memtest86 Errors, what do they mean?

sinerasis

Junior Member
Mar 10, 2012
8
0
0
Troubleshooting my problematic HTPC here, basic specs:

Intel C2D E8400 (dual core, 3ghz)
BFG 680i SLI motherboard
ATI 2600HD graphics
8GB Ballistix DDR2 800mhz (4x2GB)
Ceton Infinitv4 pcie
Single WD Caviar Blue 160GB system drive (all storage drives have been disconnected trying to isolate the problem)
LiteOn? Blu-ray/DVD/CD
Corsair 520w power supply
Win7x64 Pro

I'm getting system lockups, no blue screens. The image will stay on the screen (Panasonic Plasma 720p). The sound will go glitchy. Hard reset is required at that point.

I've been chasing demons. I originally thought there was an issue with the Ceton card, problems started up about the time I upgraded its firmware. Lockups happen fairly quickly when the TV tuner is active (longest it goes without lockup is about 1 hour). Computer will idle all day without the tuner active. Have contacted Ceton support (they've been awesome by the way). I've downgraded firmware back to the old version, upgraded to a new beta version, sent tons of diagnostic logs to them. Problem remains.

Power supply was swapped, current supply is known good. Motherboard was swapped for identical model (I had two of them). I was starting to think there was a problem with the graphics card. I wanted to upgrade it anyway so I have placed an order for a new one. While playing around tonight I decided to run some torture tests to see if I could replicate the lockup without running the TV tuner.

Ran several quick Furmark test, all came back normal. I played with the settings each time and let it go for about 10 minutes each time. Temp maxed out at 90C. Seemed stable, no issues, no errors reported. (High temps is because this is a fanless video card, never had any problems with high temps on it)

I ran all three Prime95 tests. No errors. I let each test run for about 10 minutes. Max CPU temp was about 50C (water cooled rig, so no worries there).

Decided I'd check the memory with Memtest86 and low and behold I found a new trail to explore. I've gotten failures on test 3-8 so far. I think I'm looking at two different failed addresses that keep popping up:

8385.2MB and 4609.3MB

Do these values tell me anything in particular? Am I looking at the 2nd and 4th stick of RAM that have bad parts, or do I just have to guess and check to find the bad stick(s)?

So happy to have found SOMETHING... any help appreciated. :)
 

bononos

Diamond Member
Aug 21, 2011
3,923
181
106
2 & 4 sounds likely. In any case, memtest shouldn't take long to rerun if you need to swap dimms. Memtest isn't that intensive so if you find errors, its pretty bad.
 

BrightCandle

Diamond Member
Mar 15, 2007
4,762
0
76
if you are failing on test 3 that is really bad, the normal edge case tests are 4 and 7. Anything outside that the failure rate is pretty high. It could just be the settings for the RAM in your bios are selected wrong, wouldn't be the first time the bios read the SPD wrong or set the wrong RAM clock speed automatically.
 

Compman55

Golden Member
Feb 14, 2010
1,241
0
76
Can you borrow different ram? Or take all sticks out and test them single? Crucial ballistix ram is very fast, but can be absolutly junk for durability. They are not like they used to be.
 

Charlie98

Diamond Member
Nov 6, 2011
6,294
64
91
If it's a question of RAM, you'd have to test each RAM stick separately, and then in each slot (if it's a question of the RAM slots being bad.

I would tend to agree with what BrightCandle said as well... my system was reading my RAM incorrectly too, and I had to go in and set it manually.
 
Last edited:

BrightCandle

Diamond Member
Mar 15, 2007
4,762
0
76
If it's a question of RAM, you'd have to test each RAM stick separately, and then in each slot (if it's a question of the RAM slots being bad.

I would tend to agree with what BrightCandle said as well... my system was reading my RAM incorrectly too, and I had to go in and set it manually.

My Asus X79 Pro keeps setting the RAM command rate to 1T instead of 2T when its at 1600 whenever I change anything to do with the overclocking settings. It causes about 1 crash a day and then I realise what I forgot to do, which is change a setting that had absolutely nothing to do with why I entered the bios in the first place.

I want to find the Asus engineer responsible so I can ram this board up his behind!
 

KingFatty

Diamond Member
Dec 29, 2010
3,034
1
81
Check the manufacturer information on your RAM, and make sure you are using the right voltage etc. Maybe your computer defaulted to a voltage that was too low, or maybe the speed or something is out of whack.
 

Charlie98

Diamond Member
Nov 6, 2011
6,294
64
91
Awesome, I didn't even think to check the memory settings in BIOS. Good suggestions.

You can use CPUz to check what the readings are, but you'll have to go into the BIOS to verify some of the other settings and timings.

I want to find the Asus engineer responsible so I can ram this board up his behind!

...so you can RAM it... :biggrin: (I crack myself up...)
 

murphyc

Senior member
Apr 7, 2012
235
0
0
FWIW you will test more memory by booting a linux LiveCD (or rescue CD or USB stick) and running Memtest86+ from the troubleshooting menu, because the system and app will run in an insignificant amount of RAM, rather than booting a full operating system and then running memtest.
 

