safe micron DDR3 voltage

[saaya]

Hey guys,

Micron ddr3 has been out for over 6 months now and there have barely been any sticks killed. I have however heard from several people killing their micron ddr3 memory and those few reports are enough to draw some conclusions.
I thought id share them with you to make sure you dont kill your precious D9GTR GTS GTN etc

micron ddr3 doesnt seem to slowly degrade like micron ddr2!
it slightly degrades at high vdimm even with good cooling, but then runs more or less stable at the same speeds without any problems... until it suddenly dies.

the problem is that each chip and thus each stick acts diferent and might tolerate more or less voltage. since the chips and sticks dont degrade steadily its almost impossible to predict what stick will be able to run with what voltage safely. seeing as most of the memory barely performs any faster at high voltages compared to reasonable voltages i highly recommend you to think twice at what speed and thus at what voltage you run your micron based ddr3 with.

WARNING
Whatever voltage you decide to use, make sure you meassure the voltage on your mainboard using a multimeter or at least check how much other boards of the same model and revision overvolt vdimm. some mainboards are overvolting vdimm by as much as 0.1v (2.2v bios = 2.3v real)

1.9v officially max voltage CellShock recommends/warrants
2.0v seems very safe for 24/7
2.1v seems safe for 24/7
2.2v can damage/kill memory regardless of cooling
2.3v can damage/kill memory regardless of cooling
2.4v+ can damage or kill memory regardless of cooling after hours-months, depending on the board, the batch of chips and the cooling used

sticks that were reported damaged/dead or degraded with the respective vdimm values
vdimm= REAL vdimm, not what was set in BIOS

2.15v 1
2.20v 1
2.25v 1 1
2.30v 2 2
2.35v 2 1
2.40v 4 7

temperature:
despite of ddr3 running much cooler than ddr2, the chips are still sensitive to high temperatures and will degrade when they get too hot!
even at high voltages a silent casefan will be enough to cool the memory, so its really not worth letting your ddr3 sweat

[eva2000]

Interesting Sascha, thanks for the info

how hot have you seen DDR3 run at compared to DDR2 ?

[metro.cl]

I was told long time ago that max for 24/7 was 2.15v

[DMH]

So can we use 2.3volts for benching for let's say 2 hours without damaging the chips?

[M.Beier]

Sascha, not that this will suprise you, but I totally disagree

I think alot more is safe, however, I think it varies alot from each chip, I've done a little calculation, but I need some backup on my theory, and I dont dare testing it with my next shipment, as its ES the lots of them...

- Im very busy atm, but I hope to catch you one of the upcoming days... (Doing a report... And my heart beats for some Italian chicks that keeps writing.... )

#4
I have personally benched more then 5 hours using above 2.5v...
Results werent submitted, as we had other problems... Lets just say... NB doesnt like -70C

[kiwi]

Who killed ddr3 with 2.3V+? I have seen a few guys running 2.3V for a while (weeks+) just fine, one of them was BenchZowner if I remember right who run 6+ months @ 2.3V or so. It should also depend on PCB and chips of course but still

[Kasparz]

Sascha, not that this will suprise you, but I totally disagree

I'd say Sascha had more than one memory kit to share his experience. Lot more. Even if i had feeding 2.4Vdimm+ for few DDR3 for short time, i totally agree with Sascha. Maybe 2.0Vdimm would me maximum for rating memory by manufacturer if memory has custom heatspreader with fins/extra cooling.
Also that depends of Drive/Data etc strength values.

[harleybro]

Who killed ddr3 with 2.3V+? I have seen a few guys running 2.3V for a while (weeks+) just fine, one of them was BenchZowner if I remember right who run 6+ months @ 2.3V or so. It should also depend on PCB and chips of course but still

Not me well thanks for the heads up on this now. Had them 3-4 months on the last mobo never ran them higher then 2.15 on max extreme was 2.2 boosted to 2.5 while testing E8500 now one stick will no longer boot. (Mobo stopped booting checked everything and finally pulled one stick and it booted again, tried putting in the bad stick and no boot, good stick only it boots, also tried it on new mobo still no boot with one stick) Sticks used for benching only. As for temps. a constant 120mm fan blowing on them (right on top) and either near open window or cold air from dice.

