Maximum X2 Voltage

[iSkylla]

Hey guys, this is my first post on XS and I would like to start out with a question :) I have gotten different replies on [H]ardForums so I thought I would try it here.

I recently bought an X2 3800+ and got the stepping LDBHE 0604WPAW. I have been overclocking it (very addictive I might add), and I managed to get to 3.0Ghz with 1.623V. It wasn't stable but at least I got into windows and validated CPU-Z (http://valid.x86-secret.com/show_oc?id=85571). I have a scythe ninja that gives me quite good temps, at 1.5V (32C/41C) and at 1.623V (39C/49C). I want to try to get a stable 3.0Ghz for benchmarking purposes and maybe to run it, but I am concerned about giving it 1.65-1.7V. Some people say it is bad to give a CPU that much voltage because it will damage it (no matter what the temps) and others say that it is fine as long as the temps are below 55C-60C. So I thought I would ask here. I know there are more experienced and more overclockers here so I would appreciate your responses. Lately I noticed the steppings have been really good clockers, I just saw a post about a 3.5Ghz X2, incredible. Maybe the X2s are making a comeback!

Thanks,
iSkylla

[iSkylla]

C'mon, I could use an answer ;) I know people have opinions concerning this at the very least.

[Delirious]

The general consensus is 1.5 on air and 1.5-1.6 on water.

If you have money to buy another if this one craps out then run it at whatever you want. :D

[SoF]

here is an amd section as well - better to ask such questions there my friend ;)

anyway I agree with Delirious

[MoTiG]

i think that the best choice for normal user is the max OC at default vcore.
Those X2 chips are too much expensive to risk them putting 1.6v or more to squeeze 50mhz from them.

I am currently at 2600MHz 1.36v and it's just fine.
To get 2700MHz stable i have to put 1.45v, but it's not worth it imo.

[ady1989]

this might be a bit off topic, but fits in here and i didnt want to waste a new thread on it. i just bought an X2 4400+, what would be the max save vcore for this chip with high end air cooling? would it be the same as the x2 3800+? thanks!

[iSkylla]

Well those chips have more heat output, 104W vs 89W. I say 1.5V. For my chip, I usually say 1.55V but that is why I am asking.

[Delirious]

i think that the best choice for normal user is the max OC at default vcore.
Those X2 chips are too much expensive to risk them putting 1.6v or more to squeeze 50mhz from them.

I am currently at 2600MHz 1.36v and it's just fine.
To get 2700MHz stable i have to put 1.45v, but it's not worth it imo.

That is a very unextreme thing to say :p:

There is a point of diminishing returns, but 1.45v is good for 2.7, wanna trade with me? :D

[pete5990]

I've heard that temperature is the only thing that truely matters, not voltage.

[madgravity34]

Not true at all, too much voltage can kill a cheap even if its phase/ln2/DI cooled. I wish we could get chew to comment in this thread, but he doesnt run his cpu's ON PHASE past 1.6v for 24/7 use.

[iSkylla]

Yeah, let's get Chew in here. Honestly, I was just doing it to get a CPU-Z screenie and see if I could get a SuperPi run, but I couldn't. I currently run 2.8Ghz @ 1.5V which is fine for me. I am going to try to see how low I can set the voltage to and still run at 2.7-2.8Ghz.

[afireinside]

Not true at all, too much voltage can kill a cheap even if its phase/ln2/DI cooled. I wish we could get chew to comment in this thread, but he doesnt run his cpu's ON PHASE past 1.6v for 24/7 use.

On the other hand I ran 1.68v 24/7 for months on air and now water and HK police has ran 1.7v 24/7. I've benched up to 2.05v. I've owned 15 AMD CPUs and ran each to 2v+ without a single death or degrade in performance.

iskylla is that the 3800+ I sold you? Glad to see it's working out :)

[iSkylla]

It is! I thought, hmm, cherry hill, insane ocer, wait it has to be him. It's been great. You should come over to my house lol. Help me out a bit. I got it up to 3Ghz on 1.62V but it crashed shortly after. I would like to run 24/7 @ 2.9Ghz but it needs like 1.6V. I am afraid to run it because I don't want to kill it. But as I said, the chip is very cool.

One other thing, some people say that the DFI temp sensor is off. I actually happen to disagree because the one time I forgot to plug in the fan for the ninja, it was running passivly 45C/53C and the heatpipes got very warm/hot. But it normal use the heatpipes are quite cool and during dual prime sometimes they get slightly warm.

[Delirious]

I dont think you can use that as proof that the temp sensor is accurate.

Its a known fact that motherboard temp sensors are far from accurate.

A better test would be to do what maxxracer did, cut a groove out of the ihs, place thermal probe in the groove, apply hs and measure temps.

