Hi All,
Just a heads up, heard from a friend that his daneb died after a couple of weeks, probabaly cause of vdimm.
vcore 1.5-1.7v
vdimm 2.3v at all times
Printable View
Hi All,
Just a heads up, heard from a friend that his daneb died after a couple of weeks, probabaly cause of vdimm.
vcore 1.5-1.7v
vdimm 2.3v at all times
yikes, this is good to know as i'm currently running with vdimm at 2.2 for my 4 sticks of gskill. i also went as far as 1.55 on vcore last night, but i had to back down. i think 1.55 vcore on air cooler is a bit too much. i tried 1.55 to get my 940 to go to 3.8Ghz, but it wasn't stable either :(. i should be happy at 3.7Ghz
I doubt it......Ive had mine at 2.8v since january 15 ......it would be dead by now. Probably a bad cpu or heat issue.Quote:
Originally Posted by saaya
ooopps you said Vdimm........how can you kill a cpu by giving your memory 2.3v?..to my understanding voltage that goes directly to a memory slot has nothing to do with the cpu........you mean VDDA??
i have ran 2.25 volt on the dims from the start. comp is on 24/7.
only during benching i have done a few times above 1.5v, but this a total for a of few minutes.
probably has been to high temp of to much voltage on the cpu.
Depending on the cooling, i'd say 1.7v vcore looks much more like the potential culprit there.
Yeah @ 1.7v propably, or his imc just couln't handle memory voltage....
Mine still runs fine at 1.6v for both vcore and vnb, mem at 2.2v. Then again I have the chip running under phase, but seeing as though it runs at 3/4 load constantly doing gimps with 4 threads for the one number I'm still surprised its alive.
I wasnt aware that voltage thats sent to a dimm slot for memory has anything to do with the IMC. Vdimm from my understanding only give voltage to the actual dimmslot and memory itself and does not send voltage to the cpu or to the IMC for that matter. I would think that AMD would have something to say if memory voltage specs for high performance memory .....(which do sometimes list voltages like 2.3v) actually effected the IMC itself and would prolly put limited specs or restrictions on those voltages.
The internal memory controler (fsb) is set at 200mhz and is seperate from the voltage the memory needs to run.
1.7v was a bit high back in the 90nm days, now this is a 45nm CPU so it is definitely too high of volts... That is probably what killed it.
1.7Vcore is indeed a lot, no matter what cooling. You just cant expect a CPU to keep on running at such high Vcores as long as the cooling is good:shakes:
However, some more details would be welcome, what other hardware?
I'm with the others on the vcore... I wouldn't push 1.7 through 90nm, even 1.6 is rather high IMO. 1.5 is the absolute highest I've pushed through 65nm. 45nm and that high of volts don't mix too well :(
1.55V with air cooling on Deneb is OK.
1.7V on the other hand is not...
Too much vdimm is a possibilty to kill a cpu. The first A64 cpu's (socket 754) suffered from this problem real bad. Rule of the thumb was to have at least 1:2 for Vcore:Vdimm back then.
Mind you, most were running BH-5 memory with plenty of volts on them.
That rule is completely invalid now days, that was only for DDR1 generation. Think about it, even if you ran 1.35v on the core, if you went by 1:2 ratio, thats 2.7v on mem... A good rule of now days is know the hardware you are using and dont push more than is considered "safe" :P
for 90nm, 1.6 was considered max safe for water or high end air
for 65nm, 1.5-1.55 was considered safe
for 45nm, I would have to say 1.5 is max safe.
And if you use stock cooling, I wouldnt go more than .5 higher than spec...
But I dont know what is considered to be the max safe voltage for the memory controller in the CPU, maybe someone else can answer that one.
First Rule... Dont use to low vdimm with Bh5 the Bh5 dont like it and also not the 939 cpu ;)Quote:
Originally Posted by Zeus
4.2vdimm did never kill anything for me...
My Opteron 165 have been @ 3.0ghz a longtime with 4x512mb Bh5 @ 3.6v
Now days...
Phenom II....Vcore thats work best on air 1.45v-1.55v, Vdimm dunno yet got some trouble there, Vnb my bios was @ 1.55v default and i did see it first one month later(you always learn new things :p ) , soo Vnb @ 1.55v looks safe on air maybe or may i have been lucky? ;)
Don't you mean socket 939 and vdimm =< vcore + 2.0v?Quote:
Originally Posted by Zeus
I was hoping we'd seen the last of this crap on AMDs :(
LV I've personally destroyed the dual channel on one of my chips by using high vdimm :(
I never said that rule is still valid nowadays, it was back then. It had something to do with the integrated memory controller.
I didn't say that that rule applies to this K10 platform, only that it is possible to kill a cpu with an imc by high Vdimm.
Last Viking, those were the the days, good old BH-5, hungry for voltage like a pig. :)
With this 45nm process, i think high Vcore will be your worse enemy though.
vdimm can kill an i7 because imc and the ram use the same voltage but i thought that phenoms used seperate voltages for each. i would say having the chip running at 1.7V would be the most likely cause of why it doesn't work anymore.
Congratulations dude! :rofl:Quote:
Originally Posted by saaya
I've been using 2.26V VDIMM since weeks and I have no any problem with it.
