realistic Expectations with a 2500k and Z68 mobo

[Duvie]

I have been out of the game a bit....

Still milling the Kentsfield Q6700@3.2ghz on a 780sli mobo (overkill in hindsight since i dont game)....

However now I have been thinking about getting a 2500K sandy brdige cpu on a Z68 mobo.

What I am wondering is...

1) If my intent is just to raise the multiplier what is the max people seem to be getting with the current run of chips at a default vcore?
2) What is the limit of vcore most feel is safe?
3) Do most boards automatically raise vcore as you OC it, or do you need a high end board to control those aspects?

I would be happy with 4-4.2ghz as close to stock vcore as possible to run as cool as possible....With he IPC dis-advantages from my long in the tooth b0 stepping Kentsfield I imagine a 4ghz Sandy bridge would fill like a 4.8-5ghz kentsfield. Is this right?

I would like to get better raw clock speed as some apps I run really only take advantage of one core most of the time. When I do final renderings then it can take advantage of 85-90% of my current setup.

 

[Hulk]

I'm no expert but have the rig you are considering so I'll give you my opinion. I'm running 4.2GHz at 1.30Volts full load. I can go to 4.4 or 4.5GHz but I need more like 1.40 Volts, which I'm not comfortable with. My personal max for this CPU is 1.3 Volts. Most of these 2500k's will do 4-4.2GHz at stock volts.
My temps will maybe hit 60C so it's running pretty cool.

 

[Duvie]

I am not familiar with your cooler.....How would that compare to a TRUE120...copper core....with dual 120mmm tri-cools?

60c is fine by me....

 

[cantholdanymore]

Read my signature. Not an expert, but I think safe would be about 1.35

 

[Silverforce11]

I run 4.3 without vcore tweaks.

SB is pretty beastly when it comes to OC.

 

[fastamdman]

1) If my intent is just to raise the multiplier what is the max people seem to be getting with the current run of chips at a default vcore?
2) What is the limit of vcore most feel is safe?
3) Do most boards automatically raise vcore as you OC it, or do you need a high end board to control those aspects?

1. Overclocking sandybridge is done by simply changing the multiplier, the way you are going to do it is the way we all do it. EVERY chip is different, however with that said, EVERY chip can do 4.5ghz 100% stable 24/7.

2. I personally like to keep vcore under 1.45 for 24/7 use, some say 1.4, either way it doesn't really matter. As long as temps are fine you could be running 1.6 vcore. No one is going to have the chips long enough to matter. We are talking about degradation changing the chip after 5-10 years instead of 15-20. These chips are to new to worry about and with how fast technology changes if you plan on upgrading in a year to IB, then you can basically run whatever you want

3. Almost every board has an "auto" option which will run to much vcore for any given multiplier. For example you only need roughly 1.25 vcore *also known as stock* for a 4.0ghz overclock, aka 40x multiplier. However on auto, you could be seeing 1.3 to 1.35 vcore which is enough for a lot of people to see 45x.

I am a HUGE advocate of 2500k's and overclocking. I feel as if anyone who has one should be running AT LEAST a hyper 212+ cpu cooler and overclocking to 45x. I would venture to say ALL 2500k's can do it under 1.45, and 95% can do it at 1.4 and under which is 100% safe.

Thing you have to remember is that every chip is different. Some chips can run a 57x multiplier, some can only run a 52x. Some need 1.4vcore for 45x, while some need 1.37 for 50x. EVERY chip is going to be different, even from the same batches. Nothing is the same.

Silverforce11 - "without vcore tweaks" are you implying that you run 1.25vcore or are you running "auto" vcore aka, way to much for your OC?

 

[Diogenes2]

Where are you getting these 52x and 57x multipliers ?

 

[Corporate Thug]

when tweaking the vcore, why do people insist on using offset voltages? Why not set the voltage to fixed to the highest you are comfortable with, with medium LLC, and see what the best number you end up with?

 

[Diogenes2]

That's fine for achieving stability, but once established, it doesn't make sense to leave vcore at 1.35 or higher, even when idling.

With offset, you get a very low .9 or less when idle..

 

[Silverforce11]

Silverforce11 - "without vcore tweaks" are you implying that you run 1.25vcore or are you running "auto" vcore aka, way to much for your OC?

Yup, leave it on auto. Core Temp detects 1.3vcore when loaded with prime95.

This board doesn't have manual vcore settings.

 

[Corporate Thug]

That's fine for achieving stability, but once established, it doesn't make sense to leave vcore at 1.35 or higher, even when idling.

