i5 2500k currently at 4.5ghz at 1.3V - (why) should I go higher?

O

ordovician

2[H]4U
Joined
Dec 28, 2007
Messages
2,625
I'm just playing around with the rig in the sig. 4500mhz at 1.3V using "Regular" LLC and "Optimized" VRM Phase Control is pretty damn good. I'm seeing temps of 52C during BC2 and 58C during Intel Burn Test. It just passed 10 runs of IBT.

I got a BSOD at 4700 mhz with the same voltage settings, but setting a more aggressive load line calibration and phase control will allow it, I'm sure. The question is why? This thing is already noticeably faster than my i7 930 at 4ghz, and more mhz do not seem to be improving fps...

I'm paranoid about pushing voltage much past 1.3, and I'm paranoid about messing up what I have achieved by damaging or degrading either the proc or the mobo by pushing it a little further, running the stress test, etc, etc. But that magic 5G[H]z number beckons...

What to do!??
 
Honestly, if you don't see a real-world improvement and you're happy with your current speed, voltage, and temps, why push it? 4.5ghz is still a pretty respectable overclock for that voltage.
 
i settled on 4.5ghz at 1.3V as well. i'll attempt to push it further if the time comes when i feel i need more processing power. i did something similar with my Q6600; ran it at stock for a couple years and then oc'd when i felt i needed more oomph.
 
I was running 4.8 at 1.39, but did 1.43-1.45 to 5ghz lol. just to have it. Its a preference
 
1.3V and 58C load gives you plenty of room to play with. 5GHz represents a ~50% overclock while 4.5 is "only" a 36% overclock. Set your voltage to 1.4V and see how high that gets you, incrementing multi by 1 each time.
 
I changed the LLC to "Medium" and actually dropped my volts by 0.005... now I'm sitting at 4.6ghz IBT 10-pass stable at 61C and 1.312-1.320V.
 
ordovician said:
I changed the LLC to "Medium" and actually dropped my volts by 0.005... now I'm sitting at 4.6ghz IBT 10-pass stable at 61C and 1.312-1.320V.
Click to expand...

Prime95 blend is your friend. I can pass IBT or Linx and crash within Prime95 blend within minutes. Let Prime95 blend run for 3 hours then you are good to go.
 
In particular I've found large FFTs 1280k and up to BSOD the soonest if there is any instability. YMMV.

These tests don't even start until 8.5 hours into the blend test, so do a custom torture test with minimum = 1280k and the rest default.
 
I'm running 4.5ghz at 1.26vcore at around the same temps...

Should I put vcore to 1.3? Is it needed?
It passes IBT at max without problems.
 
I'm at 5 ghz here. Because you can? lol. It's more performance. Anything you get and keep getting without causing distress or problems is free.
 
ordovician said:
I'm just playing around with the rig in the sig. 4500mhz at 1.3V using "Regular" LLC and "Optimized" VRM Phase Control is pretty damn good. I'm seeing temps of 52C during BC2 and 58C during Intel Burn Test. It just passed 10 runs of IBT.

I got a BSOD at 4700 mhz with the same voltage settings, but setting a more aggressive load line calibration and phase control will allow it, I'm sure. The question is why? This thing is already noticeably faster than my i7 930 at 4ghz, and more mhz do not seem to be improving fps...

I'm paranoid about pushing voltage much past 1.3, and I'm paranoid about messing up what I have achieved by damaging or degrading either the proc or the mobo by pushing it a little further, running the stress test, etc, etc. But that magic 5G[H]z number beckons...

What to do!??
Click to expand...

You sort of answered your own question. You got 4.7 with the same voltage settings but bsod.
Now put the ratio on 47 again. Up the vcore by 0.025 till you get a pass in prime95. Just run the quick test for 5 min. If it fails up by 0.025 again till you get a pass that's all
 
Iching said:
Prime95 blend is your friend. I can pass IBT or Linx and crash within Prime95 blend within minutes. Let Prime95 blend run for 3 hours then you are good to go.
Click to expand...

Blahman said:
In particular I've found large FFTs 1280k and up to BSOD the soonest if there is any instability. YMMV.

These tests don't even start until 8.5 hours into the blend test, so do a custom torture test with minimum = 1280k and the rest default.
Click to expand...

I do IBT, Prime, and OCCT. Just not after every single iterative jump in volts/freq.

Spooony said:
You sort of answered your own question. You got 4.7 with the same voltage settings but bsod.
Now put the ratio on 47 again. Up the vcore by 0.025 till you get a pass in prime95. Just run the quick test for 5 min. If it fails up by 0.025 again till you get a pass that's all
Click to expand...

I think there is some misunderstanding about the point of this thread. I am not looking for help overclocking - and although I appreciate the advice, I know exactly what I am doing and how to do it. My quandry is whether it is really worth it or not.

I have concerns about pushing the Load Line Calibration, voltage, and other VRM settings too high because nobody (not even Intel, and they have admitted that) has compiled enough statistical data on what kind of voltage and amperage can be pumped through the SB procs 24/7. I'm not spreading or perpetuating FUD, but I am highly skeptical that these things should have anything over 1.325V running through them. Hell, anything over 1.3V on the 45nm i7's of yore could cause degradation and even failure if you were really unlucky.

