memory problem

I have tried to search and find an answer, but I haven't, so here is my question. I have 2 sticks of Geil Golden Dragone PC 3500 ram in my Abit NF7-S, they overclock fine, and I am getting prime stabiliity 18+ hours (had to stop it). But I am getting a ton of memory errors in games, so I ran memtest86. In dual channel mode I get thousands of errors in test 5, but with either stick seperately, or both in single channel I pass fine. I tried lowering the FSB in dual channel but even at 200 I get thousands of errors. Mt next though was to try disabling CPU Interface, but for some reason my board won't let me do that, it refuses to boot, until I reinable it. I've tried every setting I could think of to disable it. Right now I am Memtest and prime stable at 215 FSB single channel. Does anyone know why I can't get dual channel to work, or know why CPU Interface is acting like this? Thanks for all your guy's help in advance.

-Ir0nman

What is your vdimm ? What timings are you using ?

my Vdimm is 2.9, vdd is 1.7, I tried it with the laxest timings in the bios and I still got errors art even 200

I've noticed a similar problem with the CPU interface and can't suggest a solution. My board seems cantankerous in this regard and is allowing me to reset CPU interface right now. I really don't see improved OCing when it is off which is different that what most folks report. Also DC does put more stress on the memory and mobo so the behavior you report for single stick versus DC isn't unusual.

Memtest86 problems can also be associated with CPU errors, you can't tell the difference. Mushkin has a summary dealing with the issue on their web site. You might also crank up the voltage to the cpu to see if that helps. With reasonable water cooling you could probably run Vcpu at 2.0V but make sure your temps stay reasonable. Alternatively would be to crank you multiplier down to see if the errors disappear.

Also if you don't already, run only test 5 in Memtest to more quickly see any problems.

Thanks for the info, I have just been runnign test 5, and my CPU hasn't been that high in my tests, granted I should use a lower pultiplier, but I keept it below 2.0 Ghz when I was testing. still not sure why DC fails even at 400FSB with my mem mobo combo.... one time I got like 10000 errors, I have no idea though. I've been running in single channel for all day today flawlessly. so for now at least it seems my computer is usable. Can you please give me a link to the article on Mushkins site. Thanks

*edit* Only reason I was trying to disable CPU interface is because I was hoping that something was wrong there, and that I could solve my DC problems that way

-Ir0nman

Ir0nman said:

Can you please give me a link to the article on Mushkins site.

Click to expand...

here

MEMTEST86

Have you read about the different tests?

TEST #5 [Block move, 64 moves, cached]

This test stresses memory by using block move (movsl) instructions and is based on Robert Redelmeier's burnBX test. Memory is initialized with shifting patterns that are inverted every 8 bytes. Then 4mb blocks of memory are moved around using the movsl instruction. After the moves are completed the data patterns are checked. Because the data is checked only after the memory moves are completed it is not possible to know where the error occurred. The addresses reported are only for where the bad pattern was found. Since the moves are constrained to a 8mb segment of memory the failing address will always be less than 8mb away from the reported address. Errors from this test are not used to calculate BadRAM patterns.

TEST #8 [Block move, 512 moves, cached]

This is the first extended test. This is the same as test #5 except that we do more memory moves before checking memory. Errors from this test are not used to calculate BadRAM patterns.

It isnt an answer to your problem but it atleast rules some issues out.

Thanks for all the help so far, still trying some stuff.... I'll find an answer sometime I guess...lol

-Ir0nman

Maybe time to back off and start over again, i.e. determine limit of chipset and cpu separately. Can't do the same for the memory since overclocking of memory alone doesn't work or at least didn't on my Abit boards.

By the way one of my NF7 boards was real picky about the DC memory slots used. Either 1 and 3 or 2 and 3 should produce dual channel, test both combos if you haven't already. Slots 2 and 3 were much better if I remember correctly on my boards.

I was one step ahead of you on that, tried all the slot configurations, and still couldn't get it stable at even 200 :-(
it shouldn't be the chipset, but I am going to try lapping the NB sink, and putting some Ceramique on it.

-Ir0nman

Sounds like you're ahead of us on most things to try. I'm out of suggestions so can only wish ya good luck at this point.

Thanks deeppow, I'll try to keep you guys informed if I have any luck

-Ir0nman

ironman it's the memory for sure

not that it's bad/diffective memory, it just isn't good for overclocking even when you are keeping it withint specs(200mhz...)

Thanks, I am thinking on a mem volt mod... just not sure if I want too yet. Guess I'm gonna have to to get this fixed.

-Ir0nman

try updating to the newest bios, sometimes it fixes problems like that, also abit is picky about memory to O/c with. I know people who use corsair have a problem similar to this. I use mushkin, i know a couple of other people who do too. You should try different memory first, after the bios upgrade,
Not sure if this helps, but i figured i would say something

It could be a single stick going bad,I had the same problem with my setup.At first it would only fail test 5 in dual channel but allways at the same address.after some very extensive testing i found that with voltage at the lowest setting it would only fail on occasion(all testing was done at 200 fsb) but any voltage increase would start errors like a downhill snowball.Either stick would pass if the other was removed.After several days i decided to check each in slot 1 and after 3 hours 1 stick began to fail tests 3,5 and 8.After that it wouldn't pass a single run and allways at the same addresses.I advance RMA'ed it to memoryx and the new stick seems fine even though it isn't identical to the other.I beleive it was going bad to start with but the lengthy testing did it in alltogether.Dealing with the disk corruption it left behind is a whole other story.I have reinstalled a few times since then but have several patitions that i needed to keep and appearently the data corruption was spread around pretty good.
You might also want to run simms testor,it freezes less often and really puts some heat on em after several hours.