• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Memtest errors

P

pay

Golden Member
I recently built a new computer that features an E6550, Gigabyte P35 board, and 2x 1GB of Adata PC6400 memory. The CPU is clocked at 450*7 = 3150 mhz and it will run Orthos all night without a hitch. However, in memtest, it fails test #2 within 10 seconds. Do I need to be concerned? The memory is running a 2x divider for 900 mhz. I realize the memory is rated at 800, but I have a feeling it would fail even at stock speeds. The memory is running at 2.1v with relaxed timings of 6-6-6-21.

Memtest is a very sensitive program. I remember on my old AMD rig, even with my OCZ ram that was running at stock speeds, it would still fail at a similar point in test 2.
 
Try it @ 800 and see if it still errors. I also think I read somewhere about possible false fail in memtest certain versions on certain tests but I don't recall where.
 
I agree with cprince - if it fails at stock speed in memtest, then RMA it.

Orthos grinds through a fairly small set of data over and over - if it doesn't hit the section that memtest is failing, you wouldn't see it (like the part that fails is way in the back of the memory, you wouldn't hit it until you'd filled up most of the 2GB). Similarly, memtest scans various patterns through... like 1010 followed by 0101. A failure in a particularly bad pattern wouldn't fail unless you tried that exact pattern in that exact spot... which would take a while to hit. But still, a failure is a failure whether it's hard to hit it or not. All you need is to be playing Crysis and get to a large level and have it crash because you hit that one bad spot in memory, or scan in a massive photo into a scanner and then have it fail. If memtest says it's broken, odds are it will come back to haunt you sooner or later.

If you want my advice, I'd RMA it back ASAP.
 
btw, if you're running linux, you don't need to throw away the ram. just tell the kernel what areas of ram to avoid with the badram patch.
 
Memtest tests more than just memory. Try setting both the cpu & memory to stock and run Memtest again.
 
Originally posted by: pay
I recently built a new computer that features an E6550, Gigabyte P35 board, and 2x 1GB of Adata PC6400 memory. The CPU is clocked at 450*7 = 3150 mhz and it will run Orthos all night without a hitch. However, in memtest, it fails test #2 within 10 seconds. Do I need to be concerned? The memory is running a 2x divider for 900 mhz. I realize the memory is rated at 800, but I have a feeling it would fail even at stock speeds. The memory is running at 2.1v with relaxed timings of 6-6-6-21.

Memtest is a very sensitive program. I remember on my old AMD rig, even with my OCZ ram that was running at stock speeds, it would still fail at a similar point in test 2.
Click to expand...

In addition to John's excellent comment regarding testing both CPU & Ram at stock speeds/voltages I would like to comment on your ram voltage from my experiences.

I too was running overclocked ram, at decent overvoltaged (>2V) ram. What I found was that just like overclocking/overvolting a CPU will seem stable when idle but become unstable once it begins to be used...ram was the same way.

I could loop thru memtest+ a couple times with no problems with my ram at 2.15V but the longer I left it looping (the hotter they got) then the more and more errors would crop up in memtest. If I let the ram cool back down then they would "pass" memtest+ without issue until they got hot again.

I confirmed this with a thermocouple, my ram would operate fine so long as they remained <42C at my point of measure (who knows how hot the IC's actually were). Adding a couple 80mm fans, simply zip-tied to the ram clips, kept my ram temps below the magical threshold and they have been stable for >1year now under heavy utlization.

Whenever I see something post about memtest+ errors and they are overvolting their ram I immediately think the likely culprit is heat and the easy solution will likely be to ziptie a fan on top of the ram.
 
I bought new Corsair XMS 6400C4 ram roday. And it still fails at the same spot!! Even at default settings. What could be wrong?
 
Originally posted by: jhu
btw, if you're running linux, you don't need to throw away the ram. just tell the kernel what areas of ram to avoid with the badram patch.
Click to expand...

Only if you're OK with manufacturers selling you broken hardware at the price of working hardware. By giving them money for bad hardware, you're rewarding poor QA and not giving them incentive to fix problems. I'm all for using badram for good ram that got a static zap or something, but not for RAM that should have worked.

I bought new Corsair XMS 6400C4 ram roday. And it still fails at the same spot!! Even at default settings. What could be wrong?
Click to expand...

Do you have any other motherboards you could use for testing?
 
I do not have another motherboard to test. Do you think that could be the case though? If its not the motherboard, what else could it be?
 
Originally posted by: Gillbot
I also think I read somewhere about possible false fail in memtest certain versions on certain tests but I don't recall where.
Click to expand...

maybe a search for the specific spot it failed at?
 
i tried my new ram on a new IP35-E with a new E2140 and still failed at the same spot! I am going to write this one off as an error on memtests end. What a POS program
 
Which version of memtest86 are you using? I was having the exact same issue with v3.4, it would crash at about 15% on test 2 on every machine I tried it on (4 total.) I downloaded 3.3, and it appears to be working fine.

 
Originally posted by: cpacini
Which version of memtest86 are you using? I was having the exact same issue with v3.4, it would crash at about 15% on test 2 on every machine I tried it on (4 total.) I downloaded 3.3, and it appears to be working fine.
Click to expand...

Good call! I just ran 3.3 and had no errors!!! My original setup was never faulty
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top