My i7 2700k overclocking results

S

sram

[H]ard|Gawd
Joined
Jul 30, 2007
Messages
1,699
After quick adjustments in the UEFI BIOS, I was able to reach 4.8 GHz @ 1.32 intelBurnTest stable. My temps are high though according to coretemp. I reached 72 C on one of the cores at full load.

I was expecting more of my corsair H100. Idle temps are:
33
35
35
36

which isn't bad I think. The BIOS however shows 28 degrees, don't know which one is more accurate, but I will go with coretemp of course(speedfan, realtemp, and everest all show same temps). I want to reach ~ 5.2 GHz without going much more than 1.35 Vcore, but it sounds like that is ganna be tough.

Here is a shot:



And this is my build:

http://hardforum.com/showthread.php?t=1680769
 
Good luck with your target overclock. Based on what I've seen, less than 1% of SB chips can hit that frequency at that voltage. Most require more than 1.4v to just reach 5 ghz.

Those temps are in line with what is expected from the H100.

Only 60 seconds of testing? You need to be testing for far longer than that to determine stability. With 1gb runs, you should be doing at least 25-30 of them. Since you have 32gb of RAM, you should set RAM usage to max, and do at least 5 runs.
 
You're still within perfectly acceptable numbers....I'm tweaking my 2600k oc and it's taking me 1.3Vcc to pull 42 stable; currently testing 43@same and so far so good (6hrs of Prime in so far *knock on wood*).

I'm pulling average 60c off a TRUE in push/pull but then again I'm running Yate highs (D12SH-12) which are surely quite a bit louder than an H100. :D

I'll see where things go....I was hoping to go no higher than 1.35Vcc but in retrospect and knowing what I do now (especially considering the methods of Speedstep and not being at load 100% of normal use and therefore being undervolted most of the time)....I may go all the way to 1.4Vcc; still deciding on that one.
 
Those temps are well within normal range, although I'd say your idle is a little bit higher than I typically see on Hxx's - not that idle matters much, but could easily just be indicative of a slightly higher ambient climate out where you're out. Regardless, 72C on your highest core at 4.8 is really fine. And not trying to rain on your parade, but to say that 5.2 @ 1.35v is a "lofty goal" would be putting it lightly - you'd have to have a very cherry chip to make those numbers. To be fair, some chips won't hit 5.2 regardless, or at least without going way past safe voltage. That being said, don't despair, as you definitely have some headroom over 1.35v that's still very safe.

And a big +1 to

Tsumi said:
Only 60 seconds of testing? You need to be testing for far longer than that to determine stability. With 1gb runs, you should be doing at least 25-30 of them. Since you have 32gb of RAM, you should set RAM usage to max, and do at least 5 runs.
Click to expand...

Not only general stability, but to actually see where your temp goes. Don't forget the premise of how your cooling works, especially on a liquid loop like the H100 - its going to take a while for that loop to come closer to equillibrium, and 60s is NOT giving you a fair readout of what your max temp is.
 
Tsumi said:
Good luck with your target overclock. Based on what I've seen, less than 1% of SB chips can hit that frequency at that voltage. Most require more than 1.4v to just reach 5 ghz.

Those temps are in line with what is expected from the H100.

Only 60 seconds of testing? You need to be testing for far longer than that to determine stability. With 1gb runs, you should be doing at least 25-30 of them. Since you have 32gb of RAM, you should set RAM usage to max, and do at least 5 runs.
Click to expand...

If I reach 1.4v, so be it. I know many people are running their cpus @ higher voltages. Hopefully it is not going to be that bad because I think temps will go up. I'll do another intelburntest, but you know intelburntest is kinda of meant to quickly tell you if your system isn't stable. I know that even if I don't pass IBT 60s test, i could do prime95 for hours without without failure or at least that's how it was with my core2 quad.

zero2dash said:
You're still within perfectly acceptable numbers....I'm tweaking my 2600k oc and it's taking me 1.3Vcc to pull 42 stable; currently testing 43@same and so far so good (6hrs of Prime in so far *knock on wood*).