The Asus rep here seems to suggest people can "fire and forget" with a 1.41V at 4.8 Ghz cookie-cutter OC. That's great if you're a board maker, but not so great if you're a CPU that has to last more than 4 months.

So, my question stands: Why run 4.8-5.0 at ~1.38-1.42 V when I can do 4.5 at 1.3 V?
 
Last edited:
ordovician said:
I do IBT, Prime, and OCCT. Just not after every single iterative jump in volts/freq.



I think there is some misunderstanding about the point of this thread. I am not looking for help overclocking - and although I appreciate the advice, I know exactly what I am doing and how to do it. My quandry is whether it is really worth it or not.

I have concerns about pushing the Load Line Calibration, voltage, and other VRM settings too high because nobody (not even Intel, and they have admitted that) has compiled enough statistical data on what kind of voltage and amperage can be pumped through the SB procs 24/7. I'm not spreading or perpetuating FUD, but I am highly skeptical that these things should have anything over 1.325V running through them. Hell, anything over 1.3V on the 45nm i7's of yore could cause degradation and even failure if you were really unlucky.

The Asus rep here seems to suggest people can "fire and forget" with a 1.41V at 4.8 Ghz cookie-cutter OC. That's great if you're a board maker, but not so great if you're a CPU that has to last more than 4 months.

So, my question stands: Why run 4.8-5.0 at ~1.38-1.42 V when I can do 4.5 at 1.3 V?
Click to expand...


I do not have an answer for you, in fact, I dont know many who DO have an answer to that question. I have searched and questioned the amount of voltage to pump into the SB. I had 4.8 running at 1.38-1.39 volts and just wanted 5ghz. I am pumping the voltage only because I wanted to run it at 5. Is it safe? I do not know. I am monitoring my temperatures each day. I have had the system running for a couple weeks so far and have not noticed any decrease in performance.

Will it decrease? probably one day yes. I guess that if I keep the temperatures low I think ill be ok for the time being.

If you do seem to find an answer please post it, or if anyone has a logical explanation or reasoning to why they run it at their volts please do share.

Right now, as it stands, there are few who can argue the point of runing their chips at 1.45+ volts, hell even 1.4+ volts. Myself, ill keep mine at 5ghz as long as temps are below 65C - which they are luckily.

I was lucky and sold enough of my previous rig and worked on the side to set aside cash for a just in case scenario or CPU death lol.

So if it dies, ill purchase another one. Hell, if it does die I may have to sell some more things on craigslist and purchase a 2600k for fun.

Thats all the rambling I can do for now. Good luck to all. Im going to keep looking at this thread until it dies haha.
 
I stopped at 4.6 myself for now. My chip doesn't seem to be the greatest as I need around 1.36V to get it stable at that voltage, but I did keep turbo mode enabled so I may be losing out on 100-200MHz. But honestly, I'd rather have ~1V at near idle usage than a constant 1.36V with 4.7-4.8GHz.

Imo it's like shades of Wolfdale - we all went crazy over them and they started to degrade after a few months to a year. Not because of high temperatures... but because everyone was pushing too much voltage through the 45nm chips. (And we are now down to 32nm)
 
blade52x said:
I stopped at 4.6 myself for now. My chip doesn't seem to be the greatest as I need around 1.36V to get it stable at that voltage, but I did keep turbo mode enabled so I may be losing out on 100-200MHz. But honestly, I'd rather have ~1V at near idle usage than a constant 1.36V with 4.7-4.8GHz.

Imo it's like shades of Wolfdale - we all went crazy over them and they started to degrade after a few months to a year. Not because of high temperatures... but because everyone was pushing too much voltage through the 45nm chips. (And we are now down to 32nm)
Click to expand...

I do not recall anything about problems with Wolfdales. I am sticking to 4500MHz. I need a massive jump (1.4V) in order to reach 4.8GHz.
 
Why are you stopping at 1.35V?

1.35V was the safe-zone for 1156/1366 processors.

Sandy Bridge can hit up to 1.375-1.38V for 24/7 use.

And very few, if any, hit 5GHz 24/7. They only do 5+GHz for benches when they bump up volts to 1.4V and only have it running for a few hours.
 
Undead46 said:
Why are you stopping at 1.35V?

1.35V was the safe-zone for 1156/1366 processors.

Sandy Bridge can hit up to 1.375-1.38V for 24/7 use.

And very few, if any, hit 5GHz 24/7. They only do 5+GHz for benches when they bump up volts to 1.4V and only have it running for a few hours.
Click to expand...

I assume you're getting that info from a quick google search. The 1.38V number is the MAX voltage that the average Sandy Bridge CPU can take for a reasonable period of time without damaging itself. My understanding is that this does not necessarily mean your CPU can be pegged at that voltage for an extended period of time. It's akin to saying that a PSU's peak output is its rated wattage, which is a no-no.