There's no need to disrespect one of the founding members of this community like that :nono::shakes:
i agree with that saaya is an outstandin member of this comunity.Quote:
Originally Posted by Oj101
wouldn't hight vdimm efect these new ph11 chips the same as the old 939/940 chips due to the ondie mem controler? after all the volts going to the mem are going thru the cpu due to this ondie mem controller? just a thought feel free to correct me if im wrong....
p.s... Hey saaya was your friend on phasechange or any type of xtreme cooling when this cpu died? If these new cpu's are anything like the old 939/940's i know id be real carefull not to push the vdimm runnin air for any amount of time unless running a benchmark or going for a over max air over clock.
peace perc,
We need someone like macci to say for sure, but even if saaya IS wrong there's no need to disrespect him like that.
People,this is just one case example. There are still unknowns to why that happened,it could be the guy's PSU,a short on the board,failing board(and bad vrm),stars not alligned with the Sirius/whatever (j/k) etc.
well if one cpu dies at a certain voltage it doesnt mean all will, there are always those odd .1% chips that die at voltages way below all the rest... plus it might have been a freak accident and not caused by vcore and vdimm at all after all! :DQuote:
Originally Posted by dng
im just saying, COULD be...
just cause one chip died i wouldnt say 2.3vdimm or 1.7v vcore is dangerous... only when several chips die we can conclude something like that.
i only posted here to share the info and see if anybody else saw something similar, cause its quite common that people who see hardware dieing kep quiet about it... its nothing to brag about and some might feel ashamed cause they think they did something wrong :D
but once there is a topic people might come forward and report similar problems... good to hear that nobody seems to have any problems so i assume this was a freak accident :D
yepp, atm i dont think its time to call any voltage levels dangerous yet, just wanted to share the info.Quote:
Originally Posted by Titan7171
wait, you had your chip at 2.8v and its not vdimm? then what voltage is it?? :confused: vcore? :lol:
and yes, the memory controller is in the cpu so of course vdimm goes to the cpu. on i7 a few chips died with vdimm as low as 1.8v, then again there were very few reports so those might have been freak incidents as well...
what cpu? daneb? what vdimm and for how long?Quote:
Originally Posted by Oj101
it wasnt even me :PQuote:
Originally Posted by Oliverda
if it would be, then id admit it, im not ashamed or emberassed to admit it when i kill hardware, especially if its in a stupid way ^^
thanks for the comments guys, but i dont think oliverda meant to offend me, no offense taken in either way :p:
so nobody else has seen or heard of a dead or damaged daneb?
well that one was ES so maybe thats why... it was running on several boards and has been tortured a lot with ln2 and dry ice and compressor cooling and some might have been without the board beeing insulated, so there might have been some shortage or power spike...
good to hear that there are no other reports actually :)
so 2.3v seems to be fine for the mem after all and around 1.5vcore as well, possibly higher.
good infos for people who are going p2, thanks for the feedback guys :toast:
Saaya dude, it's Deneb not Daneb :D
Mine was a 3700+ San Diego which no longer boots with more than one stick of RAM :( Socket 939 but same thing (1.35vcore, 3.74vdimm, IMC).
my ram is rated 2.3v for 1066 runnings, 1.8v for the 800 and I have been @2.34v ever since getting the 940's, now I am beginning to worry.....
my vote is something else killed the sillycon, i would never run 1.7v unless I had the Prometia unit attached to the CPU.....
bummer, at least we are not reading of massive failures across the board, yet.........
laterzzzz................
Well according to Everest 5 the K10 IMC has a max of 2.9 vdimm and AMD specs for PII is max 1.6v, so I think it was probably the vCore that killed it .........
oh really? lol, my bad :DQuote:
Originally Posted by LowRun
lol, for how long? 3.7 is really pushing it :DQuote:
Originally Posted by Oj101
nah, dont worry man, just a one off so far, and like most people said, it was very unlikely that vdimm killed it, maybe vcore, or it was a freak accident.Quote:
Originally Posted by bldegle2
2.9 max for 45nm? thats impressive... i wonder if its going to be the same for the ddr3 imc, and if yes, then why intel cant get the same vdimm range working.Quote:
Originally Posted by Russ_64
Probably process/design differences would be my guess.Quote:
Originally Posted by saaya
Two months?:DQuote:
Originally Posted by saaya
They're Kingston HyperX, very nice PCBs because they're still 100%. I'm not worried about hurting the BH-5 ICs at all, they're rated at 5v but PCBs that can handle 3.6v+ 24/7 are veeeeeery nice :D
http://www.xtremesystems.org/Forums/...d.php?t=217888Quote:
Originally Posted by saaya
killed an X3 with a 3.5vdimm cpu@1.4v my X4 ES has been running several weeks @vcore1.5-1.65v vdimm@2-3.1v no troubles so far
IIRC the early Phenoms had a rash of dead CPUs due to high vdimm over 2.3v
Thats why I held back on the DDR2 8500 because of the 2.0+ Vdimm required.
In fact I bought nothing but the Gskill 6400 from the start of am2+ because it ran at 1.8v. Try and track down where I read that.