With offset, you get a very low .9 or less when idle..

honest question: realistically, what difference does it make if its idling at .9v or 1.35v?

 

[ashishmishra]

honest question: realistically, what difference does it make if its idling at .9v or 1.35v?

On my machine it makes my processor run 10c cooler and pull about 30-40W less at idle. It idles at about 1.2Ghz @ 0.856V and loads at 4.6Ghz @ 1.375V. Low power consumption when I don't load it and lots of power when I need it.

 

[Kenmitch]

What I am wondering is... 1) If my intent is just to raise the multiplier what is the max people seem to be getting with the current run of chips at a default vcore? 2) What is the limit of vcore most feel is safe? 3) Do most boards automatically raise vcore as you OC it, or do you need a high end board to control those aspects?

Most people will say to stay below 1.35v's some will say 1.40v's with good cooling is fine. Of course you don't wanna run your chip at 1.40v's at 90*C under load so your voltage and overclock will be thermally limited by your cooler.

Pretty much just put rig together load bios defaults then up the multiplier to overclock the 2500k's. Leaving the vcore on auto. The only downside is on some MB's the board will give more vcore than is needed to be stable at your overclock. Best to do some research on the bios options of the MB your interested in before purchasing. I'm not sure what others call it but the Asus bios has an offset mode with a +/- option for vcore which helps keep some controll on the vcore. Another thing to note is on some MB's without this feature you'll need to run with the xtra vcore or just hardlock the vcore in bios to what is needed for stability. Depending on the MB this may also mean that your sitting at the desktop at full overclock and full vcore....Not sure which ones do this but it a problem with some MSI boards and may effecto others also..

honest question: realistically, what difference does it make if its idling at .9v or 1.35v?

On some MB's doing this will also kill the 2500k's ability to throttle down the cores. You'll be sitting at the desktop full throttle full voltage all the time....Not sure which boards do this but I know some MSI's have this issue. http://forums.anandtech.com/showthread.php?t=2205832

 

[Hulk]

I am not familiar with your cooler.....How would that compare to a TRUE120...copper core....with dual 120mmm tri-cools?

60c is fine by me....

I don't know how my cooler compares to that one. I bought it because it has good performance and is virtually silent with the Noctua fan on the lowest setting. As I wrote above I might hit 60C and it's silent. Very happy. Of course I'm not pushing a huge O/C but I'm good with 4.2.

 

[RavenSEAL]

Personal opinion Op, ride it out until Ivy Bridge

 

[OILFIELDTRASH]

Sounds about right. I thought I had a dud of a chip but it looks normal from the comments in here. Getting 4.3 at 1.29v which seems to be the norm. I'm running a 2600k.

 

[fastamdman]

Where are you getting these 52x and 57x multipliers ?

I was talking about finding the max multiplier that the chip can do. 2500k's and 2600k's for that matter both have multiplier walls where you simply can not get into windows at X mutli. For example my 2500k will run 5200mhz, 52x multi but 53 will not get into windows, hence the wall, no matter what vcore. My 2600k has a 56x multi.

honest question: realistically, what difference does it make if its idling at .9v or 1.35v?

There is a BIG power consumption difference there.

Sounds about right. I thought I had a dud of a chip but it looks normal from the comments in here. Getting 4.3 at 1.29v which seems to be the norm. I'm running a 2600k.

We can't say whether or not your chip is a dud until we know what it can do. All chips can do 4.3 lol, push it a bit and see where it takes you. If you want to get there quick and you have aftermarket cooling, simply set your vcore to 1.55 and boot up with a 47x multiplier, if that passes do 48, then 49, then 50, on up. Wherever you stop will be the highest multiplier that chip can run.

As far as your 4.3 at 1.29volts, thats barely any voltage. Until you are running 1.4 or 1.45volts 24/7 its not a big deal. If your chip is stable at 4.3 with that low of a voltage, you can probably run 4.5 with around 1.35 vcore and might even have a 5ghz chip at a higher 1.45 vcore. You don't know until you push it. As long as temps are within reason, there is nothing wrong with running 1.4 or 1.45 vcore 24/7.

My 2500k runs 24/7 at 1.4vcore for a 4.5ghz overclock. Not the lowest vcore setting by any means, but its what she wants for stability lol. Everyone has a different opinion of stable too. P95 is absolutely nothing, you can't say you are stable tell you run a real program like LinX with the avx pack. Luckily though its still a 5200+mhz chip

 

[zlejedi]

Mine goes to x42 using stock cooler and x45 with +0,005V offset (well actually x46 too but I dropped it a bit to be sure of stability) with Mugen 2.