Strange RAM problem

Phandalyon

Guest
Joined
Jun 12, 2001
Messages
5,840
OK, so here is the deal. The computer I am working on has an ECS C19-A SLI Motherboard in it. I have 2 1GB sticks of Corsair DDR2 667. When I have both sticks in the computer in and configuration, I just see a flash of the POST and then a reboot loop over and over. I can not get into the BIOS. If I have just one stick of RAM in the board in any slot, the system boots fine (It doesn't matter which stick I use). Anyone have an idea whats going on?

I am just starting to troubleshoot this, but it is a weird issue, so I thought I would ask. Well, back to the grindstone on this one.

edit: Crap. Both sticks of RAM work fine in the system if I remove one of the Video cards. I have a 500W Antec PSU in the system but I am starting to suspect it may be the culprit.

edit2: OK. I have just tried a different PSU and I get the exact same results. If I have 2 sticks of RAM AND 2 Video cards plugged in the computer restarts during POST and I can not get into the BIOS even. If I remove 1 video card OR 1 stick of RAM, everything boots fine.
 

Phandalyon

Guest
Joined
Jun 12, 2001
Messages
5,840
The BIOS on the site is still the original version, so there have been no updates. One of the first things I checked and forgot to mention. I have the computer formatted and Windows installed with no issues with one of the Video Cards pulled. I am going to run Memtest for a while and see if I get any errors. Once that is done I am going to switch to the SLI setup with one stick of RAM and stress the video cards. It could be a weakness in one component maybe. But I doubt it. I have also sent an email off to ECS to see what they have to say. We shall see soon enough I suppose.
 

Phandalyon

Guest
Joined
Jun 12, 2001
Messages
5,840
OK. I have been running memtest on the system for about 4.5 hours now and it has had 1 error. Is it possible that this could be the problem. It seems like a fairly miniscule thing. I think I may replace the RAM anyway and see if that fixes the problem, but I don't know if it will.

It was 1 error on pass 50 in the LRAND test with cache on. It returned 390f6b94 instead of 390d6b94 at address 065ca9a4

Does anyone know if one error in memtest is enough to say that RAM is truly bad, or is it tiny enough to be a fluke?
 

EnFoRcEr!!

Limp Gawd
Joined
Jan 29, 2003
Messages
472
I can't really say for sure that it is a fluke with the memory. Normally that isn't the case. Although if the systems works fine when you take either of the RAM sticks or the vid cards, maybe its a problem with the board. But you also mentioned our Antec 500w, and it seems as if you have a dual video card system, so that could very well be the problem. The only way to know for sure is to try a 550 or better and see how it works.
 

Phandalyon

Guest
Joined
Jun 12, 2001
Messages
5,840
EnFoRcEr!! said:
I can't really say for sure that it is a fluke with the memory. Normally that isn't the case. Although if the systems works fine when you take either of the RAM sticks or the vid cards, maybe its a problem with the board. But you also mentioned our Antec 500w, and it seems as if you have a dual video card system, so that could very well be the problem. The only way to know for sure is to try a 550 or better and see how it works.

Well, I can do that. I have a OCZ Powerstream 560 in my system. I will give that a shot and see if it helps. These aren't highen video cards (GeForce 6600 256MB) so I don't think the power draw would be that great, but I will give it a shot and see. I need to pull my beast of a computer out to clean it anyway. The Antec PSU I got is an SLI compatible PSU, so I figured with the power requirements of this system it would be OK, but maybe it is not enough.

ECS wrote back to me today and they told me to check the RAM and update the BIOS. Of course the BIOS version is the only one available, there are no updates. Real helpful there.

Memtest is still running (on pass 76) and still has only one error.
 

Phandalyon

Guest
Joined
Jun 12, 2001
Messages
5,840
Well, Memtest ran for 36 hours with a grad total of 8 errors, all of them on the same test and all similar. So the RAM is going back one way or the other. I guess we will see if that fixes it when the replacement RAM arrives.
 
Joined
May 4, 2006
Messages
53
Five bucks says it's a noisy mobo design or a bad bios, and I'm leaning towards the mobo. From what you've posted, it looks like there's some kind of noise working through the northbridge circuitry and is likely crosstalk that only shows up under specific conditions. One of the least fun problems to diagnose and correct.

It's not at all unusual to see this - especially in low volume boards. First adopters always get screwed with the joy of sorting out the mistakes of the people who designed and built the parts.
 

