Personal computing discussed

Moderators: renee, Flying Fox, Thresher

 
MatiasMGS
Gerbil
Topic Author
Posts: 11
Joined: Sun Nov 04, 2012 10:51 pm

Memtest86+ test failed with good working RAM and MOBO

Sun Nov 04, 2012 10:58 pm

Hi, about a week ago I bought new parts for my pc, and apparently it didn't go so smooth. At first I noticed BSODs when first using it, and I thought it was the HDD drivers reacting at the new motherboard, so I formatted the disc. When I try to install windows 7 (tried with x86 and x64) I get reboots and BSODs and the inability to install it. So I decide to run a Memtest86+ test. 300000 errors within the first minute approximately. I sent the ram sticks to my provider for testing: zero errors. Then I sent the motherboard for testing: running perfectly fine. Maybe the BIOS has a wrong ram config? Not a chance, I checked it myself. Maybe the motherboard isn't compatible with the RAM? Nope, according to the MOBO website, the mobo is compatible with the exact same model of the RAM. Maybe Memtest is throwing false positives for no reason and the windows installation disk is corrupted? Again, no; I Installed windows with the same USB drive with the same official Windows ISO image on my old computer, no problem. I also tried with one stick individually in every slot and no difference.

New parts I bought:

CPU: Intel i7 2600k
MOBO: Gigabyte GA-Z77X-UD3H
RAM: 2x AMD 4gb DDR3 1600MHz Entertainment Edition AE34G1609U2-U
PSU: Coolermaster Silent Pro 600w.

parts I kept:

HDD: WDC WD3200AAKS-00B3A0.

I was adviced in another forum to play a little bit with the ram timings and voltages. I tried with 9-9-9-27 1.55v cmd 2 and 9-9-9-24 1.50v cmd2. Still throwing errors...

Even though I have the latest BIOS, an XMP option doesn´t appear, so I assume that my ram doesn´t have it.

Thanks for reading !
 
Captain Ned
Global Moderator
Posts: 28704
Joined: Wed Jan 16, 2002 7:00 pm
Location: Vermont, USA

Re: Memtest86+ test failed with good working RAM and MOBO

Sun Nov 04, 2012 11:34 pm

RAM is cheap right now. Buy a couple of sticks and retry the test.

Memtest doesn't throw false positives.
What we have today is way too much pluribus and not enough unum.
 
UberGerbil
Grand Admiral Gerbil
Posts: 10368
Joined: Thu Jun 19, 2003 3:11 pm

Re: Memtest86+ test failed with good working RAM and MOBO

Sun Nov 04, 2012 11:49 pm

These days memory controller is in the CPU, and if it is bad you'd get a lot of errors. Swap it for a known good one and try again.
 
Cyco-Dude
Gerbil First Class
Posts: 130
Joined: Sun Jul 13, 2008 5:45 am

Re: Memtest86+ test failed with good working RAM and MOBO

Mon Nov 05, 2012 1:38 am

try 10-10-10-30 @ 1.6v (i'll assume you know what all those settings are?). if it still has problems with those relaxed timings and high voltage, get new ram. newegg has ram on sale all the time (i can post links to their promotions if you want). you can get 8gigs for $35.

if it's fine, you can start tightening up the timings (usually 9-10-9-30, 9-10-9-27) and then lower the voltage. some ram does not respond well to even timings like 9-9-9-27, and needs 9-10-9-27 to be stable (for example).
 
Krogoth
Emperor Gerbilius I
Posts: 6049
Joined: Tue Apr 15, 2003 3:20 pm
Location: somewhere on Core Prime
Contact:

Re: Memtest86+ test failed with good working RAM and MOBO

Mon Nov 05, 2012 2:01 am

Try running the memory at auto-settings and make sure that it is getting the correct voltage. You also try to play around with memory multiplier/divider. I had encounter a motherboard and memory combo that would refuse to work with memory at certain multiplier (400Mhz strap) even through both memory and motherboard are capable of handling it on paper. I decide to change it to another setting (300Mhz strap) and the problems went away.