[M.Beier]

Not me well thanks for the heads up on this now. Had them 3-4 months on the last mobo never ran them higher then 2.15 on max extreme was 2.2 boosted to 2.5 while testing E8500 now one stick will no longer boot. (Mobo stopped booting checked everything and finally pulled one stick and it booted again, tried putting in the bad stick and no boot, good stick only it boots, also tried it on new mobo still no boot with one stick) Sticks used for benching only. As for temps. a constant 120mm fan blowing on them (right on top) and either near open window or cold air from dice.

Harley, get inspiration from us danes, look at the post on OCX or here at XS...
Ghetto solution always works
... Apart from that, within a couple of months we will see some new cooling solutions for ram, some were seen at Cebit, but not all.... Its gonna be interesting!

[saaya]

Interesting Sascha, thanks for the info
how hot have you seen DDR3 run at compared to DDR2 ?

ive seen micron ddr3 degrade from running 2.1v without heatspreader and fan on it. the sticks ran notably worse after 1 day.
im not 100% sure about this, but im pretty sure...
i will try to reproduce this after the eester weekend.

it doesnt really make sense as micron rates the chips 85 max, and this usually means 90+ will still be fine as micron keeps some safety range above what they rate. lets be careful and say 85 is the max, above that the chip gets damaged, when you reach 85 die temp you will be around 95 or higher package surface temp, and i highly doubt thats possible with 2.1v.
so either i am wrong, micron isnt careful enough with their rating, or at higher voltages than 1.5v even lower temperatures already damaged the chip.

im pretty sure the latter is true, but how more sensitive the chips react to heat with increased voltage is hard to say...

I was told long time ago that max for 24/7 was 2.15v

im also pretty sure 2.15v is safe... but id recommend 2.1v max
again, IF your willing to void your warranty by using more than 1.9v or whatever other memory manufacturers recommend or rate their memory to.

So can we use 2.3volts for benching for let's say 2 hours without damaging the chips?

if you dont care about losing warranty and not beeing able to get the chips replaced if they burn...
with good cooling id say yes... but make sure its really 2.3v and not 2.35 or more as some boards overvolt, some by quite a bit (highest ive heard of was 0.8v more than what was set in bios)

Sascha, not that this will suprise you, but I totally disagree

no, you dont totally disagree, you think that some chips and sticks can run higher voltages safe, which i never questioned.

I think alot more is safe, however, I think it varies alot from each chip, I've done a little calculation, but I need some backup on my theory

yes yes, like you said micron d9gmh can take 2.8v all day and then it turned out you were actually referring to your fatbody micron chips? :P
sure, some chips and sticks can take much higher voltages, maybe even 24/7, but how can you encourage people to use higher voltages when half of them might end up with dead sticks after a month or two? will you replace them? then please stop trying to convince people to use higher vdimm than what can be considered safe.

about the theory, go ahead and tell us, maybe we can confirm your theory

and I dont dare testing it with my next shipment, as its ES the lots of them...

stop bragging about ES

I have personally benched more then 5 hours using above 2.5v...

yes and you most likely damaged your memory in doing so, whats your point? :P do you want a freekin medal?

Who killed ddr3 with 2.3V+? I have seen a few guys running 2.3V for a while (weeks+) just fine, one of them was BenchZowner if I remember right who run 6+ months @ 2.3V or so. It should also depend on PCB and chips of course but still

not all of theese are cellshock ddr3:

fredyama killed 2 sticks with 2.35 real
harleybro killed 1 stick with 2.25 real
kingpin killed 3? sticks with 2.30? real
i badly damaged one stick with 2.28v real
i damaged one stick with 2.13v real
some guy in japan damaged one stick
some guy in germany damaged one stick
some guy in hungary damaged 2 sticks running 2.4v for weeks
victorwang reported degradation after running 2.28v real for 2 months
hipro5 reported degradation after running 2.38v real for minutes

i never said all chips and sticks will die at 2.3v, but ive seen 2 sticks go bad myself and heard of several sticks die or go bad from 2.3v or even less.
most people dont know about this since barely anybody creates a thread here and says "hey! i killed my memory", thats why i posted this thread.

from what ive seen pcb has little or no impact on how much voltage memory cant take. if you really screw up your design then yes, chips might die at lower voltages, but besides that it doesnt seem to matter.