I'm pulling average 60c off a TRUE in push/pull but then again I'm running Yate highs (D12SH-12) which are surely quite a bit louder than an H100. :D

I'll see where things go....I was hoping to go no higher than 1.35Vcc but in retrospect and knowing what I do now (especially considering the methods of Speedstep and not being at load 100% of normal use and therefore being undervolted most of the time)....I may go all the way to 1.4Vcc; still deciding on that one.
Click to expand...

It is the same reason I might do 1.4 Vcore. After all, those chips are supposed to be all about overclocking and being stressed.

Tiporaro said:
Those temps are well within normal range, although I'd say your idle is a little bit higher than I typically see on Hxx's - not that idle matters much, but could easily just be indicative of a slightly higher ambient climate out where you're out. Regardless, 72C on your highest core at 4.8 is really fine. And not trying to rain on your parade, but to say that 5.2 @ 1.35v is a "lofty goal" would be putting it lightly - you'd have to have a very cherry chip to make those numbers. To be fair, some chips won't hit 5.2 regardless, or at least without going way past safe voltage. That being said, don't despair, as you definitely have some headroom over 1.35v that's still very safe.

And a big +1 to



Not only general stability, but to actually see where your temp goes. Don't forget the premise of how your cooling works, especially on a liquid loop like the H100 - its going to take a while for that loop to come closer to equillibrium, and 60s is NOT giving you a fair readout of what your max temp is.
Click to expand...

It is good to hear that I'm within range of temps. Regarding Vcore, wish me good luck. At least for only one time, I want to feel that I picked a very special CPU.

And I will run IBT for more time and see what happens.
 
On my 2600k build it was taking around 1.42v to get 4.8Ghz, and that seemed to be the brick wall no matter how much voltage I threw at it. Temps were similar to what you're reporting (mine's also on a H100). So, I think your oc is quite good :cool:
 
My 2700k does 4.8 at 1.35v, it takes 1.40+ to hit 5GHz. Your voltage / temps are fine and pretty standard. You would have to have a beyond exceptional chip to hit 5.2GHz on an H100 at 1.35v.
 
sram said:
If I reach 1.4v, so be it. I know many people are running their cpus @ higher voltages. Hopefully it is not going to be that bad because I think temps will go up. I'll do another intelburntest, but you know intelburntest is kinda of meant to quickly tell you if your system isn't stable. I know that even if I don't pass IBT 60s test, i could do prime95 for hours without without failure or at least that's how it was with my core2 quad.



It is the same reason I might do 1.4 Vcore. After all, those chips are supposed to be all about overclocking and being stressed.



It is good to hear that I'm within range of temps. Regarding Vcore, wish me good luck. At least for only one time, I want to feel that I picked a very special CPU.

And I will run IBT for more time and see what happens.
Click to expand...

K1tty said:
On my 2600k build it was taking around 1.42v to get 4.8Ghz, and that seemed to be the brick wall no matter how much voltage I threw at it. Temps were similar to what you're reporting (mine's also on a H100). So, I think your oc is quite good :cool:
Click to expand...

Do you guys have Hyperthreading enabled or disabled? I disabled it hoping that it will help lower temps but it didn't really help or so it seems.
 
sram said:
Do you guys have Hyperthreading enabled or disabled? I disabled it hoping that it will help lower temps but it didn't really help or so it seems.
Click to expand...

Enabled on mine; not going to disable HT when I paid for a 2600K (rather than a 2500K) just to get HT. :)
(that's my philosophy at least)
I'd rather be a few hundred MHz less on an OC with HT rather than be running slightly higher with no HT because the HT is definitely going to be of a greater benefit than a few hundred MHz.
 
zero2dash said:
Enabled on mine; not going to disable HT when I paid for a 2600K (rather than a 2500K) just to get HT. :)
(that's my philosophy at least)
I'd rather be a few hundred MHz less on an OC with HT rather than be running slightly higher with no HT because the HT is definitely going to be of a greater benefit than a few hundred MHz.
Click to expand...