[Plywood99]

Been running my opty 165 @ 1.65v bios on water for several months now. Not a single problem...

Lol, just noticed , opty 165 at 1.65. Get it???

[albumleaf]

I've been running Boinc (100% Load) @ 1.72v for three months. No issues on a XP-90/Venice 3000+

[aug1516]

I sort of thought that anything above 1.5v on air cooling was pushing it but I guess that really just depends on your temps. Some of the voltage numbers posted so far as quite surprising.

[cantankerous]

I have 1.65v going through my Single core Winchester and have done so since Feb. 2005 24/7. Still going strong!

[knob]

for me 1.55 is the max. I dont wat to go any higher eventhough i am using water cos heat can be an issue

[Raphie]

Venice boyz, the question is X2 only... I run my 4400+ @ 250 * 10 @ 1.45v. It needs 1.575 to get 2600Mhz which ain't worth it. BTW i only consider a system stable when it's dual affinity Prime95 stable for 12+ hours.

Running a dual 32M SuperPI means crap, running 3 hours of Prime (though you getting close then) also means nothing.

[knob]

Venice boyz, the question is X2 only... I run my 4400+ @ 250 * 10 @ 1.45v. It needs 1.575 to get 2600Mhz which ain't worth it. BTW i only consider a system stable when it's dual affinity Prime95 stable for 12+ hours.

Running a dual 32M SuperPI means crap, running 3 hours of Prime (though you getting close then) also means nothing.

i agree. Its up to a person when they talk abt stability. It only gets me is when ppl try to sell their chip and they put screniees of 32 M SuperPI and say that i can hit tht speed and its fully stable :slapass:

[Raphie]

Yep, and that all at custom and not SMALL FFT's:slapass:

Same thing with the Vcore, people running 1.52v in CPU-Z or Smartguardian had to put 1.586v in their BIOS Vcore settings. So again what are we talking about?

What Smartguardian or CPU-Z (or BIOS) reads, OR what yuo've put in to the BIOS.

with DFI a 1.55v readout means a 1.65v setting

[masterofpuppets]

Been running 1.612v in BIOS almost 24/7 for a week or two, and it's fine.

[albumleaf]

with DFI a 1.55v readout means a 1.65v setting

Thanks for your blanket statement. This isn't true all the time, esp. with my NF3

[Delirious]

with DFI a 1.55v readout means a 1.65v setting

Have u even measured the vcore on the board with a digital multimeter?

when i set 1.5v vcore in my bios the true vcore voltage according to a multimeter is 1.50, under full load is when dfi's overvolt and then its only by 0.03v on my board. cpu-z says its 1.472v.

[Raphie]

I haven't measured it on my Ultra-D, i only notice that the values as showing in the CPU VID special tables, are higher than the voltages shown in BIOS or in CPU-Z / Smartguardian. I have no experience on the other models, so point about the blanket statement taken.

[lawrywild]

Have u even measured the vcore on the board with a digital multimeter?

when i set 1.5v vcore in my bios the true vcore voltage according to a multimeter is 1.50, under full load is when dfi's overvolt and then its only by 0.03v on my board. cpu-z says its 1.472v.

correct

[RLM]

Yep, and that all at custom and not SMALL FFT's

Actually Raphie, not sure if I'm reading you right, but I have found the reverse to be true. As per my results in the OC database, I had to up my vcore by 0.025 to achieve stabilty in a custom test over small FFTs.

I'm only guessing but that seems to make sense to me. After all wouldn't the cpu come into play with ram work, especially considering that the mem controller is on the cpu? In addition how do you limit your 24/7 usage to "mainly cpu/a bit of ram". You know what I mean? Custom seems to be more relevant to 24/7 usage. Just my thoughts anyway.

RLM

[Raphie]

We're understanding eachother perfectly, CUSTOM (blend with specified memory settings i.e. 800Mb per core with 2GB) is a lot harder than small FFT's so we're on the same page:toast:

for me custom dual prime 12h+ is the only real life benchmark. because if you can make something fail, that doesn't fail on a lower clockspeed your system is not stable.

[RLM]

Oh that's good. LOL:toast:

[Delirious]

Speaking of custom prime testing, i have 2gb of memory and have been using 820mb per core, how much does everyone else use?

We're understanding eachother perfectly, CUSTOM (blend with specified memory settings i.e. 800Mb per core with 2GB) is a lot harder than small FFT's so we're on the same page:toast:

for me custom dual prime 12h+ is the only real life benchmark. because if you can make something fail, that doesn't fail on a lower clockspeed your system is not stable.

[RLM]

I think I'm cheating 700mb.:D