Also try to play around with memory slots to see you can find anything here.
Gigabyte X670 AORUS-ELITE AX, Raphael 7950X, 2x16GiB of G.Skill TRIDENT DDR5-5600, Sapphire RX 6900XT, Seasonic GX-850 and Fractal Define 7 (W)
Ivy Bridge 3570K, 2x4GiB of G.Skill RIPSAW DDR3-1600, Gigabyte Z77X-UD3H, Corsair CX-750M V2, and PC-7B
 
MatiasMGS
Gerbil
Topic Author
Posts: 11
Joined: Sun Nov 04, 2012 10:51 pm

Re: Memtest86+ test failed with good working RAM and MOBO

Mon Nov 05, 2012 5:53 pm

1º: I live in Argentina and things here are pretty expensive. I can´t buy ram just like that.
2º: Increasing memory voltage higher than 1.575 volts will void my cpu warranty. We wouldn´t want that since CPU damage could be the cause of my problem.
3º: About decreasing memory frequency, if someone can confirm that doing this couldn´t damage my ram, I´ll try lowering it to 1333Mhz. Would I have to change the timings and voltage too?
4º: About CPU damage. If someone can confirm that fabric CPU damage is both probable and could be the cause of my problem, I´ll RMA it (The place it´s pretty far from home, so I have to be sure I won´t go for no reason :P)

Thanks for all the answers. I posted this problem in 3 other forums and I never got so much participating users :).

Edit: As for the last post, I tried with both modules individually in each and every slot, no difference.
 
nerdrage
Graphmaster Gerbil
Posts: 1314
Joined: Thu Mar 06, 2003 2:49 pm
Location: Raleigh, NC

Re: Memtest86+ test failed with good working RAM and MOBO

Mon Nov 05, 2012 6:36 pm

I'd say the odds of you having TWO bad memory sticks is rather low...

My advice would be to:
1. Turn off the PC and unplug the power
2. Install one stick of RAM in DDR slot 1 (according to your motherboard manual, it's the slot furthest away from your CPU)
3. Clear your CMOS using the Clear CMOS button
4. Go into BIOS screen and choose "Load Optimized Defaults", save and exit
5. Run Memtest

If you get errors, swap the RAM sticks.
If you don't get errors, add the second stick in DDR slot 3 and repeat steps 1,2,5.
i5-9600K · Z390 Aorus Pro · 16GB G.Skill DDR4-3200 · MSI TwinFrozr HD7850 · Intel 760p 512GB · WD 1TB · HP ZR24w · Corsair VX450 · Antec Solo · Win10/Ubuntu 18.10
 
MatiasMGS
Gerbil
Topic Author
Posts: 11
Joined: Sun Nov 04, 2012 10:51 pm

Re: Memtest86+ test failed with good working RAM and MOBO

Mon Nov 05, 2012 7:01 pm

nerdrage wrote:
I'd say the odds of you having TWO bad memory sticks is rather low...

My advice would be to:
1. Turn off the PC and unplug the power
2. Install one stick of RAM in DDR slot 1 (according to your motherboard manual, it's the slot furthest away from your CPU)
3. Clear your CMOS using the Clear CMOS button
4. Go into BIOS screen and choose "Load Optimized Defaults", save and exit
5. Run Memtest

If you get errors, swap the RAM sticks.
If you don't get errors, add the second stick in DDR slot 3 and repeat steps 1,2,5.


Ok, I did exactly as you said (even though I did the same thing a while ago, and I said it in the first post) and the 2 ram sticks throw errors. I also noted that I sent them to my provider for testing, they kept them running memtest for the whole weekend and it was clean.
 
cynan
Graphmaster Gerbil
Posts: 1160
Joined: Thu Feb 05, 2004 2:30 pm

Re: Memtest86+ test failed with good working RAM and MOBO

Mon Nov 05, 2012 8:18 pm

As others have said, it can either be the motherboard slots (and connecting traces) that are defective, the CPU or CPU socket, or the dimms themselves (assuming that you are running the correct voltage and are not overclocking the ram). There's also the small possibility of it being the power supply.

If each stick individually causes so many memtest errors, together with the fact that you have confirmation that they do not in another PC, then it's likely not the RAM. If you've tried these ram dimms separately and in different slots and get the same error rate, then it's likely not the motherboard either. This leaves the CPU. The other possibility might be that pins linking the memory controller to the ram slots on the CPU socket got slightly bent out of place when you installed the CPU and is not making a consistent contact.

Unfortunately the only way to really trouble shoot a problem like this, after you've already tried each individual ram dimm in different slots, is to swap with a known working part, starting with the least expensive/most likely of culprits, which would be the ram.
 
MatiasMGS
Gerbil
Topic Author
Posts: 11
Joined: Sun Nov 04, 2012 10:51 pm

Re: Memtest86+ test failed with good working RAM and MOBO

Mon Nov 05, 2012 8:36 pm

cynan wrote:
As others have said, it can either be the motherboard slots (and connecting traces) that are defective, the CPU or CPU socket, or the dimms themselves (assuming that you are running the correct voltage and are not overclocking the ram). There's also the small possibility of it being the power supply.