Pretty much this - if you're going to disable HT, no reason to not have gone with a 2500k (or 2550k).
 
Enabled on mine as well. I encode a lot of stuff and have been trying folding so the HT is needed :)
 
BF3 has a strange choppiness/stutter with HT enabled, its the only time I ever turn it off
 
I reached 5GHz now @ 1.41 Vcore but the temps are insane. It reached 82 on one of the cores. It goes back to 75 when the AC is one however. Summer is starting where I'm and room temps nowadays fluctuate between 30 and 35.

How can I improve my cpu cooling? i really expected more of the H100.


Thanks.
 
thats odd, I'm running 5.0 @ +0.230 offset which is ~1.52v under prime95 load, and my temps are only around 68'C

whats your room temp?

edit: and also, how do you have the H100 mounted? I noticed that in my RV02, if I had the CPU block mounted a certain way, it wouldn't seat flush with the CPU and I'd get horrible temps even though all 4 screws were tightened down normally
 
Last edited:
Reality said:
BF3 has a strange choppiness/stutter with HT enabled, its the only time I ever turn it off
Click to expand...

Interesting I seem to get smoother play with HT turned on (Note, this is with a i3 2100 ).
I will say that I used to have a similar issue with my OC'ed 920 and even though it was stable @4.4 in all the test for some games I needed a slightly higher voltage to get smooth playback with HT turned on. BF3 wasn't out back then so I have no other comments that that. :)

@ O.P.

Are you temps really up in the 70s with the H100 or is that a improper reading from Core Temp?
 
Reality said:
thats odd, I'm running 5.0 @ +0.230 offset which is ~1.52v under prime95 load, and my temps are only around 68'C

whats your room temp?

edit: and also, how do you have the H100 mounted? I noticed that in my RV02, if I had the CPU block mounted a certain way, it wouldn't seat flush with the CPU and I'd get horrible temps even though all 4 screws were tightened down normally
Click to expand...

I said that my room without the AC on is pretty warm. It ranges from 30 to 35. Now, it is your turn to tell me: What is your room temps? Since you are running the same frequency @ relatively much higher Vcore than me, and your temps don't even reach 70, your ambient should be really cold. That or there is something wrong with my H100 installation.

Interesting note about H100 mounting. You can look at my photos in the other thread and see how I mounted it. Tell me if it is similar to what you have.

ythe1300 said:
Interesting I seem to get smoother play with HT turned on (Note, this is with a i3 2100 ).
I will say that I used to have a similar issue with my OC'ed 920 and even though it was stable @4.4 in all the test for some games I needed a slightly higher voltage to get smooth playback with HT turned on. BF3 wasn't out back then so I have no other comments that that. :)

@ O.P.

Are you temps really up in the 70s with the H100 or is that a improper reading from Core Temp?
Click to expand...

Well, coretemp, speedfan and everest all report the same temps. It is only the BIOS that reports a temp that is less by ~5 degrees. Coretemp should be right because TJ.max is 98 which is right for the i7 2700k.



This shot will help:

 
Forgot to say that I'm prime95 stable, but not IBT stable. IBT is just too stressful.
 
If you're not IBT stable, then you're not stable, period.

30-35 C ambients is really warm. In general, most people have 20-23 C ambients, especially air conditioned. I would say those temps with those voltages at that overclock with your ambient temps is what is to be expected of the H100. After all, the H100 still gets beat out by the top end air coolers like the good old NH-D14.

Pretty much the only way I can see you improving your temps with the H100 is getting faster fans or getting fans in push/pull.
 
My room temp is around 21'C

The NHD-14 only beats the H100 at idle according to OC3D tests (tested on an [email protected])

At "High" fan speed on the unit, it beats the NHD14 by ~1'C and is equal to the silverarrow. Put the fans on 12v, and it will pull ahead of both.
 
Last edited:
Tsumi said:
If you're not IBT stable, then you're not stable, period.

