actually its not common knowledge because there are several people that have been running 3.2v+ with no problem.
Thread resurrection is bad.
Printable View
actually its not common knowledge because there are several people that have been running 3.2v+ with no problem.
Thread resurrection is bad.
No, many people vdimm mod and go higher. I've been running mine at 3.5v for a month without any signs of a problem. Also you I've read that you're looking at the Chaintech 250. Just to let you know I've never seen anyone happy with that boardQuote:
Originally posted by Geforce4ti4200
I didnt bother reading the whole thread, but at this point it is now common knowlege that DO NOT raise the vdimm past 2.9v and dont do any mods, especially not vdimm mods! The a64's ondie memory controller hates 3v and higher. I wont tell you not to vdimm mod, some will for high benchmarks but they pay the price in dead cpus. keep the vdimm at 2.9 or below even if you can get higher
so I guess me running 3.3vdimm for a few months 24/7 is just a fluke? :rolleyes:Quote:
Originally posted by Geforce4ti4200
I didnt bother reading the whole thread, but at this point it is now common knowlege that DO NOT raise the vdimm past 2.9v and dont do any mods, especially not vdimm mods! The a64's ondie memory controller hates 3v and higher. I wont tell you not to vdimm mod, some will for high benchmarks but they pay the price in dead cpus. keep the vdimm at 2.9 or below even if you can get higher
Quote:
Originally posted by rro
would it be possible so summarize the general findings of this thread?
It a bit of a long read 8pages!
yes.. high vdimm (3.3v) is okay for 24/7 use.. high 3.3v line (3.6V +) is bad but up until a few month ago it was needed. At least for the gigabyte series :)
eshbash needs to read this thread, I can vouch for him, he lost 2 a64s to high vdimm volts of 3.2v or $700+ of cpus in 3 months. hes mad and done with vdimm mods! what is so bad about the chaintech? the only other alternative is the epox at $40 more and that one seems to be even worse :(
hmm my 3 months at 3.6V 24/7 on an AN50R must be a miracle then :rolleyes:Quote:
Originally posted by Geforce4ti4200
I didnt bother reading the whole thread, but at this point it is now common knowlege that DO NOT raise the vdimm past 2.9v and dont do any mods, especially not vdimm mods! The a64's ondie memory controller hates 3v and higher. I wont tell you not to vdimm mod, some will for high benchmarks but they pay the price in dead cpus. keep the vdimm at 2.9 or below even if you can get higher
It's just like the guy at OCAU... Pulling unsubstantiated facts from thin air to make a point. And on both occasions, from people who don't have A64 setups :rolleyes:
My AN50R is Vdimm modded, i'm pushing 3.15V trough the mem 24/7 for 3 months now, have been as high as 3.3V for benches, still everything works 100% ok.
The difference is in the Vmod itself and the way of increasing voltage methinks.
Oh, and indeed, that Chaintech board sucks, why not just go for the older NF3 150?
Anyone ever noticed a difference?
I think I may have killed my 3000+. I was bored this morning and wanted to do some benchmarking and stuff so I pumped up my vcore to 1.82v and vdimm to 3.1v using my trimmers (normally I run 1.75v and 3.0v) to try for my maximum benchable OC. After having not much luck improving on my current 2.5ghz I quit and turned back down the voltages. Then I started up again at my normal speeds and I was playing BF Vietnam and my system went to a black screen. I thought that was wierd and rebooted and tried again and the same thing after about 10 minutes. I rebooted and did Prime 95 which failed in under a minute when I am normally stable. Dropped my OC down to 2.4 from 2.5 and it failed as well. Tried again at 2.2ghz and after a few minutes my system went to a black screen again. Temps never above 45C even under load with my watercooling. Sounds bad doesn't it?
Update: Looks like it was a false alarm. Unplugged everything and let it sit overnight and now everything is working fine again. Weird.
I've been running 3.3Vdimm for about 2 months no probs, recently bumped upto 3.4Vdimm. I doubt the 3.3Vrail affects the CPU at all. Maybe the mobo.
Well, I really hope my FX53 is not going to die from high Vdimm voltages (I don't plan to go over 2.9v) and hopefully it gets along well with the NForce 3 250 chipset. *Crosses fingers*
i know a dude who is running 3.8vdimm for..... 4 months on his a64 now, no problem at all...
Hm I'm guessing there's still no real conclusion here, some say high I/O voltage is bad, others say that this voltage doesn't affect the cpu at all. So what's right and what is wrong ? help me ! :(
who said a high 3.3v rail is bad? :confused: only hjas_bjorth mentioned that and im not sure what he means but afaik a high 3.3v rail doesnt matter at all.Quote:
Originally posted by sai
Hm I'm guessing there's still no real conclusion here, some say high I/O voltage is bad, others say that this voltage doesn't affect the cpu at all. So what's right and what is wrong ? help me ! :(
just like most people here mentioned most of the cpus probably died from either bad mods ,bad mobo design or stupidity from using the bios to set a higher vdimm when you have a vdimm mod in place. all 3 things probably resulted in vdimm fluctuations wich killed the cpus.
my theory is that its a vref/vdimm problem. as most of you know all the new boards (afaik) have a fixed vref , or vref doesnt increase past ~1.4v (2.8v vdimm)
sometimes (:D) mobo makers know more than us and im sure they found out that an unbalance of vdimm and vref killed the onboard memory controller. that would perfectly explain all ive heard about it so far. a bad mod (too high pot=too big increasment from a small turn) a bad mobo design or chaning vdimm in bios when a mod was in place probably resulted in a fast jump in vdimm and short but strong fluctuations. vref probably couldnt keep trck and was still low while vdimm jumped up. then vref jumped up to keep track, but vdimm dropped down (fluctuations) and vref and vdimm bounced like this very fast for just a milisecond wich was enough to get vref high and vdimm low.
we all know about the dead video memory on videocards from too much vref, and i think its the same thing we are seeing here. maybe it wasnt really the memory that died on our videocards like everybody always believed, maybe it was the onboard memory controller of the gpus that died from too much vref, just like the memory controllers of the a64s are possibly dioeing from too much vref nowadays.
again vref alone doesnt matter. it only gets dangerous when it is more than 50% of vdimm, this also explains why some people can run such a high vdimm without a problem. either vref doesnt keep track and stays low while vdimm is high=no problem , or vref does keep track, but it always stays at 50% of vdimm, so its fine.
if this is true then on most boards running vdimm above 2.8v is even safer than running 2.8v! because on most boards vref doesnt increase afaik, so if you increase vdimm the cahnces that vref wich is fixed at ~1.4v gets higher than 50% than vdimm by accident or a fluctuation gets lower the higher vdimm is :)
just an idear :)
Can anyone put their word that running 3.3volts through the ram will NOT kill the cpu?
Mixed opinions here, what out ways what?
just read this last page, there are more than 6 people here who have run 3.3v or higher for months.
thats safe enough for me...
now i wish i had an a64 to prove it as well :/
good enough for me :D
good enough for me :D
been running 3v + since january and 3.3+ since march.. and yes saaya.. thats what I meant :)
hehe, I benched my 3400 with 3.9 running through the ram and it didn't die :banana: And I run it at 3.5v 24/7 and it still lives :)