Micron D9 and high voltage, electromigration effects

[zsamz_]

Hello.

I did buy 2G from Mushkin 8500 2G.

They are D9 Micron Chips.

I bought them in August 2007, running 24/7, with 2.2V, at 440 FSB 1:1, on my Asus PK5-E....

3 Months ago, i star to have problems in games, and finally, after runing Prime 95, i discovery, that one of the modules was defective, failing the Prime 95, at any voltage and timming....

I did RMA, at the store, here in Portugal.......

Sorry the OFF, but.....SHAME on Mushkin RMA here in Europe/Portugal...

No product stock, and i can`t get back my money!!!

Glory days, of DDR 1.....

i'm on my 4th rma with my mushkins
funny how its only my mushkin d9 that have died on me 2.24v

[saaya]

i think its all about the chip batches...
I dont think it has to do with the mfg...

the latest batches of gmh were bad i think... they die with really low vdimm already.
luckily we had some older stock at cellshock
but even with those there is already a higher rate of chips dieing than with the very old gmh and gkx batches...

[zsamz_]

i think its all about the chip batches...
I dont think it has to do with the mfg...

the latest batches of gmh were bad i think... they die with really low vdimm already.
luckily we had some older stock at cellshock
but even with those there is already a higher rate of chips dieing than with the very old gmh and gkx batches...

what you think about the last batches saya will they die too lol?
2.03v @ 1230mhz bad for 24/7?

[saaya]

i dont know, we didnt buy small quantities of the latest batches at cellshock, so we didnt have a constant impression about all the different batches...
I can only say that roughly the newer the chips the higher the chance they die with 2.4v or even less.
its still small percentages, keep in mind that if a stick fails or dies usually only one single chip went bad!
so its a small amount of chips that actually fails... its usually more recent batches tho from what ive seen.

2v 1200... very nice

[saaya]

i dont know, we didnt buy small quantities of the latest batches at cellshock, so we didnt have a constant impression about all the different batches...
I can only say that roughly the newer the chips the higher the chance they die with 2.4v or even less.
its still small percentages, keep in mind that if a stick fails or dies usually only one single chip went bad!
so its a small amount of chips that actually fails... its usually more recent batches tho from what ive seen.

2v 1200... very nice

[zsamz_]

latest batch i meant double sided saya not the higher density chips

even if some new batches would clock high my gut tellin me they will die quick

[saaya]

double sided chips? dont get whatya mean bro

[zsamz_]

double sided chips? dont get whatya mean bro

i meant i got one of the last batches before they swithed ics lol
i just hope 2.03v is safe for 24/7

[saaya]

yeah, 2.03v should be 100% safe from what i know, even on 975 or 680 which kill memory at notably less vdimm than other chipsets

[Feriuz]

DFi DK p35 now kill my OCZ Reaper 1066 i'm set vdimm 2.35 not 24/7 - -. Sadddddd

[TheGanG]

I use my 4-4-4-10 PC-6400 Micron Team Extreems on EVGA 780i/ABIT AW9D Max for some benchs up to 2,425V for 3-3-3-3/800 and 2,35V for 5-5-5-15/1200...

I'm still curious if that voltages could hurt them at Super Pi tests or 3DMarks...

What kind of test should i use to know if they are still alive or degraded?

[saaya]

I use my 4-4-4-10 PC-6400 Micron Team Extreems on EVGA 780i/ABIT AW9D Max for some benchs up to 2,425V for 3-3-3-3/800 and 2,35V for 5-5-5-15/1200...

I'm still curious if that voltages could hurt them at Super Pi tests or 3DMarks...

What kind of test should i use to know if they are still alive or degraded?

the best is to write down the exact configuration memtest or some other test passed. then every now and then get back and configure everything exactly like that, and check if the memory is still stable or if it degraded.
and checking if the memory can still run a certain speed with low vdimm is also helpful, some chips show first signs of degradation at low vdimm.