[saaya]

Harley, get inspiration from us danes, look at the post on OCX or here at XS...
Ghetto solution always works
... Apart from that, within a couple of months we will see some new cooling solutions for ram, some were seen at Cebit, but not all.... Its gonna be interesting!

his memory didnt die cause he didnt cool it properly.
and what cooling solution do you mean?
the aluminum strap-on ln2 pot? :P

we are talking about 24/7 here, not benching, and if you think we are all crazy and 2.5v and higher is totally safe then go ahead and run 2.5v and stop posting in this thread :P

[Yoxxy]

I ran 2.33v for 5 months on Maximus Extreme.

Not saying it is wise to do so, but that was 24/7 and sticks showed no signs of degradation. 1800 6-6-6-18. Only lowered the volts now because I am running 4x1GB and mch will only do 6-6-6-18 @ 1600 mhz.

[M.Beier]

That sure was a long one...

1) Please behave, you dont have to be unpolite, you can keep that elsewhere then public - next time it will be reported.

2) as for the D9DQT / D9GKX, the reason was I didnt buy GKX before late, as my D9DQT did very well..

3) Im fortunate, Im not a representative from any company, thereby I couldnt care less about warranty.. - Its common knowledge that no users are held responsible for information given on a public forum - Further, this is XtremeSystems, and it has always been about doing extreme things... (or atleast for the last 3-4 years)

4) How do you define damaging memory? Electromigration happens all the time, the speed of this process is quite hard to measure, do you disagree on this?

5) No the theory wont be submitted, as I think its too risky, and I sure as hell will get alotta' people on my neck, however, I recommend out of this for every bench session...

6) Do you wish to announce why its the max that Cellshock recommends, and how CellShock tests their kits? This could actually be very interesting reading.
Please tell me if Im wrong, but some companies test at quite extraordinary conditions, to keep RMA as low as possible..
- But my knowledge is based on two companies, and none of them is CellShock..

7) Hmm, yes, the "strap-on" from A-Data is one of them, I wont publish anything else, as I dont know if Im allowed to.. Some of them is only in scetch form for now... Very innovative in my oppinion, it was showed in Planet Reseller during cebit.

8) Saaya, DMH actually asked if a given voltage was safe for 2 hours of benching, then I mentioned that I've tried 0.2v above for 5 hours, and they didnt die... But no, this isnt related to your post #1.

9) Harleybro actually mentioned how he did cool them, and I just mentioned he could have a peak at how we did, one learns from one another... Jesus, without you guys I would have killed all my hardware by condensation ages ago

10) I havent decided what voltage I want to go for 24/7, I use the very max on my motherboard atm.... Hopefully, I can try much more.... How long time do you want me to test...? Is 2 months fair?

[saaya]

I ran 2.33v for 5 months on Maximus Extreme.

Not saying it is wise to do so, but that was 24/7 and sticks showed no signs of degradation. 1800 6-6-6-18. Only lowered the volts now because I am running 4x1GB and mch will only do 6-6-6-18 @ 1600 mhz.

yeah, several people are running 2.3v 24/7 without any problems, but as i said, not all chips and sticks can take this in the long run.
and the real problem is, there is not really any sign of degradation, they just stop working.

1) Please behave, you dont have to be unpolite, you can keep that elsewhere then public - next time it will be reported.

where was i unpolite?

3)I couldnt care less about warranty..

then why do you encourage people to use higher voltages and potentially kill their memory by claiming high voltages are safe? there have been dozens of people posting here that killed micron ddr2 with less than 2.4v, are you calling them all idiots? do you think they are all making this up? indirectly you do, cause you claim 2.8v is totally safe.
and you base this on your experience with one set of fatbody and d9gkx.
thats just ignorant... and it seems to me you only post this to get attention and to disagree just for the sake of disagreeing.
your not helping anyone with what you do, your just annoying people.

