[HassanDiab]

Hello every body in this forum.
i have a question about the safe voltage for i7 920 to run at least 2 years??
for me i set my one at 1.25V is in the safe margin or it will kill my cpu fast ???

BTW: my i7 920 is OCED to just 3.6 GHZ :S

[Bobnova]

Intel's warranty is good for 1.375, as per:http://processorfinder.intel.com/det...px?sSpec=SLBCH

That being a three year warranty, i'd say there is a 99.9% chance or better that your cpu will make it 2 years or more at that voltage or less, and an even better chance if you keep >30*c from TJmax.
The intel warranty allows that voltage sitting on tjmax for three years, which means they think it'll live through that. Personally, i wouldn't try it.

[Seanie's Show]

personally for such a small overclock you should beable to get those speeds out of all your stock voltages, I didnt find any need to increase voltages at all until I hit 3.8ghz on my i7.

If you did want to push further though and came to the point where you need to raise your voltages, VTT no higher than 1.35v mines at 1.3v, my NB is at 1.26v I prolly wouldnt put anymore than 1.3v through this, CPU PLL no more than 1.9v, mines at 1.84v, and CPU itself I wouldnt go higher than 1.4v, mines at 1.31v, all this gives me a nice 4.3ghz on my CPU, dont forget no matter how good your board is most of the time whilst stressing it voltages will fluctuate a bit especially vcore, just keep an eye on your temps, under stress like prime95 and LinX mines hit 85oC, tjmax is 100oC but like the other gentleman said, I wouldnt like to see mine up their.

A slight raise in CPU PLL will compensate for VTT and should either allow you to keep it down or low, but its a very dangerous voltage to be playing with and too much will kill your CPU for sure.

The general rule I have always stuck by for CPU voltage and NB voltage is never anymore then 0.2v increasement max.

Just so you know the i7 is a 45nm core just like my previous E8400, 1.45v VTT degraded that E8400 in 6 months to nothing more than a paper weight, I was getting memory errors left right and centre, I thought it was my ram as you do, I RMA'd it, there was nothing wrong with it, turned out to be the cache on the CPU, thats why im very very careful with voltages like VTT and PLL now, I can get 4.4ghz out of my i7 but the extra voltage jump to the CPU and the extra heat it produces isnt worth it, 95oC in LinX and a whole extra 0.1v for 100extra mhz, I can live without that.

[Brolloks]

Up till 4Ghz all you need to raise is v-core and you will probably need around 1.35 v, leave the rest of voiltages on auto and you will be good.
Voltages per se won't dagrade your CPU unless you go crazy, heat on the other side will, if your cooling solution is ineffectve then you might see some issues down the road...maybe.

[Seanie's Show]

Sorry I disagree with the above, to leave your voltages on Auto is one of the most stupid and worse things you can do, leaving them on auto means you dont know what your board is putting through the CPU to keep things stable, at least if you hard set them you have some idea, ok leaving vcore on auto you can monitor that with software, but not killing voltages like PLL, NB, VTT etc.

the following data sheet from intel pretty much backs up what I said in my first post so please dont be so stupid and put your voltages on Auto, your chip wont last you 2 months let alone 2 years.

[wingman99]

Originally Posted by Bobnova

The intel warranty allows that voltage sitting on tjmax for three years, which means they think it'll live through that. Personally, i wouldn't try it.

OEM PC's and laptops see that heat all the time that's why they built tjmax then shut down.

Originally Posted by Brolloks

Up till 4Ghz all you need to raise is v-core and you will probably need around 1.35 v, leave the rest of voiltages on auto and you will be good.
Voltages per se won't dagrade your CPU unless you go crazy, heat on the other side will, if your cooling solution is ineffectve then you might see some issues down the road...maybe.

I have a question about heat, where do you get your studies on electrical migration cased by heat. here on oveclock forms people keep there cpu's real cool?

[MattNo5ss]

Originally Posted by Seanie's Show

Sorry I disagree with the above, to leave your voltages on Auto is one of the most stupid and worse things you can do, leaving them on auto means you dont know what your board is putting through the CPU to keep things stable, at least if you hard set them you have some idea, ok leaving vcore on auto you can monitor that with software, but not killing voltages like PLL, NB, VTT etc.

the following data sheet from intel pretty much backs up what I said in my first post so please dont be so stupid and put your voltages on Auto, your chip wont last you 2 months let alone 2 years.