30-35 C ambients is really warm. In general, most people have 20-23 C ambients, especially air conditioned. I would say those temps with those voltages at that overclock with your ambient temps is what is to be expected of the H100. After all, the H100 still gets beat out by the top end air coolers like the good old NH-D14.

Pretty much the only way I can see you improving your temps with the H100 is getting faster fans or getting fans in push/pull.
Click to expand...

Well, I'm trying to convince myself that your above statement in bold is false. Before IBT, everybody used prime95 and considered themselves stable. I tried it myself, if you fail IBT but pass prime5, you will have a really hard time crashing/freezing your system doing normal tasks. One time, I kept using my other PC doing regular daily work while priming at the same time for days and it just kept going.

You can call that my way of convincing myself that I'm stable even If I don't pass IBT:D

I know I know, passing IBT can make you rest assured that stability issues aren't existent in your system. There is almost no better way to tell. It is one of the most stressful programs out there, and if you pass this strict test, then you don't have to worry about a thing.

I'll see if I can pass it with higher Vcore.

To take care of temps, I think I'll put some ice packs close to my intake fans.:p
 
Actually, I've had times where my computer would fail P95, and pass IBT. They stress different parts of your system in general. While P95 is good for testing the core stability, IBT will highly stress the IMC and your RAM as well. Instabilities in IBT might be the result of RAM instability, not enough QPI volts, or not enough CPU volts. That is why I run both tests and make sure it's stable on both of them in order to call my system stable. I also run the different P95 tests, usually the small FFT and max memory ones.
 
:) I went through all this not long ago and in the end decided to stick with 4.5Ghz 24/7 overclock. The amount of voltage I could get away with at 4.5Ghz was more attractive than the extra 300Mhz.
 
Reality said:
thats odd, I'm running 5.0 @ +0.230 offset which is ~1.52v under prime95 load, and my temps are only around 68'C

y
Click to expand...

Is it just me? Not saying its not possible, but 5.0 @ 1.52v ---> 68C full load sounds...well, way too good on a H100. The H100 is a nice cooler, but it doesn't work magic. A chip that is truly at 1.52v under full load not tipping 70C is quite possibly the best H100 I've heard of. Is this after a real amount of burn-in, or a 60s load?

sram said:
I reached 5GHz now @ 1.41 Vcore but the temps are insane. It reached 82 on one of the cores. It goes back to 75 when the AC is one however. Summer is starting where I'm and room temps nowadays fluctuate between 30 and 35.

How can I improve my cpu cooling? i really expected more of the H100.


Thanks.
Click to expand...

As is said plenty of times in plenty of reviews, the H100 is a good cooler, but its not anywhere near a true custom loop (which costs quite a bit more). A big portion of your limit here could very well be your ambient though - if your inside temp around your case is truly 30-35C (~90+F - I don't how you stand that in your house, high 70s is borderline intolerable for me inside), then you're dealing with a deficit to overcome to start with.

Tsumi said:
If you're not IBT stable, then you're not stable, period.

30-35 C ambients is really warm. In general, most people have 20-23 C ambients, especially air conditioned. I would say those temps with those voltages at that overclock with your ambient temps is what is to be expected of the H100. After all, the H100 still gets beat out by the top end air coolers like the good old NH-D14.

Pretty much the only way I can see you improving your temps with the H100 is getting faster fans or getting fans in push/pull.
Click to expand...

+1 to everything stated here.
 
I'll post a video when I get off work since you're questioning my cred.
 
One thing I love about using offset is that the chip will only max out on the vcore when it needs absolutely all of that juice. Was playing FF14 last night, had msi afterburner + rivatuner for the extra cpu stats. At 5000mhz, voltage was around 1.28 to 1.4, nowhere near 1.52. Same in Forged Alliance (which surprised me, as its an RTS)
 
sram said:
Well, I'm trying to convince myself that your above statement in bold is false. Before IBT, everybody used prime95 and considered themselves stable. I tried it myself, if you fail IBT but pass prime5, you will have a really hard time crashing/freezing your system doing normal tasks. One time, I kept using my other PC doing regular daily work while priming at the same time for days and it just kept going.