As an Intel engineer (but not in any way am I a company spokesperson), I have to say that I appreciate the implied compliment, but the real world is not so clear. Quality and Reliability Engineering (QRE) is a difficult statistical job and Intel does the best to set limits which will ensure a long and reliable life for it's products, but even so every CPU is different. While you could do a statistical analysis and determine a median value down to a microvolt, in the real world it's unclear where any individual part would fall in that broad statistical spectrum beyond a probability.

All that said, I look at people who claim to know the maximum safe voltage for a processor who have no large dataset of information or long term analysis and I personally think these "experts" have no idea what they are talking about.

* Not speaking for Intel Corporation *
Click to expand...
http://forums.anandtech.com/showpost.php?p=31120367&postcount=23
 
ordovician said:
I assume you're getting that info from a quick google search. The 1.38V number is the MAX voltage that the average Sandy Bridge CPU can take for a reasonable period of time without damaging itself. My understanding is that this does not necessarily mean your CPU can be pegged at that voltage for an extended period of time. It's akin to saying that a PSU's peak output is its rated wattage, which is a no-no.


http://forums.anandtech.com/showpost.php?p=31120367&postcount=23
Click to expand...

thanks for the link. interesting read.
 
ordovician said:
I assume you're getting that info from a quick google search. The 1.38V number is the MAX voltage that the average Sandy Bridge CPU can take for a reasonable period of time without damaging itself. My understanding is that this does not necessarily mean your CPU can be pegged at that voltage for an extended period of time. It's akin to saying that a PSU's peak output is its rated wattage, which is a no-no.


http://forums.anandtech.com/showpost.php?p=31120367&postcount=23
Click to expand...

That was a very interesting read. Even though the post(s) are a few months old.
Pretty much clarifies that Intel doesnt know the absolute max, and the VID range is not the "functional" range of the processor--only the functional range of the VRD 12 spec. And also that the CPU's VID it requests depends on its multiplier, and the CPU's are calibrated for high/lower vids depending on how they perform in testing.

The only valid conclusion is keep it SOMEWHERE below 1.52v on air, watch the temps, and that low VID chips may clock higher than high VID chips...

One thing that's unclear is if chips with lower default VIDs can handle voltages better over time (aka degradation) than higher VID chips...maybe the whole reason some chips have a higher VID is because they require more voltage to stabilize in Intel's tests, and thus don't run quite as hot, but may also have trouble clocking high (again because they might not respond as well to voltages)...
 
Porter_ said:
it certainly won't hurt. i'd be willing to bet that you won't notice much of a performance difference between say 4.6ghz and 5.0ghz.
Click to expand...

I'm down to 1.28v @ 4.6 I'll try 1.25v next time lol. The vdroop kicks in or well up and volts go to 1.3 during high load. Will see later on.
 
ordovician said:
I do IBT, Prime, and OCCT. Just not after every single iterative jump in volts/freq.



I think there is some misunderstanding about the point of this thread. I am not looking for help overclocking - and although I appreciate the advice, I know exactly what I am doing and how to do it. My quandry is whether it is really worth it or not.

I have concerns about pushing the Load Line Calibration, voltage, and other VRM settings too high because nobody (not even Intel, and they have admitted that) has compiled enough statistical data on what kind of voltage and amperage can be pumped through the SB procs 24/7. I'm not spreading or perpetuating FUD, but I am highly skeptical that these things should have anything over 1.325V running through them. Hell, anything over 1.3V on the 45nm i7's of yore could cause degradation and even failure if you were really unlucky.

The Asus rep here seems to suggest people can "fire and forget" with a 1.41V at 4.8 Ghz cookie-cutter OC. That's great if you're a board maker, but not so great if you're a CPU that has to last more than 4 months.

So, my question stands: Why run 4.8-5.0 at ~1.38-1.42 V when I can do 4.5 at 1.3 V?
Click to expand...

For gaming? No you can leave it on stock you'll get the same performance especially since the gpu are the important factor unless you play at 720p.

Btw Asus and Gigabyte boards have larger circuits than the other boards. So they're voltage regulation is a lot better and they follow intels spec. I know on a Gigabyte board with LLC enabled and disable there ain't much of a difference thanks to the number out filter caps they have
 
Last edited:
ordovician said:
I assume you're getting that info from a quick google search. The 1.38V number is the MAX voltage that the average Sandy Bridge CPU can take for a reasonable period of time without damaging itself. My understanding is that this does not necessarily mean your CPU can be pegged at that voltage for an extended period of time. It's akin to saying that a PSU's peak output is its rated wattage, which is a no-no.


http://forums.anandtech.com/showpost.php?p=31120367&postcount=23
Click to expand...

No, I'm getting that info from overclockers over at TPU.
And you just said the exact same thing I did.
1.38V is the MAX voltage it can take without damaging it self.
Does that mean it's the peak? No, you can easily hit 1.5V if you want to.
That's why I recommend 1.375.

So basically what you're saying, is that the highest SAFE voltage is 1.38V, but you're saying you can't go there because it's unsafe?
-.-
 
Last edited:
You must log in or register to reply here.
Back
Top