Voltages set to Auto is fine for low-mid OCs, but once you start to really push a chip to need to try to reduce them manually for your above reasons.

I ran 1.85v PLL and 1.6 Vtt on my EP45-UD3P for 24/7 use for over a year. That's what the board sets when the FSB was set to 500 or higher.

[Brolloks]

Originally Posted by Seanie's Show

Sorry I disagree with the above, to leave your voltages on Auto is one of the most stupid and worse things you can do, leaving them on auto means you dont know what your board is putting through the CPU to keep things stable, at least if you hard set them you have some idea, ok leaving vcore on auto you can monitor that with software, but not killing voltages like PLL, NB, VTT etc.

the following data sheet from intel pretty much backs up what I said in my first post so please dont be so stupid and put your voltages on Auto, your chip wont last you 2 months let alone 2 years.

I will not comment on your "stupid" comment so let that go...
I have worked with a fair share of CPU's and boards over some time and monitored voltages as they run under different types of load, bios' are programmed in 99% of the cases to be conservative when assigning voltages, so there is nothing wrong with setting on auto. I still have several OC'd CPU's running like that for a couple of years.

Please do not shoot statements or comments down because you read some data sheet, experience counts actually quite a bit on most people's books and maybe you should consider learning from that.

Originally Posted by wingman99

I have a question about heat, where do you get your studies on electrical migration cased by heat. here on oveclock forms people keep there cpu's real cool?

Firstly cool is relative, secondly read this ... LINK (scroll to the top of the book)

[bing]

The Auto setting actually is quite safe, although they're not at the optimal value.

Leave all those technical discussion apart, just common sense, imagine "IF" you're the product manager or whoever in charge for fixing the 'auto' value.

The question is, do you willing to risk your career by having tons of complains at one of your mobo product line that is killing cpu/ram/etc because of overvolted or "far" beyond standard specification for each components ? Also remember, you got yourself tons of other mobo product line/variant as well.

Ok, have to admit those auto could be wrong because of buggy or ill behaved bios, but I believe they should be minority. Still most of them will be implemented and set to a conservative value or at least not far from the component's specifications. Those components include cpu, chipset, ram , nic, audio chipset etc..etc.

Especially in these days in cut throat mobo business, the last thing you want to hear is nasty finding and publication at the net that your mobo AUTO value is spotted and measured by many2 fierce reviewers out there that they're far beyond standard spec and killing stuffs ... and tons of RMA as well.

My 2 cents

[Seanie's Show]

Originally Posted by Brolloks

Please do not shoot statements or comments down because you read some data sheet, experience counts actually quite a bit on most people's books and maybe you should consider learning from that.

So what you are saying is intels own data sheet about their own highly lab tested before sent to production CPU's is wrong, and I did learn from my own mistake 1.45vtt over 6 months killed my E8400 as per my first post in this thread, maybe you should consider reading that before dissing me too.

[Jizzzared]

You are the only person that "dissed" anyone, Seanie's Show. Brollocks calmly and respectfully explained where he was coming from.

Don't get so personally invested in this. It is only a computer forum.

I also agree with Brolloks, as I've had my PLL voltage on auto since I've had my curent computer with no issues. In the past, I have used voltages (vcore, nb, etc.) over Intel's (and AMD's) spec and have never had a cpu die on me in 7+ years of overclocking. Experience does play a role in this. I don't blindly believe everything I read.

Just because Intel says a certain voltage is safe and not to go over it does not mean that a little (or sometimes more than a little bit) over it will end up in disaster. If that were the case, you'd hear about a ton of cpu deaths. That is simply not the case, even on this forum, where people tend to use way higher voltages/frequencies than casual overclockers.

[HassanDiab]

so 1.25 V is safe for 24/7 machine with max Temp. 70??
i think it's under the Max Intel recommended VCORE.

VCORE: 1.25V
PLL: 1.81V
VTT: 1.35V
DRAM: 1.6V
ICH:1.3V
ICHPCIE:AUTO
IOH:AUTO
IOHPCIE:AUTO

BCLOCK:200MHZ
CPU MULTI.:18X (3600MHZ)
DRAM:1600MHZ
UCLOCK:3600MHZ
QPI:AUTO = 7200 MT/S
PCIE: 100MHZ
DRAM CLOCK: 7-8-7-20-1T

[Jizzzared]

Hassan,

I would say that those voltages are just fine. Does that overclock require 1.35 vtt for stability? I would think at your modest overclock you wouldn't need that much.