You can call that my way of convincing myself that I'm stable even If I don't pass IBT:D

I know I know, passing IBT can make you rest assured that stability issues aren't existent in your system. There is almost no better way to tell. It is one of the most stressful programs out there, and if you pass this strict test, then you don't have to worry about a thing.

I'll see if I can pass it with higher Vcore.

To take care of temps, I think I'll put some ice packs close to my intake fans.:p
Click to expand...

The whole point is that there should be literally no software that will crash your system because it was "overtaxing" your cpu/bus. If something crashes because of "too much use," it simply is not stable, whether that software ends up being prime95, ibt, or the windows experience index =)

edit: yes, there are exceptions like futuremark for gpu, but that's a different story imo.
 
What's offset called in my maximus iv mobo? I want to try it.
 
Some nice clocks so far. To get a true measure of how well your CPU is you need to enable HT. I never understood why someone with a 2600/2700 would not.
 
Uploading video now. For reference, when I tested earlier today, ambient temperature was 20.3'C and Cores were 65/68/69/69

Ambient now is fluctuating from 21.6-24.9. Temps are in line with the current ambient (66/72/70/72), with highs of 76 and low's of 70 (on Core's 2, 3, and 4)

edit: could've made the video longer I guess, but I lost interest

http://www.youtube.com/watch?v=b7Asm1A4g0c
 
Last edited:
Tiporaro said:
You choose either manual overclocking or offset under the same option in Asus bioses.
Click to expand...

I'm at work now so I can't play with it. But, can you please expand? It is the extreme tweaker page? I don't remember seeing an option with the work offset anywhere.
 
Reality said:
Uploading video now. For reference, when I tested earlier today, ambient temperature was 20.3'C and Cores were 65/68/69/69

Ambient now is fluctuating from 21.6-24.9. Temps are in line with the current ambient (66/72/70/72), with highs of 76 and low's of 70 (on Core's 2, 3, and 4)

edit: could've made the video longer I guess, but I lost interest

http://www.youtube.com/watch?v=b7Asm1A4g0c
Click to expand...

Ok, so not trying to say anything argumentative, but your temps highs were reading 76 in a very short period. Given the general function of a cooling loop, that would likely continue to rise somewhat over the course of a longer burn-in, as is usually seen with any simple rad setup. So not that your temps or rig aren't fine, but saying that your CPU is 68 under prime95 load isn't quite accurate - when people talk about temps under prime95, they're referring to the actual thermal load it will acclimate to, and the highest core is usually the representative as you're seeking the "max temp" that was pegged during burn in. The temperature build is more gradual in a liquid loop than a straight hsf for obvious reasons, and as far as any review I've seen the h100 still climbs up over a period of time until it reaches equilibrium.

In any case, it's a very nice rig you have, and the only thing that caught me off guard was more the claim that a full prime95 1.52v burn in didn't even top 70C at all, as that's exceptional for an Hx series.

Edit: Couldn't tell 100% from the clip, but that looked like a rv02 case, was it? One of my favorite cases to work with.
 
Last edited:
sram said:
I'm at work now so I can't play with it. But, can you please expand? It is the extreme tweaker page? I don't remember seeing an option with the work offset anywhere.
Click to expand...

I just double checked the Asus rive manual, but I believe it's the same place for any of this Asus line. Under the first tab (extreme tweaker) there's the option for CPU voltage. You should be able to change that to offset, and then have the options for adjusting offset available. I believe there's also a thread somewhere in the motherboard forums that was posted by Asus about overclocking this series of motherboards if you're curious about where some of the additional tweaks are.
 
K1tty said:
:) I went through all this not long ago and in the end decided to stick with 4.5Ghz 24/7 overclock. The amount of voltage I could get away with at 4.5Ghz was more attractive than the extra 300Mhz.
Click to expand...

