Micron D9 and high voltage, electromigration effects

[mad_skills]

not 1 of my sticks have been degraded since i went dfi

This is so true, it's the same case with me. None of my kits perform differently since when i bought them.

[ziddey]

my dfi p965-s ate a pair of gkx and gmh so far. i'm more or less ok with it though since the gkx may have been on its way out already, and i got about a year from the gmh. voltage for both was 2.06v, actively cooled.

[CryptiK]

Was that 2.06v real or bios set?

I got 8 months with no issues out of some corsair 8500 (GMH) that was actively cooled running 2.16v (real) 24/7. That was on a P5B vanilla.

[Shiranui Gen-An]

Ugh, one of my aXeRAM 1200 sticks just went bad. I ran those things at 2.2V REAL for a month!!

[mad_skills]

They are also D9GMH and perhaps 2.2V is a bit too much.

[Shiranui Gen-An]

2.2V doesn't seem to have hurt my Ballistix... I'm *almost* about ready to give up on Micron D9, but I don't see what else there is that's as overclockable.

[ziddey]

Was that 2.06v real or bios set?

I got 8 months with no issues out of some corsair 8500 (GMH) that was actively cooled running 2.16v (real) 24/7. That was on a P5B vanilla.

real. 2.1v set

[CryptiK]

I don't think anything is a fun, overclockable or flexible as Micron D9's. I love them, as do many, but they are quite easily damaged.

I think it comes down to the voltage regulation of the board, and many other factors. Some people have run 2.3v+ 24/7 for months on end with no degradation, others loose sticks in 1 month at only 2.2v.

[mad_skills]

Voltage regulation and pwm unit.

[CryptiK]

I think either my board, or my new Teams, are degrading.

They were stable at 1112 MHz, 1.99v (real) load, now they are at 2.10v (real) and failed orthos after 49mins.....I'm retesting at 2.14v (real), but this isn't looking good.

They are about 3 weeks old.

[mad_skills]

That's not that much time for a kit to degrade. Perhaps some other factor is causing the problem?
Try memtest, i don't use orthos for testing ram stability.

[CryptiK]

OK thanks I forgot about memtest - by comparison to Large FFT test in orthos isn't it a weak test though? I have had sticks that pass memtest loops but fail rapidly in orthos or when in windows and running apps.

Still, I shouldn't be failing anything now at the same settings that I was stable at before. I'm concerned, but as you said it is a remarkably rapid degradation.

I moved the sticks during an orthos stress test, and the pc froze. I have just now pulled the sticks and cleaned the contacts with a cotton bud (q-tip) and isopropyl alcohol. They were quite dirty, leaving residue on the end of the cotton bud.

They are testing now, I'm gaming while running orthos large FFT.

[mad_skills]

I also had several cases failing orthos and passing memtest. I used that as stable because if they pass memtest they never gave me any error in real work/job/gaming etc.
It's always good to have another pc nearby, i try my rams at my brother's pc if they give me any trouble to see if they still work normally. It would be wise if you could to do the same, perhaps at a friend's or so.
Do tell if you seen any progress, i am really eager to see what could be the issue.

[Ch@pS]

Mine D9GMH set has been run at 2.2v for most of its really long life, been up to 2.9v under active cooling etc, and been used on a variety of boards from DFI 965s to Asus and MSI P45s, and still work now as well as they did when they were 1st overclocked (a year, maybe more?). i run them at 2v 1200 5-5-5-5 for 24/7 use.

[nfm]

I have my Ballistix PC8500 @ 2.09v DDR1200 running for over 14 months now on DFI LP UT P35 and PC Power & Cooling 750. There were times where I ran it at 2.2v for few weeks and benched at 2.35v. Before buying them and reading the threads, I was almost sure they will die on me within weeks, but here they are keeping strong .

[zsamz_]

i damaged my best set goin 1300+

[mad_skills]

Wasn't 1240MHz enough?

[zsamz_]

Wasn't 1240MHz enough?

no are you serious
i wanted to break 12000mb/s
but i got 2 more sets of ballistix 8500 so i'll try again
i guess 2.45 for 10 min was too long lol
i shoulda put in my d9gkx

[CryptiK]

After some arduous hours testing and fiddling, I have found my ram has not degraded. It was my cpu not being stable and causing the crashes, and some of the secondary timings when left on auto were too tight for high MHz stability. I am 20 passes of linpack (Intel Burn Test) stable now @ specs in sig.

I actually booted at DDR1333 Mhz at 2.24v (nowhere near stable obviously) so my sticks can do high freq very well it seems, my board just has a massive hole between 1199 and 1277 MHz where it refuses to boot.

[mad_skills]

no are you serious
i wanted to break 12000mb/s
but i got 2 more sets of ballistix 8500 so i'll try again
i guess 2.45 for 10 min was too long lol
i shoulda put in my d9gkx

Of course i'm not serious.
Be sure to put some active cooling while benching at that voltage.
But that is kinda low (high) voltage. My Team's benched like half an hour on 2.75V with a 120mm fan and didn't degrade.

[zsamz_]

Of course i'm not serious.
Be sure to put some active cooling while benching at that voltage.
But that is kinda low (high) voltage. My Team's benched like half an hour on 2.75V with a 120mm fan and didn't degrade.

i had 2 90mm fans on em lol they never got above 25c

[garson]

My Team's benched like half an hour on 2.75V with a 120mm fan and didn't degrade.

I tought one of your Team is dead.

[mad_skills]

I kinda nursed it back to life.
But i just keep them safe now, don't bench with them anymore.

[zsamz_]

i just retested my stickx they are back
but i need 2.05v for 1220mhz

well i had to punish my d9gkx they still rock solid

[koli.fx]

Guys,

I don't think it is the voltage killing these chips. After first two kits of my Ballistix 6400 died (ran @ 2.2v 4-4-4-12 @ 800mhz for 5 months) I thought it was due to my Maximus overvolting. So when I got replacement kits I tried to find lowest voltage that would run 4-4-4-12 @ 800mhz. I started at 1.8v (in bios), upping until 1.86v. At 1.86v my 4x1gb modules would pass the memtest so I left it like that.
These modules are meant to run at 4-4-4-12 @ 2.2v. I got it running @ 1.86v. I don't care how much my Maximus overvolts, it does not matter because I got my ram to run at latencies and frequency as specified by manufacturer at the lowest possible voltage. I hope it makes sense.

Now after just TWO months, one module out of those two replacement kits is dead too. It never ever EVER ran above 800mhz or 1.86v. Second kit runs without problems @ cas 4 @1.86v.

Also I would think that Asus Maximus is using high quality voltage regulators.

I RMAed the bad kit (thank God for brilliant customer service at Crucial). I am considering getting something else. Can you tell me what manufacturer doesn't use D9? I am thinking about going with OCZ Reaper. All I need is ram that will do 1000mhz (I will get C2D 8600 soon) and won't F***ing die in the process.

Thanks.

koli