DFI NF4 Ultra-D & 3200XL memtest errors

Hey all.

 

I just built this system about a week and a half ago, but for the life of me i can't get memtest to run cleanly without errors on my system. I'm using optimized defaults and the memory is on auto-detect at 2-3-2-5@2.7v I've tried upping the voltage up to 2.9v, but ended up getting more errors on test #8 than i did at 2.7v Test #8 seems to bring out the worst in the memory as i haven't had it run past 2 successfull passes without error.

 

On a side note, when i initially installed my OS, i had to only use 1 stick of ram, otherwise i would just get BSOD's everytime. I'd just like to get this thing running stable and make sure i don't have defective memory.

 

Thanks for your help in advance!

Ok, I ran the tests on each stick by itself at the settings you said. One 512MB stick performed 4 passes on memtest without errors and the other stick performed 5 passes on memtest with 1 error (ran an extra one because of the error.) Either way the results were much better, but here's the problem:

 

I thought i'd throw both sticks of ram in to see if they worked on those settings also, and after the first pass of memtest, it brought up 13,000 errors.

 

Also, my temperatures jumped up to 52C on the chipset, 55C on PWM and 48C on the CPU and the fans were screaming like a wind tunnel. So i'm feeling like i've hit a dead end. Any ideas?

Well, here's an update:

 

I went ahead and ran prime95 overnight to test my system stability. I let it run for about 7 hours overnight, then in the morning i checked it and it was still going fine. I come back 4 hours later and i moved the mouse to turn on the monitor (have it on power savings) and the system just reboots automatically.

 

So it starts rebooting and it makes it into the login screen at windows and once I login, it brings up my wallpaper then shows a bluescreen for a split-second and restarts. It did this 3 or 4 times until i went into bios and loaded optimized defaults again and changed the dram voltage to 2.7. I checked the temperatures right after the first reboot happened and they were normal at around 44C.

 

Do you have any other suggestions before i send the ram in for replacement? I just bought these a couple weeks ago so I can send them back to the manufacturer, but I'd be without a working computer :laughing: I do have the latest bios, but should i perhaps try reflashing it again? I just find it weird for the memory to go nearly error free by itself then into dual channel mode where it blows up with 13000 errors.

 

Thanks alot for your help!

Sounds like you've tested them in dual channel together, but what about single channel together?

 

I'd personally lean more towards the memory controller (on the CPU) having a problem, which could also explain the higher temperatures.

the memory controller is on the CPU or the motherboard? Also, how exactly would that affect my temperatures?

 

I ran the tests in dual channel with both sticks in and not single channel. I'll have to give that a try. Unfortunately I don't know anyone with an AMD64 bit system that i could really test my ram on at the moment. I'm away at college now and it would kinda suck to have to send stuff back, so i'm trying to make sure I have all areas covered and I know it's defective. My cpu is running at 34C atm, which isn't really that high, it's just my PWM area that i can't seem to get below 40C.

 

Do you think i should reflash to the latest bios just to make sure everything was updated?

 

EDIT: At home, I have my old system which was an AMD XP2600+ with an ASUS a7n8x mobo that i'm 99% sure is dual-channel. Would that be sufficient for testing even though it doesn't have the juice of the AMD64?

On Athlon64 systems the memory controller is part of the CPU. On AthlonXP systems the memory controller is in the chipset. I'd test it in single first on your board, then in the older system.

hmm.. for some reason I couldn't get my system to boot the memory in single channel mode. I set the ram in dimm1+2 and it wouldn't boot up, also tried the pair in dimm3+4 with no luck. With the ram in dimm1+3 or 2+4 it was in dual channel, but for some reason I just can't get it to bootup with both running in single channel.

 

Am I doing something wrong here?

 

edit: also is there any specific way I could check to see if the memory controller on the cpu is bad?

ok thanks, i'll have to give that a shot after i get back from work. By running the modules in single channel, I'll be testing the memory controller to see if it has trouble with dual channel?