I'm pretty much at this point myself....I'm Priming 47 @ 1.39 Vcc and it's still going (hasn't popped 'offline' in Teamviewer yet) so on one hand I'm like "yay 4.7" but OTOH I'm like "well it only took 1.3Vcc to get 4.4 P95 8hr stable and I think 4.5 took 1.32Vcc and 4.6 took 1.35Vcc....where do you draw the line? The e-peen says "go all the way" :D but the logical side says "run lower, enjoy the lower volts and temps". I think I'll probably see how much I can get at 1.4Vcc or less but probably back down a few hundred MHz and just run 4.6, 4.5 or 4.4 in the end.

The biggest question/issue I have now (on mine) is temps. I'm generally hovering around the 68-72 mark (with a TRUE and 2 D12SH-12's) but my max temps reported by CoreTemp on 1 core reached 80. The fluctuating ones aren't hitting that now, but the idea of a max going up to 80 doesn't really sit well with me...though then again, this is while stressing, which is going to push those temps hotter than any core loading in CS5 or dBpowerAMP will do. I'm debating on re-seating and reapplying compound. I really was hoping to have lower temps than this considering I'm running these fans...good Lord, 88CFM @ 40dB each (+ I'm push/pull so I'm running 2 of them). :) I would expect some lower temps in an 'ideal situation' basis. :D

ShuttleLuv said:
P95 stresses Sandy more than IBT, this is a known fact.
Click to expand...

My recent checking on things has pretty much said this.....I know I used LinX religiously for Nehalem but it seems like the general consensus is the P95 is a much better stresser at least for SB, so I've been running P95.
 
Last edited:
It's better to run both tests. If you're stable on both tests, then your chances of a BSOD with any program is slim to none.
 
zero2dash said:
I'm pretty much at this point myself....I'm Priming 47 @ 1.39 Vcc and it's still going (hasn't popped 'offline' in Teamviewer yet) so on one hand I'm like "yay 4.7" but OTOH I'm like "well it only took 1.3Vcc to get 4.4 P95 8hr stable and I think 4.5 took 1.32Vcc and 4.6 took 1.35Vcc....where do you draw the line? The e-peen says "go all the way" :D but the logical side says "run lower, enjoy the lower volts and temps". I think I'll probably see how much I can get at 1.4Vcc or less but probably back down a few hundred MHz and just run 4.6, 4.5 or 4.4 in the end.

The biggest question/issue I have now (on mine) is temps. I'm generally hovering around the 68-72 mark (with a TRUE and 2 D12SH-12's) but my max temps reported by CoreTemp on 1 core reached 80. The fluctuating ones aren't hitting that now, but the idea of a max going up to 80 doesn't really sit well with me...though then again, this is while stressing, which is going to push those temps hotter than any core loading in CS5 or dBpowerAMP will do. I'm debating on re-seating and reapplying compound. I really was hoping to have lower temps than this considering I'm running these fans...good Lord, 88CFM @ 40dB each (+ I'm push/pull so I'm running 2 of them). :) I would expect some lower temps in an 'ideal situation' basis. :D



My recent checking on things has pretty much said this.....I know I used LinX religiously for Nehalem but it seems like the general consensus is the P95 is a much better stresser at least for SB, so I've been running P95.
Click to expand...

I seriously don't see p95 stressing my 2700k more than IBT. I passed prime95. And I also passed IBT high test after increasing Vcore, but I still can't pass the maximum test.

Do you think it is the memory preventing me from passing IBT? How does it work actually? When you set the test to maximum, you use all your memory. Does that mean also more stress on your cpu? Or is it just testing all available memory vs testing part of it?

Thanks.
 
Which P95 tests are you running?

If you're stable on small FFTs, and stable at high on LinX, then there's a good chance that your RAM is the culprit. Or, it could simply be because it takes more time for the errors to show up.

Use a program called HCI Memtest to stress your RAM and test for stability.
 
You must log in or register to reply here.
Back
Top