[MattNo5ss]

Originally Posted by Jizzzared

You are the only person that "dissed" anyone, Seanie's Show. Brollocks calmly and respectfully explained where he was coming from.

Don't get so personally invested in this. It is only a computer forum.

I also agree with Brolloks, as I've had my PLL voltage on auto since I've had my curent computer with no issues. In the past, I have used voltages (vcore, nb, etc.) over Intel's (and AMD's) spec and have never had a cpu die on me in 7+ years of overclocking. Experience does play a role in this. I don't blindly believe everything I read.

Just because Intel says a certain voltage is safe and not to go over it does not mean that a little (or sometimes more than a little bit) over it will end up in disaster. If that were the case, you'd hear about a ton of cpu deaths. That is simply not the case, even on this forum, where people tend to use way higher voltages/frequencies than casual overclockers.

Yeah.

Typically, when anything is designed by engineers they make room for 15-20% over what is needed, this is b/c there is always a small chance whatever they built could be loaded more than expected. So, going a little over spec shouldn't destroy your CPU.

[Seanie's Show]

Originally Posted by HassanDiab

so 1.25 V is safe for 24/7 machine with max Temp. 70??
i think it's under the Max Intel recommended VCORE.

VCORE: 1.25V
PLL: 1.81V
VTT: 1.35V
DRAM: 1.6V
ICH:1.3V
ICHPCIE:AUTO
IOH:AUTO
IOHPCIE:AUTO

BCLOCK:200MHZ
CPU MULTI.:18X (3600MHZ)
DRAM:1600MHZ
UCLOCK:3600MHZ
QPI:AUTO = 7200 MT/S
PCIE: 100MHZ
DRAM CLOCK: 7-8-7-20-1T

Voltages look fine but like Jizzzared said VTT is a little high, you can counter balance VTT with PLL so if you raise your PLL slightly say something like 1.84v / 1.86v you should beable to drop your VTT slightly down to around 1.28v'ish

[Brolloks]

Originally Posted by Seanie's Show

So what you are saying is intels own data sheet about their own highly lab tested before sent to production CPU's is wrong, and I did learn from my own mistake 1.45vtt over 6 months killed my E8400 as per my first post in this thread, maybe you should consider reading that before dissing me too.

I'm did not say Intel's specs are wrong, did I?...maybe you should read my post again and think about it before you post.

Just because you misjudged your settings and killed your CPU with 1.45vtt, which is crazy in any event for a dual core...any noob could tell you that, does not make you and expert does it now...

[Seanie's Show]

???

[wingman99]

Originally Posted by Brolloks

Firstly cool is relative, secondly read this ... LINK (scroll to the top of the book)

read page 137 they talk about temps in the range 300c 150c 200c
LINK:http://books.google.com/books?id=CoE...age&q=&f=false

[Brolloks]

The temps reported by the bios is not at substrate level where the micro circiuts reside and transistors are integrated, so it is much higher than reported, also long exposure to elevated temps have the same effect than short bursts of temps you referring to.

[Seanie's Show]

Originally Posted by Brolloks

I'm did not say Intel's specs are wrong, did I?...maybe you should read my post again and think about it before you post.

Just because you misjudged your settings and killed your CPU with 1.45vtt, which is crazy in any event for a dual core...any noob could tell you that, does not make you and expert does it now...

and this is from you guys: http://www.overclockers.com/forums/s...8&postcount=44

Didnt misjudge my CPU at all, their was nothing noob about it at all, things I dont care too much about I push to death, costly things on the other hand like my 920 i7 I do care about, but when I can pick a replacemnt up for £100 I will stop caring about it and push it to death again, I was overclocking PC's and components donkeys years before E8400's were thought of, I was unlocking AMD XP's 2500's to 3200 speeds and selling cheap on ebay and all sorts.

Take my Corsair x64 SSD for instance, I gave up waiting for corsair to get their firmware flash tool right and enable TRIM support so gave up caring bout the drive all together, Grabbed myself a Crucial M225 x64 SSD and flashed my Corsair with Patriot Torqx Firmware, thankfully I survived the flash and I now have the 2 drives RAID'd together. http://ssdtechnologyforum.com/Thread...triot-firmware