If each stick individually causes so many memtest errors, together with the fact that you have confirmation that they do not in another PC, then it's likely not the RAM. If you've tried these ram dimms separately and in different slots and get the same error rate, then it's likely not the motherboard either. This leaves the CPU. The other possibility might be that pins linking the memory controller to the ram slots on the CPU socket got slightly bent out of place when you installed the CPU and is not making a consistent contact.

Unfortunately the only way to really trouble shoot a problem like this, after you've already tried each individual ram dimm in different slots, is to swap with a known working part, starting with the least expensive/most likely of culprits, which would be the ram.


I ran my old pc with the same PSU for 2 weeks approximately, and it performed perfectly. Would that fact confirm that the PSU is fine?

It´s also not likely that the pins on the cpu socket got bent out. I was really, really, really careful when installing the cpu. I also sent the motherboard to my provider, and they said it was fine...

So, that leaves CPU, right?
 
MatiasMGS
Gerbil
Topic Author
Posts: 11
Joined: Sun Nov 04, 2012 10:51 pm

Re: Memtest86+ test failed with good working RAM and MOBO

Tue Nov 06, 2012 11:52 am

I just read that my CPU is supposed to handle up to ddr3-1333 ram. Mine runs at 1600mhz. Could that be the problem? Should I decrease the frequency? How should I proceed?

Edit: Ok I got more into it and looks like a lot of people has i7 2600k working with 1600mhz ram and higher. The odd thing is that the BIOS doesn´t automatically reduces the memory frequency, it stays at 1600mhz. Is that what is supposed to be doing? Shouldn´t it be running at 1333 auto? Everything is set to auto and there´s no XMP...
 
MatiasMGS
Gerbil
Topic Author
Posts: 11
Joined: Sun Nov 04, 2012 10:51 pm

Re: Memtest86+ test failed with good working RAM and MOBO

Wed Nov 07, 2012 5:52 pm

OK, I finally solved the problem. I changed the frequency multiplier so the memory runs at 1333mhz. Since I did that, Memtest doesn't throw a thing, and I could finally install windows. I have no idea why this works, I hope someone can explain to me.
Edit: I also wanted to thank Krogoth because he was the one who convinced me to try this out, and the only one in 4 different forums that told me to do the right thing.
 
Ryu Connor
Global Moderator
Posts: 4369
Joined: Thu Dec 27, 2001 7:00 pm
Location: Marietta, GA
Contact:

Re: Memtest86+ test failed with good working RAM and MOBO

Wed Nov 07, 2012 6:15 pm

The memory controller in your CPU just can't overclock much is the gist.

If you really want the 1600MHz you're going to have to pump more voltage into the memory controller. Going past 1.575 isn't recommended by Intel, but your mileage may vary.
All of my written content here on TR does not represent or reflect the views of my employer or any reasonable human being. All content and actions are my own.
 
MatiasMGS
Gerbil
Topic Author
Posts: 11
Joined: Sun Nov 04, 2012 10:51 pm

Re: Memtest86+ test failed with good working RAM and MOBO

Wed Nov 07, 2012 8:02 pm

Ryu Connor wrote:
The memory controller in your CPU just can't overclock much is the gist.

If you really want the 1600MHz you're going to have to pump more voltage into the memory controller. Going past 1.575 isn't recommended by Intel, but your mileage may vary.


Why didn't anyone tell me that before? I posted this on 4 different forums a while ago!
 
cynan
Graphmaster Gerbil
Posts: 1160
Joined: Thu Feb 05, 2004 2:30 pm

Re: Memtest86+ test failed with good working RAM and MOBO

Wed Nov 07, 2012 9:24 pm

MatiasMGS wrote:
Ryu Connor wrote:
The memory controller in your CPU just can't overclock much is the gist.

If you really want the 1600MHz you're going to have to pump more voltage into the memory controller. Going past 1.575 isn't recommended by Intel, but your mileage may vary.


Why didn't anyone tell me that before? I posted this on 4 different forums a while ago!


Well, it depends on the individual processor, motherboard, ram, and combinations of the 3. I bet most here probably assumed that ram rated for 1600 would likely run fine in most scenarios. But sometimes, due to the memory controller on your specific CPU and combination of motherboard and ram, you can be unlucky.

For example, I have a processor and ram that many people seem to have not too much difficulty overclocking to well over 1600Mhz for the ram. However, no matter what settings I play with, or voltage I give the memory controller or ram, I just can't get it to run stable at anything over 1600Mhz...

It could be that the ram you have just isn't very suitable for the processor you have. Some DDR3 chips run better with different memory controllers, etc. When choosing ram, it's safest to do a bit of research to make sure that others have had luck using it with the same processor/motherboard chipset.