EDIT: keep in mind, if you change the cpu, mainboard, upgrade the bios, the temperatures changed, the psu is diferent... then the results might slightly vary.
from one board to another you will see big differences sometimes. keep in mind that variations do not necessarily mean the memory went bad!

[TheGanG]

the best is to write down the exact configuration memtest or some other test passed. then every now and then get back and configure everything exactly like that, and check if the memory is still stable or if it degraded.
and checking if the memory can still run a certain speed with low vdimm is also helpful, some chips show first signs of degradation at low vdimm.

Should 1-2 hours of Prime95/Blend Test a good idea?

And later, i will try the same benchs at same timings/voltages as i did before...

[saaya]

in my experience 30mins is already enough to have a reproduceable landmark, but yeah, 2hours sounds fine.

[18 is # 1]

I think I'll keep mine at 2.3v and below. I kinda like 'em!

[zsamz_]

I think I'll keep mine at 2.3v and below. I kinda like 'em!

i lost 3 sets @ 2.24v lol on gigabut p35
i keepin mine @ 2.03v 1220mhz

[18 is # 1]

i lost 3 sets @ 2.24v lol on gigabut p35
i keepin mine @ 2.03v 1220mhz

Were all your loses on the same mobo?

[jcool]

I lost 2 pairs of cellshock 800 running 24/7 at 2,1V on 2 different mobos with different PSUs. Talk about bad luck
I dunno if it was electromigration though, they worked fine and one day, after a reboot, I only got C1 error. Tried the sticks in 3 other mobos to be sure, but one of the two sticks was dead. Same with the 2nd one
My crucial values seem indestructible though...

[zsamz_]

Were all your loses on the same mobo?

not same mobo lost 2 sets on 2 different p35c-ds3r 1 set on x38-dq6
i'm on dfi now n 6 months not 1 dead stick

[onelivestarfish]

hmmm...seems that 2.4v may be a bit much for those D9 owners out there who want to save their memory

[18 is # 1]

hmmm...seems that 2.4v may be a bit much for those D9 owners out there who want to save their memory

What were we talking about?

[Schmetterling]

I use my 4-4-4-10 PC-6400 Micron Team Extreems on EVGA 780i/ABIT AW9D Max for some benchs up to 2,425V for 3-3-3-3/800 and 2,35V for 5-5-5-15/1200...

I'm still curious if that voltages could hurt them at Super Pi tests or 3DMarks...

What kind of test should i use to know if they are still alive or degraded?

They will die, i had two sets of them dying on me already whilst running stock speeds.

[mrcape]

I have 3 sets of d9gmh, two that don't boot up at default bios. But they all clock well and stable between 2.0 - 2.45v on air (120mm fan from above and side) and have only had the issue booting up. One has had this issue for a year and hasn't seemed to get any worse at all other than needing a tad more to boot. All can hit 1200 for benching and go high and tight. So I guess I have seen no problem other than a pain if I need to reset bios. I have a cheap low v stick just for that reason. So what is that, is there a break in with these or mild degredation?

[saaya]

I lost 2 pairs of cellshock 800 running 24/7 at 2,1V on 2 different mobos with different PSUs. Talk about bad luck
I dunno if it was electromigration though, they worked fine and one day, after a reboot, I only got C1 error. Tried the sticks in 3 other mobos to be sure, but one of the two sticks was dead. Same with the 2nd one
My crucial values seem indestructible though...

yeah, those must have been from a bad batch then :/
as i said, some batches work really great with high vdimm for months or years even, some seem to have a high failure rates at high vdimm after weeks to months.

[mrcape]

So I take it there's no real clear way to identify if a stick is on it's way to die, or just needs more juice and will be fine. I choose to look at the benching sticks as somewhat disposable, giving them what they need to clock high, and keep one kit under 2.2v for a solid backup. That's the beauty of cheap DDR2 these days. But I will be damn careful with my DDR3 lol.