Memtest нашел ошибки что делать

Saphire HD 7850 OC (at default OC)
i5 2500k (turbo enabled, i.e. 3.7 ghz while gaming)
16 GB Corsair Vegeance @ 1333 CML16GX3M4A1600C9B (but can activate XPM for 1600 but would this really help at all?)
Maximus IV Gene-z/Gen 3 Motherboard BIOS — 0402 (first released BIOS)
Blue Caviar WD 500 GB
Seasonic Semi Modular 620 W PSU

I made a thread here: http://www.tomshardware.com/forum/311741-30-memtest-error-replace

Because I found several errors in the same address while running memtest. So this is what I did: I tried each stick individually in slot #1, all passed except 1, then tried one known good stick on the remaining slots and no problems at all. Which pointed to bad RAM, so I got a replacement from Amazon, I was extremely careful in placing them in the motherboard, ran memtest got error on the same address with same pattern on test #2 all the time, if I let it run more I also get similar addresses errors on test #5.

So at this time, I am not sure what the problem is, could it a problem with something else?. In between waiting for my replacement I took out the bad stick from the first memory I got from amazon and ran memtest 86+ with only 2 sticks and got no errors at all, which reinforced my theory that the problem was with the RAM.

I guess it is possible that I got yet again another bad stick of RAM from Amazon, possibly due to lousy packaging?. I was thinking since I have all 8 sticks here, I am going to try and find the bad stick on this other set, and mix and match?. I know that is not a good idea, which is why they are sold as a set. But I figured if it works… What do you guys think?.

I also tried setting vram to 1.53 and memory controller to 1.2 and same results.

Other notes: Other than the memtest errors, I have not experienced any other stability or BSOD issues. I only started to suspect RAM because some files got corrupted twice in Starcraft II that is it.

Any help will be greatly appreciated.

Thank you,

Oscar

I was going to RMA to Amazon again, but I have heard Amazon sometimes bans costumers because of too many returns, and I do buy a lot of stuff that works from them… How long is the process to RMA to Corsair?.

Anyway I used my 3 new sticks, and one of the good sticks from the first batch and ran memtest with all together for 6 hours a little over 3 Passes and all passed with 0 errors, would it be safe to keep it this way to avoid RMA again? And just RMA the other 4 mixed sticks back?.

I know you should not mix sticks, but this ones seem to work well together, what do you think?

Thank you

Oscar

Thank for your reply, in this case my memory is fully supported for my cpu and mobo, as many people inthe asus rog forums are using my exact setup with no issues. I could be wrong, but I think it is faulty RAM as it fails individually, in adittion when I replaced the bad stick with a good one and ran all 4 it ran with no errors for 6 hours, while it fails within minutes with the faulty stick.

You discharged yourself from any static before touching the memory? You didn’t walk on carpet at anytime during the install right?

Also you need to discharge your motherboards capacitors before removing/installing memory, this is done by powering down the computer, and turning off the power supply switch and then pressing the computer power button 3 times.

You will see the lights come on and the fans jump even though everthing is unplugged and turned off, this is remaining power in the capacitors.

Check the memory contacts for any burn spots.

I only handled it from the edges and I did not walk in the carpet, I had the psu connected to the ground but the swich OFF, an i was touching the psu with one hand and installing the memory woth the other.

I did check throughly for burn spots before installing the new memory but found none.

I had no idea you had to discharge the capacitors I read about 10 guides and none mentioned it =(, thank you for telling me. Could it be that is what messed up one of the sticks from each batch??

I dont think it can be a burn slot since they were tested in the same slot #1 and only 1 failed but I could be wrong. Adittionally it is working fine now that I replaced the bad stick with a good one from my first order.

The scariest part is the corruption you may not know about.

RAM and HDDs are stuffed with non-critical data like text, images, geometry, numbers, etc. where a couple of flipped bits may cause typos, visual glitches and other minor problems but not necessarily crashes.

Would it be ok to use it like it is now with 3 sticks from the second set and 1 stick from the first set?, it ran memtest fine for 6 hours.

I guess at this point I am so frustrated with getting 2 bad sticks from Amazon that I think either the third one will also be bad or Amazon will ban me for too many returns (yes I read that ramdomly happens

My third option is RMA to Corsair do they pay for shipment?, how long is the process?

Yeah I was talking my friend though installing ram on the phone. He said it sparked and he smelled smoke, turns out he burned one of his dimm slots and we had to replace the mobo.

I make sure I mention emptying the capacitors first from now on.

That’s one odd procedure since the DC power rails lose power within milliseconds from using soft-off — a 10A constant-current drain on 10 000uF capacitor bank will go from 12V to 0V in 1ms. What does take a while is draining the primary storage caps in the PSU, which can take 15-30sec with only the 5VSB supply draining power.

Turn off the PSU’s main supply (if it has physical switch) or disconnect from the wall, wait for 5VSB LED on motherboard to go out, now it is safe to start pulling stuff out.

Since DIMMs do not receive any power while the motherboard is in soft-off state, turning off the PSU is not absolutely necessary. It is mostly a precaution against accidental turn-on.

I will do that, unless do you think mixing the sticks now tha they work is fine if no errors show up?

Yeah, that should be fine. You can look at the dimms and see what date they were manufactured. If they are a day or two apart, it should be pretty much the same chips.

Wrong. Very bad advice.

If this is true, then why do the fans spin and all the lights come on and hd spin up even though the pc is unplugged and the psu is off?

If the lights come on and all the fans start spinning and the hd spin up, there’s voltage getting to the dimms.

If you can make a disconnected PSU turn on, you did not wait long enough for the 5VSB LED present on most motherboards and some PSUs to turn off and therefore you did not follow my instruction/advice.

5VSB is the last power rail to go out in a properly designed PSU that gets unplugged, which can take more than 20 seconds with some PSUs and light/no 5VSB loading. Once it goes out, the line input caps (200-400V) have insufficient energy/voltage left to start the main converter(s) again.

That said, I do use the unplugged turn-on ‘trick’ to shave 10-20sec off the 5VSB fade delay.

I tried to find a manufacturing date on the DIMMS but could not find any, all I found was a version and it is the same version. Where would the date be?. They are corsair vengeance, anyway the 3 new and 1 old stick seem to work without issues, so I think I will just keep them like that.

As long as the internal structure of the DRAM chips is the same (same number of row and column addresses) and they have the same timings, mix&match is often possible.

Who said ANYTHING about making a disconnected psu turn on??? You my friend have bad reading comprehension skills.

I’m talking about stored capacitor power, who wants to wait? I said just hit the power button 3 times and your ready to start pulling/inserting parts, no waiiting.

You did. I recommend proofreading.

What a coincidence, I was talking about stored capacitor too, your leet reading skills apparently missed that. You also missed my original point that the 5VSB LED is a more reliable indicator of whether or not the primary caps have been discharged than whether or not the PSU will do something when you push the soft-on button.

While the PSU may not react to soft-on due to under-voltage on line caps, the 5VSB supply may still be live due to having a lower turn-off point. That’s why I said the 5VSB LED is the more reliable indicator of when it is safe to proceed… whether you let the 5VSB drain primary caps or push the soft-on button 50 times, your final check will always be the 5VSB LED.