 

I'm guessing the best way is to test them in another system, but even though that system will be only running the chips at 266fsb, will that really be testing the integrity of the ram since it will be quite a performance decrease? It's an AMD XP2600+ but i can't remember if that's 266fsb or 333 fsb. hehe

 

I really appreciate the help you guys have given me. I'm just trying to understand exactly what i'm testing for.

By running the modules in single channel, I'll be testing the memory controller to see if it has trouble with dual channel?

Correctamundo.

 

I'm guessing the best way is to test them in another system, but even though that system will be only running the chips at 266fsb, will that really be testing the integrity of the ram since it will be quite a performance decrease? It's an AMD XP2600+ but i can't remember if that's 266fsb or 333 fsb. hehe

XP 2600+ has a 333 Mhz FSB. If the memory truly is bad, it'll still fail at a lower FSB.

Ok, I've been thoroughly running memtest for the past 36 hours to see if I could shed anymore light on the situation.

 

I ran 4 tests:

2 tests for each module by itself

1 test with both modules running in single channel

1 test with both modules running in dual channel

 

Each test consisted of 8 hours running the normal default test selection and then an additional 4 hours of running test 8 loops. Posted screenshots of each test are below.

 

Test results:

 

Module 1

Module 2

Both Modules - Single Channel

Both Modules - Dual Channel

 

As you can see module 1 had absolutely no errors like the first time I had tested it alone before. Module 2, however, had 4 errors durring the default test selection, but no errors from the test 8 loop. I was surprised that no errors happened at all with both modules running in single channel since module 2 generated a few errors by itself.

 

I didn't let the dual-channel test run the full 12 hour length because it didn't even make it through one pass without spitting out errors as usual. I'm still planning on testing the memory regardless on another system next weekend, but this seems to be looking more and more like a faulty memory controller.

 

What do you all think?

It's odd that Module 2 got errors at all considering it passed in single channel mode, but it could be a fluke. Testing the memory on another system should tie it all up.

are there any other tests i should consider running on this system to figure out the problem?

Testing in another system is really the only choice at this point.

in all likelyhood, could the motherboard have anything to do with these problems?

Possibly. Testing one stick in each slot would solve that.

Thought I might horn in on this post as I am having the same problems with the same mb and memory and an amd 53fx. I have not tried anything but dual channel. My symtoms are the same. I have found I can run stable with one stick in the dual channel slot. I'll keep watching

Thought I would paste a link I was following at DFI.

 

http://www.dfi-street.com/forum/showthread.php?t=8377

Well guys, I just got in my replacement 3200XL modules and popped em in and once again I got errors within a few minutes. I'm testing the new pair one at a time again just to cover all areas, but this is pretty frustrating. Apparently i'm not the only one having trouble running dual-channel mode on this ram with the DFI NF4 Ultra-D, so I dont' believe the integrity of the ram is the issue. I also don't feel the need to send back my cpu unless there is fairly certain evidence that it is defective.

 

My overall experience with this ram has been less than satisfactory to say the least, especially since this was recommended off corsair's website for my motherboard. Something needs to be done about these issues because the XMS3200XL isn't performing anything like it's hyped up to be, nor is it stable.

 

What other options do I have to consider now?

Re-doing the tests will help. One stick at a time in the orange slot farthest from the CPU (as per DFI), then single channel with 2 sticks, then 2 in dual channel. In DC, are you in slots 1 and 3 (which per their manual needs to be @ 2T), or slots 2 and 4 (which can be set to 1T)? BTW, which version of memtest were you using again? I think that board has memtest built into the BIOS (not 100% sure), try the latest version of memtest86+ from a boot disk if you're not already doing it. Don't try voltages above 2.9 like you mentioned in the following link, as that voids the memory warranty.

 

BTW:

http://www.dfi-street.com/forum/showthread.php?t=7919

 

It doesn't seem to be a compatibility issue since there are people reporting this combo works together fine, although you already know that since you posted in that thread :)

I'm using the orange slots (2 + 4) for my testing at 1T 2-2-2-5@2.8v I already tested the new set and got errors right off the start. I'm using the built in memtest v1.51 I believe. I'm downloading the latest memtest version now like you recommended. Also, how many passes of memtest do you think is sufficient? I've had one module of memory going for 2.5 hours and has passed 10 times without errors.