How do you define damaging memory?

it runs stable at notably lower speed and timings than before or doesnt run stable anymore at whatever speed.

5) No the theory wont be submitted, as I think its too risky, and I sure as hell will get alotta' people on my neck, however, I recommend out of this for every bench session...

if your not going to share it then why do you post about it here?

[aaug1232001]

But I always felt Cell Shock is specially durable

My friend Cell Shock DDR3 1800 use 1.62v in Bios for 24/7

[DMH]

But that's using 1,62V they are talking about 2,1/2,2V which is 0,6V and 0,7V more than the JEDEC standard

[aaug1232001]

But that's using 1,62V they are talking about 2,1/2,2V which is 0,6V and 0,7V more than the JEDEC standard

2.1v-2.2v for 24/7? I have not certainly had the electric power company share @_@

[harleybro]

Well I have good news at least for me. The cellshocks live! Aparently it was the mobo on it's way out. Last night was a new trial, new mobo. I love my cellshocks booting up easily at 1000 cas 7. So 2.5 did not kill my ram afterall.

[saaya]

great news!

[nut311]

My other RAM D9GTR can't boot DDR3-1000 after run Vdimm 2.44v (real) BIOS 2.35V

[dinos22]

great news!

please change your earlier statement





joke


seriously though saaya sounds as if you guys are providing guesstimates here....................is that the best you can do I thought you guys could test a decent sample to determine what effect different volt range has on durability of your products or do you guys sell only few hundred kits a month or something :S

[saaya]

My other RAM D9GTR can't boot DDR3-1000 after run Vdimm 2.44v (real) BIOS 2.35V

daym, your board is overvolting a LOT... which one is it?

seriously though saaya sounds as if you guys are providing guesstimates here....................is that the best you can do I thought you guys could test a decent sample to determine what effect different volt range has on durability of your products or do you guys sell only few hundred kits a month or something :S

what more can i do than make the line in my first post bold?
should i make it in size 20 letters flashing red and blue with big arrows pointing at it? hehehe

every chip is diferent, all the infos i posted about micron ddr3 dieing are not estimates, this memory really died or is really damaged. my comments about what voltage is how safe are nothing but a guess, but its based on what ive seen and heard of micron ddr3 dieing. and again, every chip acts diferent, so even if id burn a few dozen sticks id still only be able to guess, and id only be able to guess for that batch of chips i burned a few from...
and even then, it might be related to that mainboard, and another one has cleaner more safe vdimm so those chips would have burned on slightly higher vdimm on another board...

and before you complain any more about me/cellshock not posting more details and not testing this more detailled... i havent seen any other manufacturer giving any feedback whatsoever about what voltages are dangerous to their customers to give them a heads up so they dont damage or fry their mem

[nut311]

daym, your board is overvolting a LOT... which one is it?

this A-Data 1600X, in normal condition or 1800 7-6-5-18 1T my ram was no problem but i try again 2000 doesnt boot cas 7/8

sorry 2.34v from BIOS

yeah, several people are running 2.3v 24/7 without any problems, but as i said, not all chips and sticks can take this in the long run.
and the real problem is, there is not really any sign of degradation, they just stop working.

i agree different stick different power

[saaya]

thats similar to what hipro reported when he tried 2.4v+
so now the mem still works fine at 1800 cas7 like before?
the only difference is that higher speeds dont boot anymore?
whats the max you could boot with before and whats the max now?

do you mean boot as in reach windows, or boot as in post, the board starts and shows you the initial boot string information, cpu speeed, mem used, awardbios etc...

so has the max stable speed degraded or only the max bootable/postable speed?

[hipro5]

My benching sessions last ~8 - 10 hours each.....
I run my CellShocks PC14400 at 2.31Volts in every benching session and I don't go higher OR lower that this since I got them (VERY FIRST BATCH - FIRST DDR3 Ram modules - when DDR3 show up)......