

or you could go ECC and avoid most of this nonsense

 

[litwicki22]

Today i am going to buy new memory.

HyperX 2x4GB 1600MHz DDR3 Non-ECC CL9 DIMM Plug n Play (KHX1600C9D3P1K2/8G)

 

[litwicki22]

if I run memtest and get an multiple errors within a minute. Then run memtest the next night with the same ram, and run error free for 12+ hours, whats up with that?

Here is a log:
2015-04-18 16:14:35 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4DBF64, Expected: EFEFEFEF, Actual: EFFFEFEF
2015-04-18 16:14:36 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4DBF24, Expected: EFEFEFEF, Actual: EFCBEFEF
2015-04-18 16:14:37 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4DBC24, Expected: EFEFEFEF, Actual: EFF7EFEF
2015-04-18 16:14:39 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4DBB24, Expected: EFEFEFEF, Actual: EFEDEFEF
2015-04-18 16:14:40 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4DBAE4, Expected: EFEFEFEF, Actual: EFEBEFEF
2015-04-18 16:14:41 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4DB9A4, Expected: EFEFEFEF, Actual: EFCFEFEF
2015-04-18 16:14:42 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4DB464, Expected: EFEFEFEF, Actual: EFEEEFEF
2015-04-18 16:14:44 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4DB424, Expected: EFEFEFEF, Actual: EFCFEFEF
2015-04-18 16:14:45 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4DAFA4, Expected: EFEFEFEF, Actual: EF67EFEF
2015-04-18 16:14:46 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4DAA64, Expected: EFEFEFEF, Actual: EF77EFEF
2015-04-18 16:14:47 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4DAA24, Expected: EFEFEFEF, Actual: EFEDEFEF
2015-04-18 16:14:49 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4DA564, Expected: EFEFEFEF, Actual: EF25EFEF
2015-04-18 16:14:50 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4DA4E4, Expected: EFEFEFEF, Actual: EFE3EFEF
2015-04-18 16:14:51 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4DA0E4, Expected: EFEFEFEF, Actual: EFEBEFEF
2015-04-18 16:14:53 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4DA0A4, Expected: EFEFEFEF, Actual: EFE9EFEF
2015-04-18 16:14:54 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D9FE4, Expected: EFEFEFEF, Actual: EFE7EFEF
2015-04-18 16:14:55 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D9CA4, Expected: EFEFEFEF, Actual: EFE8EFEF
2015-04-18 16:14:56 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D98A4, Expected: EFEFEFEF, Actual: EFE1EFEF
2015-04-18 16:14:58 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D97E4, Expected: EFEFEFEF, Actual: EFDBEFEF
2015-04-18 16:14:59 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D9564, Expected: EFEFEFEF, Actual: EFECEFEF
2015-04-18 16:15:00 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D94E4, Expected: EFEFEFEF, Actual: EF8DEFEF
2015-04-18 16:15:01 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D8F64, Expected: EFEFEFEF, Actual: EFC7EFEF
2015-04-18 16:15:03 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D8F24, Expected: EFEFEFEF, Actual: EFFFEFEF
2015-04-18 16:15:04 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D8EA4, Expected: EFEFEFEF, Actual: EFE7EFEF
2015-04-18 16:15:05 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D8C64, Expected: EFEFEFEF, Actual: EFCFEFEF
2015-04-18 16:15:06 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D8C24, Expected: EFEFEFEF, Actual: EF6DEFEF
2015-04-18 16:15:08 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D8B24, Expected: EFEFEFEF, Actual: EFA7EFEF
2015-04-18 16:15:09 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D8AE4, Expected: EFEFEFEF, Actual: EFEDEFEF
2015-04-18 16:15:10 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D8824, Expected: EFEFEFEF, Actual: EF39EFEF
2015-04-18 16:15:12 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D8764, Expected: EFEFEFEF, Actual: EFBFEFEF
2015-04-18 16:15:13 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D8724, Expected: EFEFEFEF, Actual: EFCEEFEF
2015-04-18 16:15:14 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D86E4, Expected: EFEFEFEF, Actual: EFB7EFEF
2015-04-18 16:15:15 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D8364, Expected: EFEFEFEF, Actual: EFE7EFEF
2015-04-18 16:15:17 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D8324, Expected: EFEFEFEF, Actual: EF6EEFEF
2015-04-18 16:15:18 - [MEM ERROR - Data] Test: 4, CPU: 0, Address: 17E4D8064, Expected: EFEFEFEF, Actual: EF23EFEF


I am used Memtest86 v.6.0 ( without + )

http://www.memtest86.com/download.htm

 

[amenx]

Back in circles. You're basically repeating your OP question.

 

[KingFatty]

I'd suggest running memtest long enough to complete a pass, see how many errors, and then reboot the computer. Repeat this several hundred times and see if a pattern emerges.

Then, you should try another approach, by warming up the computer first, before you run memtest. Like use a CPU benchmarking program to get the CPU nice and warm, then immediately reboot and run memtest.

 

[VirtualLarry]

Based on that log file, it appears that perhaps, a single DRAM chip on your DIMM is flaky. I would replace or re-seat the RAM.

If it were the memory controller on the CPU, I would expect that the erroring addressing would be more randomly distributed. As it appears, they seem to be relatively localized.

It could be thermally-related too.

 

[Spanners]

You're not going to get an exact answer. IDC gave you the best explanation.


You're actually going to have to do work on this one. No-one from a forum will be able to give you an exact answer on what, if anything, needs to be replaced. Not without some exhaustive diagnostic work on your part.

Edit: For those that don't know, "litwicki22" is a known tech-support troll that likes to start multiple threads, gets advice, and repeatedly ignores the advice, and goes on to repeat the same question over and over again, and starts more threads.
http://forums.anandtech.com/showthread.php?t=2328724&highlight=litwicki22

Supposedly, his problem is fixed, according to his original thread in Computer Help.
http://forums.anandtech.com/showthread.php?t=2428588

But he'll just keep repeatedly asking "why", until the mods lock this thread, like so many others he's started.

Weird some kind of computer hardware Münchausen by proxy.