Phandalyon

Guest
Joined
Jun 12, 2001
Messages
5,840
I am leaning toward it being a motherboard problem myself. I have been talking with ECS through email and so far they are useless. I am going to go ahead and replace the RAM to appease them and see where they want to go from there. If they don't fix it in a relatively short amount of time I am going to return the mobo, eat the 20% restocking fee, and order a better one.
 
Joined
May 4, 2006
Messages
53
Yeah... I hate to throw ECS under the bus but, in fairness, their stuff has never really impressed me and their service has always been in the 'you get what you pay for' camp.

It's still *possible* that you have bad memory...but IMO unlikely. Since the effect doesn't seem to differentiate when any 3 of the 4 card components are installed, mobo noise seems the likely culprit. The fourth node breaks the camel's back...

I can't tell you how miserable it is to try to track down problems like that when it's your job to fix them. :D
 

Phandalyon

Guest
Joined
Jun 12, 2001
Messages
5,840
That is the truth. I am not even sure how I ended up with this board, it is the first ECS product I have ever purchased. I think someone I know ordered the wrong board and I happened to need one like it so I bought it off of them to use in this system. It is actually a really well layed out board and it looks good, but if it doesn't perform then who cares.
 

Phandalyon

Guest
Joined
Jun 12, 2001
Messages
5,840
New RAM arrived with the same issue. Motherboard is apparently bad. Ordered a new one today.

Stupid ECS. Well, thats a lesson to me.
 

Phandalyon

Guest
Joined
Jun 12, 2001
Messages
5,840
New motherboard did not fix the issue...

I am pulling my hair out with this. I just can't believe that I am the only person who has encountered this issue.
 

Tigerbiten

[H]ard|DCer of the Month - February 2007/January 2
Joined
Sep 24, 2002
Messages
5,028
Is the MoBo a different brand ?

It sound like something is pulling a rail down as it posts hence the reboot.
Can you monitor the voltages via a multimeter as it boots ?

What other hardware is attached or is it a minimum build so far ?
If not minimum then what happens with a minimum build ?

Luck............... :D
 

12Bravo

n00b
Joined
Oct 17, 2003
Messages
30
I've had the same problem before with Corsair RAM on an Asus mobo... Try up'ing the voltage to your ram... while mine is supposed to be running at 2.75v it was defaulting to 2.65 before I changed it, and it fixed the prob in my system
 
Joined
May 4, 2006
Messages
53
Ph: Can you please define 'New Mobo' a little better? I'm guessing that you now have an Exact Replacement of the original... If so, it's not surprising that you're having the same problem. :D

Tiger's got a good point - I went back and re-read what you'd originally written about symptoms and think I may have misunderstood (had it in my mind that your system would get partway through the memory test section for some odd reason - oops) the time it takes to produce the reboot. Since you're only getting a brief flicker of a post screen, it does look like something else is going on.

Does it get through the video post screen and all the way to the bios post or you just see a flicker of the video post screen?

Since it sounds like you've got plenty of PSU juice, I'm starting to wonder about the initial charge-up load that those cards place on the system...Tiger may well have it nailed. ECS is mostly low-buck stuff and it wouldn't surprise me a bit if they cheaped out there - a bios change might could fix that though.
 

Phandalyon

Guest
Joined
Jun 12, 2001
Messages
5,840
Oh crap. I forgot about this thread. Sorry guys.

First of all, the issue is fixed. But I will answer some questions first:

It was a new motherboard from a different company. It was a Gigabyte.

I tried running the system with my PCZ Powerstram 520 (That I know can take a huge load) and it had the same issue.

It was a minimum build during troubleshooting (MoBo, RAM, CPU, Video Cards)

During the issue, I could not get into the buis at all, I would see a tiny flicker of the screen and then it would reboot.

OK, are you ready for the fix now?

Someone in General Mayhem (I linked to this thread there) suggested that I go through the BIOS with a fine tooth comb. I did so. Changing settings one at a time and testing then resetting to defaults. What eventually resolved the issue was a BIOS setting.

There was a setting called SLI Broadcast aperture that defults to disable. I set it to Auto instead (the only other option) and it fixed the problem.

On the ironic side though, once the problem was fixed I figured out that the original RAM I had in the computer was fine, but that the RAM I had bought to replace it was very very bad.

Someone in GenMay also answered my earlier question about RAM and told me that ECS motherboards will frequently test RAM as faulty due to flaws in the boards and that my earlier bad RAM tests were not actually an indication of BAD RAM, but of a cheap motherboard.
 
Top