Is it as simple as turning up the multiplier and the voltages, or is there more to it? Also, what is a good cooler for the 2500k if i want to get it to 4.5 ghz or higher?
Overclocking a 2500k
- Thread starter dpk33
- Start date
MagnusTheBrewer
IN MEMORIAM
- Jun 19, 2004
- 24,122
- 1,594
- 126
MagnusTheBrewer
IN MEMORIAM
- Jun 19, 2004
- 24,122
- 1,594
- 126
All you need to do is to increase the processor clock frequency multiplier and the voltage accordingly or, you can run Gigabyte’s Quick Boost application.
So i can just turn up the multiplier and voltage right out of the box if i have either of those motherboards?
This article seems to say otherwise:
http://www.bit-tech.net/hardware/cpus/2011/01/07/how-to-overclock-the-intel-core-i5-2500k/4
This article seems to say otherwise:
http://www.bit-tech.net/hardware/cpus/2011/01/07/how-to-overclock-the-intel-core-i5-2500k/4
MagnusTheBrewer
IN MEMORIAM
- Jun 19, 2004
- 24,122
- 1,594
- 126
Depends on the BIOS version. If you're uncertain or don't want to take the time to learn the ins and outs of a specific MB, dl and run the quick boost application.
Puppies04
Diamond Member
- Apr 25, 2011
- 5,909
- 17
- 76
I can get 4.4ghz with stock voltages anything over that needs ever increasing bumps in voltage. In real terms the noticible difference between 4.4 and the 4.8-5 ghz you will max out at is very little so i would recommend staying at stock (or very close to) voltages.
BonzaiDuck
Lifer
- Jun 30, 2004
- 16,889
- 2,208
- 126
If my remarks don't provoke any heated argument, that's good. If they do -- it should be civil and I apologize for the distraction.
This has been discussed several times, and I think there is a clear agreement about what Intel has or has not chosen to publish. We can only speculate as to why one or the other.
They (Intel) always used to show two voltage-range specs: a "safe range" spec, and an "operable range" spec. Former a subrange of the latter.
For Sandy, unless something new has developed, they didn't really do that. So we come down to logical inference from known facts:
* Some Nehalem socket-1366 processors are made of 32nm silicon
* These seem to have a "safe-range" spec with an upper boundary of either 1.35 or 1.37V
* Sandy Bridge is built on 32nm silicon
From what I've seen here at Anandtech and elsewhere, there is a growing consensus among the more cautious enthusiasts that you're "safe" to keep VCORE within 1.35V.
If you follow the newer approach some of us have taken for over-clocking with EIST, C1E and some other features "enabled," and if you choose to leave VCORE set to "Auto" and overclock the "Turbo" mode, you should have two voltage settings that allow you nevertheless to control the voltage with no less precision than if you simply fixed the VCORE to a constant value. There are two "Offset" parameters -- a "+/-" signed value, and an amount in volts. Similarly, there should be a setting for "additional voltage in Turbo mode." There are also some four or five levels for using LLC, and settings for controlling the current (amperage) to the CPU as a percentage of default (100%
. In this regime of over-clocking, you are more likely to "see" the load voltage under "vDroop" during stress-testing, and less likely to see a value for "idle" voltage with the full clock speed (although it might show up at the beginning or end of a stress test in your sensor-monitoring software).
I generally take it that "load" voltage is the stress factor to watch, but Anandtech (December, 2008 -- "Over-clocking the QX9650") also points out the existence of voltage spikes at load-to-idle transition.
The "spec" is just that: a spec that addresses probabilities and expected lifespans for the processor. In this probabilisitic view of the world, chance of damage to your CPU for exceeding a VCORE spec by 0.01V to 0.02V (or so) may not be of great significance, and any risk may be infinitesimal. So if your load voltage seems to settle around 1.30 to 1.32V, and you observe a momentary "idle" voltage of -- say -- 1.36V when the clock is still at its "turbo" maximum, you're probably in pretty good shape (and some here might tell you "why so sissy, when you can push that ol' voltage higher!?")
Use your own judgment, and make your own choices. But I think what I say here is a pretty safe bet . . .
[EDIT -- some hours later:] you may want to check what I've said on another thread I just posted about "revisiting voltage limits" for the Sandy Bridge. . . .
Last edited:
So basically, whatever I can get out of 1.35 volts will be the max I can get? The heatsink and fan is a Thermalright Silver Arrow. Are there any other dangers of high voltages besides the heat?
BonzaiDuck
Lifer
- Jun 30, 2004
- 16,889
- 2,208
- 126
Yes. There is a process called "electro-migration." Higher than spec voltages will slowly eat away at the processor and degrade it. Someone published some photos taken with an electron microscope, showing the deterioration in the circuit traces.
Once again, the "spec" is Intel's choice based on their testing which is defined to eliminate RMA returns as much as possible. Thus they would strive for 0% chance of failure over a three year time-horizon -- on average. The processors, on the other hand, are supposed to have a longevity of ten years. I've seen all sorts of components -- motherboards, PSUs, memory -- go bad after some period of time, but I've never personally seen a processor go south. And that includes my (cautiously) overclocked units, but excludes anyone else's wildly over-volted CPUs.