Did you try lowering memory timings to 10-10-10-30 and running at 1600Mhz?

Good to hear you pinpointed the source of the Memtest errors though :D
 
MatiasMGS
Gerbil
Topic Author
Posts: 11
Joined: Sun Nov 04, 2012 10:51 pm

Re: Memtest86+ test failed with good working RAM and MOBO

Wed Nov 07, 2012 10:15 pm

cynan wrote:
Did you try lowering memory timings to 10-10-10-30 and running at 1600Mhz?

I'll try that tomorrow, thanks for the explanation. My guess is this happened because AMD RAM isn't really meant for Intel platforms?

cynan wrote:
Good to hear you pinpointed the source of the Memtest errors though :D

Yes, and more importantly I could finally install windows and try out how BF3 runs :lol: . Runs pretty good even with my dated Radeon 4870
 
Flying Fox
Gerbil God
Posts: 25690
Joined: Mon May 24, 2004 2:19 am
Contact:

Re: Memtest86+ test failed with good working RAM and MOBO

Wed Nov 07, 2012 10:25 pm

MatiasMGS wrote:
cynan wrote:
Did you try lowering memory timings to 10-10-10-30 and running at 1600Mhz?

I'll try that tomorrow, thanks for the explanation. My guess is this happened because AMD RAM isn't really meant for Intel platforms?
It probably just means they tested the RAM only on AMD platforms. So the fact that it is not working on Intel platforms, and on the somewhat unofficial speed of 1600MHz, is just coincidental.
The Model M is not for the faint of heart. You either like them or hate them.

Gerbils unite! Fold for UnitedGerbilNation, team 2630.
 
MatiasMGS
Gerbil
Topic Author
Posts: 11
Joined: Sun Nov 04, 2012 10:51 pm

Re: Memtest86+ test failed with good working RAM and MOBO

Sun Nov 18, 2012 2:14 am

It doesn't work with the 10-10-10-30 timings either. I posted the problem on another forum.
 
MatiasMGS
Gerbil
Topic Author
Posts: 11
Joined: Sun Nov 04, 2012 10:51 pm

Re: Memtest86+ test failed with good working RAM and MOBO

Fri Nov 30, 2012 11:06 pm

They refuse to help in the other forum. What do you guys reccomend? I already tried this:
Try command rate 2T, tRFC to 300ish and tWCL (or tWL, whatever it's called on Gigabyte) to 8. If this doesn't help, then it gets down to adjusting other subtimings via trial and error method.
 
Ryu Connor
Global Moderator
Posts: 4369
Joined: Thu Dec 27, 2001 7:00 pm
Location: Marietta, GA
Contact:

Re: Memtest86+ test failed with good working RAM and MOBO

Fri Nov 30, 2012 11:29 pm

Why does helping you make me feel like a battered wife?

My old Nehalem i7-940 couldn't do 1600MHz even with RAM rated for 1866MHz. It didn't matter how much voltage or how relaxed the timings were. It was simply beyond the switching speed of the memory controller transistors.

You've unfortunately won the lottery, which is to say you've gotten a "choice" chip from the very edge of the wafer. My suggestions is to select 1333MHz and be pleased. 1333MHz is the proper rated speed for Sandy Bridge.

If you wanted 1600MHz you should have bought an Ivy Bridge. Given the amount of time invested in trying to make the 2600K do 1600MHz factored against your average wage per hour, you probably have already made up any difference in price in frustration and failure.

The extra MHz won't even make a reasonable difference in anything except synthetic benchmarks. Just enjoy the PC as it is.
All of my written content here on TR does not represent or reflect the views of my employer or any reasonable human being. All content and actions are my own.
 
MatiasMGS
Gerbil
Topic Author
Posts: 11
Joined: Sun Nov 04, 2012 10:51 pm

Re: Memtest86+ test failed with good working RAM and MOBO

Sat Dec 01, 2012 7:55 pm

Guess I have no choice. I really regret buying that RAM. I should have bought at least a RAM with a heat spreader, so at least it'd look somewhat cool :lol:. Thanks for the answers.
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: Memtest86+ test failed with good working RAM and MOBO

Sat Dec 01, 2012 8:24 pm

At least RAM isn't terribly expensive these days so it is no great loss.

It was probably a bit of a stretch to expect this RAM to work reliably in this scenario anyway. Entertainment Edition is AMD's lowest cost RAM, and they specifically test it for compatibility with AMD systems. It should also meet all relevant DDR3 specifications, but that is no guarantee that it will work with an overclocked Intel memory controller.
Nostalgia isn't what it used to be.

Who is online

Users browsing this forum: No registered users and 1 guest
GZIP: On