Overclock.net banner
1 - 20 of 24 Posts

·
Premium Member
Joined
·
3,574 Posts
Discussion Starter · #1 ·
This has been bothering me for a long time, some have killed 45nm's with 1.4v+ while many haven't. Some report degrading while again many haven't. My sig CPU can stabilize 4Ghz at about 1.41 ~ 1.42v load with temps of 71c, sure its warm but its within specifications so iam happy with that part. Also aware of the loadline calibration debate, reguardless if enabled or disabled the vcore required for all speeds is the same for me.

At 3.8Ghz now its at 1.31v stable, I have a few games, especially one thats very CPU demanding.. so thats my reason for pushing it more. I can't make up my mind if I should go back to 4Ghz 24/7 (I did it a few months ago for about 2 weeks). People with 45nm's that are constantly above 1.4v 24/7 what are your opinions? Has yours ever been subject to this "degrading" thing? I read many articles, OCN threads, everything, its just adding to the confusion.
 

·
Registered
Joined
·
5,508 Posts
The way I look at it, what would you rather, 1.3625 with LLC on or 1.4v + without LLC.

Also, I'm pretty sure with LLC enabled even at 1.3625v the processor could handle some spikes here and there and would probably never hit over 1.45v.

I'm not exactly sure what happened to my e8400 but I ran it at 1.3625v for the longest time (on load) with a vdroop modded board and for some reason after about 6months to a year it wouldn't maintain it's 4GHz overclock.

But then again, Intel Burn Test wasn't out a year ago, so I'm not really even sure it was stable to begin with.

I guess it's all what you're comfortable with.

For me, I will never pass 1.3625v, ok, well maybe I would go up to 1.4v but I'd like to think that 45nm dual's are more sensative than 45nm Quads.

But personally I think 1.3625v even with llc is fine, so theoretically I would be ok with running up to 1.4v but nothing more.

I run 4.2GHz @ 1.31v with LLC on so I really don't need to use more vcore at the moment. But I think as a general rule of thumb its wise to stay as close to the 1.3625v max as possible, and as close to the 71c max as possible.

jam3s
 

·
Registered
Joined
·
1,710 Posts
Quote:


Originally Posted by GigaByte
View Post

This has been bothering me for a long time, some have killed 45nm's with 1.4v+ while many haven't. Some report degrading while again many haven't. My sig CPU can stabilize 4Ghz at about 1.41 ~ 1.42v load with temps of 71c, sure its warm but its within specifications so iam happy with that part. Also aware of the loadline calibration debate, reguardless if enabled or disabled the vcore required for all speeds is the same for me.

At 3.8Ghz now its at 1.31v stable, I have a few games, especially one thats very CPU demanding.. so thats my reason for pushing it more. I can't make up my mind if I should go back to 4Ghz 24/7 (I did it a few months ago for about 2 weeks). People with 45nm's that are constantly above 1.4v 24/7 what are your opinions? Has yours ever been subject to this "degrading" thing? I read many articles, OCN threads, everything, its just adding to the confusion.


In my opinion, not worth the gamble for the .3 or so GHZ you are achieving. And it is a gamble.

Benchmarking and other short runs at that voltage might be fine, but I wouldn't for 24/7 usage.
 

·
Registered
Joined
·
2,688 Posts
I don't think its worth it for 200 MHz, you won't even notice the difference.
 

·
Premium Member
Joined
·
8,040 Posts
I'm running the same CPU as you and also at 3.8ghz with 1.30625vcore in bios, 1.28v cpuz. It is 1000% stable and everything I do runs so smoothly at these settings I love it. I think about getting those 200mhz extra for 4ghz but at my current settings it does everything I need and I would hate to lose that stability I have. I have even been able to tighten my ram timings to 4-4-4-12.

So imho, 3.8ghz and good temps is where it's at!
 
  • Rep+
Reactions: GigaByte

·
Registered
Joined
·
5,508 Posts
Quote:


Originally Posted by Berger
View Post

LLC just stabilizes and corrects the voltage......it doesn't make it harder on the cpu......

There have been reports at XS that LLC is not good for 45nm CPU's.

Reason being is that when you enable LLC you are disabling intel specification of vdroop.

The purpose of vdroop is to offset any erratic voltage spikes that may occur at any given time on load.

65nm architecture can handle the effects of vdroop, while many believe that the 45nm's (especially dual cores) are more sensitive to these spikes.

I however need LLC in order for my voltage to be even somewhat stable. I am comfortable with mild vcore spikes but I wouldn't run anything over 1.3625v 24/7

jam3s
 

·
Premium Member
Joined
·
6,985 Posts
+1 to what everyone is saying!
 

·
Premium Member
Joined
·
3,574 Posts
Discussion Starter · #9 ·
@mtbiker033

Did you try 4Ghz before? If so what vcore did it pass small FFT over 6 hours? I haven't done much tweaking at 3.8Ghz for vcore, but I do know if I reduce it 1 more notch it fails at about the 5 hour mark. I may stay at 3.8 after all and try to mess around with VTT since GTL/PLL volts are hidden within it (my board hides those settings as they aren't in bios). VTT will be below 1.32v at all times since P35 absolute max is 1.32v. Chipset limit before the CPU limit 1.45v.

@jam3s LLC is disabled, since it doesn't affect vcore needed to stabilize for my system I leave it disabled, rule out any possible problems from that.
 

·
Premium Member
Joined
·
2,533 Posts
I don't hesitate on raising the voltages for benching, but I wouldn't do the same for 24/7 use. The real-world performance gain is negligible, and and doesn't really justify such a constant high voltage. That's probably because I like low temperatures, but to each his own.
 

·
Bacon & Eggs
Joined
·
2,865 Posts
I do it easy. 1.425 in bios, 1.395 in CPU-Z, 1.384 underload. Gonna kill the chip, nope. Again, a chip set to run for 10 years at 1.3625 max running at 1.4 kills a year or two off the chip?? You gonna use your chip for 10 years, no.
 

·
Premium Member
Joined
·
2,533 Posts
Quote:


Originally Posted by h00chi3
View Post

I do it easy. 1.425 in bios, 1.395 in CPU-Z, 1.384 underload. Gonna kill the chip, nope. Again, a chip set to run for 10 years at 1.3625 max running at 1.4 kills a year or two off the chip?? You gonna use your chip for 10 years, no.

With a high-end watercooling setup I wouldn't mind. But not on air.
 

·
Registered
Joined
·
973 Posts
Quote:


Originally Posted by jam3s
View Post

There have been reports at XS that LLC is not good for 45nm CPU's.

Reason being is that when you enable LLC you are disabling intel specification of vdroop.

The purpose of vdroop is to offset any erratic voltage spikes that may occur at any given time on load.

65nm architecture can handle the effects of vdroop, while many believe that the 45nm's (especially dual cores) are more sensitive to these spikes.

I however need LLC in order for my voltage to be even somewhat stable. I am comfortable with mild vcore spikes but I wouldn't run anything over 1.3625v 24/7

jam3s

This is very true.

Vdroop is all in the board. Typically, higher-end motherboards are going to allow less vdroop, thus LLC can be turned off, and you will still have good voltage running. Alternatively, LLC when on can overcompensate for vdroop and that's where the cooked chips are coming from.

For instance, I went from an Asus P5K (which is not a bad board by any means) at 1.38v with LLC enabled to my current DFI x38-T2R at 1.392v with no LLC. The difference is really in current management on the board.
 

·
Bacon & Eggs
Joined
·
2,865 Posts
Quote:


Originally Posted by Shift.
View Post

With a high-end watercooling setup I wouldn't mind. But not on air.

I am doing it, and this is on my folding box, with a Noctuna. But my load temps are not above 49.
 

·
Registered
Joined
·
5,508 Posts
Quote:


Originally Posted by h00chi3
View Post

I do it easy. 1.425 in bios, 1.395 in CPU-Z, 1.384 underload. Gonna kill the chip, nope. Again, a chip set to run for 10 years at 1.3625 max running at 1.4 kills a year or two off the chip?? You gonna use your chip for 10 years, no.

like shift said, the real world difference of say 4.0GHZ @ 1.3625v vs 4.2GHz @ 1.4v is negligble, and you may as well just leave it at 4.0GHz, since it's lesser volts.

I can run my processor at 1.225v for 3.8GHz 100% stable

I have to use 1.31 for 4.2GHz but I can afford the higher temps and the higher vcore primarily because i'm going to be watercooling it.

I guess some people just value their hardware that they buy.

Running over 1.3625v IMHO, will land you either a dud chip or a degrading one, slowly but surely over time.

/jam3s
 

·
Bacon & Eggs
Joined
·
2,865 Posts
Quote:


Originally Posted by jam3s
View Post

like shift said, the real world difference of say 4.0GHZ @ 1.3625v vs 4.2GHz @ 1.4v is negligble, and you may as well just leave it at 4.0GHz, since it's lesser volts.

I can run my processor at 1.225v for 3.8GHz 100% stable

I have to use 1.31 for 4.2GHz but I can afford the higher temps and the higher vcore primarily because i'm going to be watercooling it.

I guess some people just value their hardware that they buy.

Running over 1.3625v IMHO, will land you either a dud chip or a degrading one, slowly but surely over time.

/jam3s

I am 100% stable, but like I said, this is on my folding box. The difference between % on a 2669 WU... 4.0 is 8 minutes a percent, at 4.25 it is 6 1/2, and the only thing my [email protected] box in my sig rig does is fold.
 

·
Registered
Joined
·
824 Posts
70c is too high, I would skip the last 200mhz unless you get the temps under control. If you could get the temps down then I don't see a problem running 1.4v 24/7 mine seems just fine running there.
 

·
Registered
Joined
·
643 Posts
I think this really depends on the amount of money you have and how much you paid for the processor in the first place. I had a e7200 that I paid $60 for. However I still kept it at 1.36v. With the e8400 I have that I payed $100 for I still wont run it over 1.36v 24/7. I did do some benching at 4.6ghz 1.43v though, but only for like 20 minutes. I don't even notice much of a difference from 4ghz to 4.4ghz. The whole diminishing return factor starts to come into play at some point.
 

·
Premium Member
Joined
·
3,574 Posts
Discussion Starter · #20 ·
I think I am convinced enough now, iam going to tweak around the volts for 3.8Ghz and look at 3.9Ghz as thats looking interesting at 1.344v with 68c load. +REP
 
1 - 20 of 24 Posts
Top