Some say that voltage is more of a threat than heat. Temperatures scale exponentially with higher voltage, and only linearly with higher frequency. You could fool yourself with water-cooling, stay within the thermal spec, and over-volt the processor to a premature end.
But, as I said, I think you should be "good to go" if you keep both load and idle voltages within a 1.35V limit -- or close to it. Others will tell you I'm giving "sissy" advice about this. You could -- as in the Lou Reed song -- "take a walk on the wild side" if you have no care about replacing a processor in a couple years. And it might not be just a couple years -- it could be longer. It's a roll of the dice, varying with the variability of what rolls off the assembly-line.
As I said -- probability is all about uncertainty. If you flirt enough with the uncertainty of a processor going south, you could find yourself married to a cluster of uncertainties about "what went wrong" when you start troubleshooting the computer. And if you're going to dance with extreme settings, it would be a better bet to limit the "out-of-spec" item to a single thing. Or none at all.
The very good thing about the Sandy Bridge processors and the chipsets that support them: you won't be running the processor at its full over-clock speed and voltage 24/7 -- as we did in earlier generations of processors. Over-clocking in "Turbo-mode" with both EIST and "turbo-boost" means that the voltage and speed scale on demand. So there would be considerably less stress on the CPU over months and years.
To recap -- you could choose to see what you can "get out of" 1.40V, 1.45V, etc. It's your processor; you can do what you want. I just don't see the sense in going much out of the voltage spec or just beyond "voltage common-sense" when I use my computer on a day to day basis, expect reliability, and have long wearied of troubleshooting in an extreme-performance regime. I'll take three months to build it and tune it, but I'll want to rely on it for the expectation of four years or so -- even if I choose to upgrade after only two.
Last edited:
So with all of this in mind, would 1.35v be enough to bring the cpu to 4.5ghz? And what's the average voltage needed for 5 ghz?
toyota
Lifer
- Apr 15, 2001
- 12,957
- 1
- 0
and don't you also run your memory at 1.65? just because you are doing stupid crap and it has not killed the chip does not make it wise to do so.
I used to push the envelope too until 2 of my CPU's died. (after a year and change of use) Oh well, gotta learn somehow.
He asked how much voltage for 5ghz.1.45 volts with 3 radiators and only on load is sooo stupid!!!
Iv let it run prime for 24 hours just to burn my cpu in and not to see if its stable also.
My ram has been tested to run on sandy,it even says made for sandy in the add at newegg.my board defualts it to 2133 and 1.65 volts.
Show me one freaking sandy with a dead memory controler from running high memory and nit 1.7 volts going for 7ghz on ln2
Last edited:
Your error is that you think CPU's die overnight. Its a slower process than that, pushing the envelope continually lowers the lifespan of the cpu -- I have 2 dead cpu's as proof of this. But hey, if it works for you whatever.
Also, the only RAM "made" for sandy bridge will always be 1.5 volts. Your BIOS may apply an incorrect voltage but SB ram will always be designed for 1.5v.
Last edited:
toyota
Lifer
- Apr 15, 2001
- 12,957
- 1
- 0
please link me to where it says your ram was made for Sandy Bridge. if it was then no way should it default to 1.65 and anyone that knows anything at all about Sandy Bridge knows that. again do whatever you want to your components since they are yours to destroy. and whether its in one week or one year, they are being damaged. and I sure as heck hope you do not ever intend to sale them to anyone else later.
Last edited:
I will and this chip was bought a week after launch and has been used everyday since with most days on all day.
Its the family gaming rig and my bencher is a 2600 setup.
Your dead cpus have nothing to do with my ram at 2133
There u go.mine are a little better but they are sold out and not listed.
These are made to support intel extreme memory 2133 xpm profile
You do know that intel is going native 2133 with ivybridge right?
Anyway here is one of many links with 2133 ram made for p67
http://www.newegg.com/Product/Produc...82E16820231435
toyota
Lifer
- Apr 15, 2001
- 12,957
- 1
- 0
so why do they have 1.65 memory for Sandy Bridge when it officially calls for 1.5?
and Ivy Bridge is going to from 1333 to 2133? I have not heard of that so do you have a link to anything official?
Every chip is different and so is every MB. I need 1.375 on my Gigabyte to hit 2500k stable at 4.5G. The voltages mentioned by Bonzai are very conservative so you'll be safe if you stay with them. However that debate has raged for every processor made and historical evidence sides with it being very difficult to burn out a CPU. Google max voltage for Q6600 and you'll see similar very low voltage recommendations from posts several years ago. You'll see posts years later of people running much higher voltages for years and still going strong. The real answer is no one knows for certain so what's your risk tolerance?
I read it online a while ago that intel 2133 profile is coming out.There are a lot of articles posted online about ivy going native 2133 and will end up supporting up to 3000.
here is a pic so people dont have to see the link,you can see it says made for intel xmp certified profile and defaults to 2133 and 1.65 volts
lol it even says
INTEL CERTIFIED
,is that enough proof that maybe me running my ram at 2133 is safe?
here is a pic so people dont have to see the link,you can see it says made for intel xmp certified profile and defaults to 2133 and 1.65 volts
lol it even says
INTEL CERTIFIED
,is that enough proof that maybe me running my ram at 2133 is safe?
Last edited:
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 24K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter