Jump to content
Corsair Community

RMA request


Recommended Posts

I got a ValueSelect 2x512 VS512MB400 package today, and one of the sticks appears to be bad. My system has an AMD AthlonXP 2800 (333fsb), and Asus A7N8X-X mobo, default bios settings (2.5-3-3-8). I installed both new sticks, then had a random reboot during a game. System POST passes fine. Memtest86 showed 100+ errors during test 5, and 3 in test 8. My original 512 stick (also VS512MB400) works fine, and has been tested in all 3 slots, no errors. Only one of the new sticks gives errors, almost all during test 5, and all have "good = ffffffff" with "bad" varying between ff7fffff, fff7ffff, and fffbffff (there could be more, these were what I saw). I also tried slightly different timings (2.5-3-3-7, and -11), same results.

 

On a side note, Memtest86 reported "dual channel" mode with all 3 sticks installed, which appears incorrect. This motherboard has the nforce2 non-ultra chipset (it's the budget version), and does not support dual channel to my knowledge, nor is there any such indication (configs and such) in the manual.

 

I would like to RMA only the bad stick (rather than return both to the retailer). Thanks,

 

-Lucas

Link to comment
Share on other sites

  • 3 weeks later...

Hi again, I got my replacement part on friday, but it would seem I received another bad stick. This one fails memtest86 test #8 consistently, averaging 30 errors per pass (good=00000000, bad=80000000). It is not in the same slot as the other bad stick was, and I will try running it solo tonight just in case my system has some issue with using all 3 slots, but is there anything else I can do prior to RMAing the replacement? Should I test different memory timings, bios settings, or try upping the voltage? I think I shouldn't have to, and it should run at spec, as obviously my other memory is fine, but I can monkey with the settings if it's worth trying.

 

-Lucas

Link to comment
Share on other sites

On further testing, I was surprised that the replacement stick does fine when solo. I suspect that what I may have now is a heat-related problem. The RMA'd stick failed test 5 with 100+ errors per pass, even solo, so I was probably right to return it, but my new results suggest the current issue is something different, and possibly heat related.

 

Here are the facts:

 

Friday (day 1), nice weather, no errors occured in testing.

Saturday, it was warm (70-80) and very humid, and the replacement stick generated 30 errors per pass in test 8 consistently.

Sunday, it was cooler and less humid, and fewer errors occured, though they were still present.

 

These were with all 3 sticks installed, which is the only time I find errors now.

 

All errors on the replacement stick were from the first half (<256MB), and it was in the end-slot. This may indicate only one side of the stick having heat issues.

 

When moved closest to the case fan, the replacement stick did not generate errors, but my ORIGINAL stick, now in the middle of the 3, gave 6 errors in test 5 (one pass of all tests).

 

The 2nd stick (kept from the new pair), did fine in the middle previously, so it has been swapped to there with the original now where the replacement stick had been erroring Saturday, and I will test again overnight with this configuration.

 

My case is a LITE ON|HT761B 400W mid-tower, and additionally contains 1 seagate hd, 1 lite-on DVD/CDRW combo drive, and a GeForce 6600GT AGP card. There are 4 fans total (cpu, gpu, psu, case). Asusprobe reads the mobo temp currently at 30C/86F.

 

I think re-arranging the sticks may be enough to fix my issues, if I'm lucky, but I will have to test more. I might need some RAM heatsinks or another case fan.

 

Any information to verify or debunk my suspicions would be appreciated. Thanks again,

 

-Lucas

Link to comment
Share on other sites

Well, this gets more interesting. My overnight test turned up 100 errors per pass of test 5, on the 2nd module, now in the middle slot. All these sticks tested fine on their own, and rotating them still gave various (different) combinations of errors with all 3 installed. It also seems very unlikely that I got 3 bad modules from 3 separate batches.

 

I tried one more thing, which was to manually clock the memory down to 166 (FSB speed), instead of 200 (the "by speed" bios default). This cut down the error rate to just a handfull per pass (on the replacement stick, now in the first slot, and all ABOVE 256MB). Further changing the timings to 2.5-3-3-11 gives me a clean all-tests pass, and multiple passes of tests 5 & 8, no errors whatsoever.

 

I can probably live with this, but it's not quite ideal, and I don't know what the performance impact will be, if any. It eliminates any hope of later OC'ing the Athlon to 200mhz FSB.

 

What I might do is to try all 6 stick-slot combinations, each overnight, and keep a log of the errors, to find the happiest configuration, then de-tune the settings until it's all error-free again. It just doesn't seem worth it to play RMA-roulette at this point.

 

I also noticed the bios defaults to 2.6v on the DDR, and there are a lot of comments here suggesting 2.7v. Should I try that? It could be a power issue (cheapie mobo + psu), rather than heat. Thanks,

 

-Lucas

Link to comment
Share on other sites

  • Corsair Employees
That would suggest some other problem, and if you run the CPU at 333 MHz you would need to run the memory at DDR333 as well. Can you tell me the make and model of MB you have along with the CPU speed and it’s FSB as well? In addition, please tell me the bios settings you have set for both CPU and memory and any performance settings that you may have set?
Link to comment
Share on other sites

Interesting.

 

The system info is in my original post, but I'll repeat it: Athlon 2800+ XP (333fsb), Asus A7N8X-X, all settings default (initially), but using memory at 166mhz and 2.5-3-3-11 at the moment.

 

So you think the problem was letting the mobo default to 200mhz memory speed when the bus is at 166? It actually has a multiplier you can set manually (currently 100%, for 166), but the auto defaults to 200mhz (by memory speed), and the board does support that (just not with this cpu).

 

-Lucas

Link to comment
Share on other sites

Wow, thanks very much. I had not read that anywhere! This should go right at the top of your dumb-things-to-doublecheck list. My AUTO setting happily picked the by-memory-speed 200 setting and botched itself up.

 

My mobo manual states "Make sure the memory frequency and bus frequency setting in the BIOS are the same or set to [AUTO] to ensure system stability."

 

I guess they presume you wouldn't put DDR400 in a 333FSB system. Doh! I went for the very-cheap upgrade to 400 since I may eventually be able to pick up an Athlon 3200+ for peanuts when this system is near it's retirement age, and I don't want to replace the RAM as well.

 

My testing should be much more productive now. I'll let you know how it turns out.

 

-Lucas

Link to comment
Share on other sites

Even at 166, I was still getting infrequent errors with different module arrangements and settings. Yesterday I rearranged them one more time, but also thought to move the various fan/power/etc cabling away from the modules, and have not seen an error since. I left memtest86 overnight running all tests over everything, and cleared 13 passes successfully.

 

Now that it's all working error free, I don't want to mess with it, but I remain curious. Could the cabling have caused enough interference to generate memory errors? It would appear that the errors appearing <256MB on the left slot and >256MB on the right slot correspond with the previous cable positions (hanging around the slots on either side). This could also suggest the error rates changed by rearrangement were related, as it was not possible to access the modules without moving the cables slightly.

 

Anyhow, mission accomplished. Thanks for your help!

 

-Lucas

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...