Elixer

Lifer
May 7, 2002
10,371
762
126
FWIW you will test more memory by booting a linux LiveCD (or rescue CD or USB stick) and running Memtest86+ from the troubleshooting menu, because the system and app will run in an insignificant amount of RAM, rather than booting a full operating system and then running memtest.

o_O
Memtest86+ can't be run on top of anything else (well, I suppose you could use a OS, then do a VM, then use memtest86+, but that would be nuts).
 

murphyc

Senior member
Apr 7, 2012
235
0
0
o_O
Memtest86+ can't be run on top of anything else (well, I suppose you could use a OS, then do a VM, then use memtest86+, but that would be nuts).

I happen to have a compiled version on Mac OS X that runs on the command line. I'm reasonably sure it can be run on Windows as well.
 

sinerasis

Junior Member
Mar 10, 2012
8
0
0
I think there is a Windows app version, but I'm just using the disc image available on the memtest86 website (burns as a live cd, which I'm fairly certain is linux based).

...at any rate... I checked the BIOS out, voltate and timings were auto detected incorrectly. Voltage was particularly low (detected at 1.8v, memory spec is 2.2v).

With the correct settings I re-ran memtest86 and still got the same errors. I ran through each stick individually in slot 1 and ran through test 5. Only 1 stick showed errors, again starting with test 3.

I re-ran the through test 5 with a stick in slot 1 and 2 with no errors, then with a stick in slot 1, 2 and 3 with no errors.

So it's looking like one bad stick at this point. I'm gonna try to watch lots of TV tonight (super hard grueling work I know, I might need to take regular beer breaks) and hopefully I won't get any lockups. If all goes well tonight I'll see if I bring myself to order a replacement stick or two at the newly inflated DDR2 prices :colbert:

Thanks for the help guys! :)
 

murphyc

Senior member
Apr 7, 2012
235
0
0
a.) Warranty. It's usually lifetime on decent memory.
b.) Run memtest86+ overnight. It shouldn't fail any of the tests, and they get more aggressive. And I've seen memory fail a test on the 3rd pass, but not on pass 1 or 2.
 

serpretetsky

Senior member
Jan 7, 2012
642
26
101
If it's a question of RAM, you'd have to test each RAM stick separately, and then in each slot (if it's a question of the RAM slots being bad.

I would tend to agree with what BrightCandle said as well... my system was reading my RAM incorrectly too, and I had to go in and set it manually.
this is what i would do too.

also, when you are using prime95, 10 minutes is not enough, and you have to enable "round off error checking" under the advanced toolbar. I usually run a min of 8 hours, and for memory testing i use "blend test". I have actually had memory pass everything (including memtest86) but fail prime95 blend test.
 

bryanl

Golden Member
Oct 15, 2006
1,157
8
81
Memtest86+ and Memtest are different programs. Memtest is a Windows application from HCI Design and has some fanboys who say it's the best, without offering proof, of course.

Very old Crucial Ballistix memory rated for 2.20V and is known to fail over time, but even newer Ballistix DDR2 rated for 2.00V is less reliable than Crucial's regular DDR2 and should be operated at 2.00V, rather than at the default 1.80V for DDR2, to minimize errors.
 

Seeg88

Junior Member
Mar 10, 2013
3
0
0
hello everyone..

lately i've been getting some blue screens with error msg of MEMORY MANAGEMENT. I have downloaded and burned Memtest86 image to a cd and ran it on start up overnight. Here are some specs on my PC

HP Pavilion Elite M9350F
Phemon4 9850
6 GB of RAM (slots 1&3 are 2GM, slots 2&4 are 1GB), DDR2 6400mhz

the results of the Memtest, as I am looking at them now, basically are showing failures in the 3243.5 to 3303.5 MB range. All of these failures occur in Test 5, Pass 6. If I am reading this correctly, it appears that stick 2 is going bad, Can someone please confirm this for me?
 

murphyc

Senior member
Apr 7, 2012
235
0
0
HP Pavilion Elite M9350F
Phemon4 9850
6 GB of RAM (slots 1&3 are 2GM, slots 2&4 are 1GB), DDR2 6400mhz

The RAM is from whom?

the results of the Memtest

memtest86+ 4.20? You can get bogus errors using the wrong version of memtest for your system/CPU.

it appears that stick 2 is going bad, Can someone please confirm this for me?

With one test there isn't a way to know if it's the slot or the stick. I'd removal all RAM except the suspect stick and retest. If it doesn't fail, you've guessed the wrong stick but at least that stick and slot are OK. If it fails it's either the slot or stick, so move the stick and retest. If if doesn't fail, it's the slot. If it does fail, it's the stick.

If you end up with a contradictory results, i.e. test passes on individual sticks but start to fail with multiple sticks, then it sounds like a voltage issue which could be BIOS settings, or RAM that's not actual to spec. So you have a lot of testing to do, which is why I ask where you got the RAM. If it's HP RAM, and the system is newish, I'd call up HP and tell them it's failing memory tests with memtest86+ 4.20 and see if they'll swap out all new RAM so you can reduce your testing. I mean, there's got to be a reason why you'd pay more for brand name memory, and I'd consider this one of them. When you buy really cheap RAM, you get to test and troubleshoot it.