Bought an OCUK '4.8Ghz' chip, can barely get to 4.6Ghz.

not really.
the guaranteed 4.8GHz CPU gives the end user piece of mind that it has hit those speeds before they buy. we do state the hardware that the overclock is achieved on for the customers own reference. if they choose to use different hardware then that's obviously their own decision.
the main thing is that 4.8GHz can be a difficult clock to achieve on low volts so it is a much better idea having an advantageous starting point. i.e. a cherry picked CPU.

i get that, but if the purchaser (like in the instance of the OP) doesnt know what settings you used, then its not really beneficial. and even if they purchased the exact same models as you, does that guarantee those products will perform exactly the same as they did in your workshop?
i dont know enough about OC'ing, so im not trying to say you are wrong, im merely asking.
 
no worries, happy to answer your questions :)

the reason we don't put the settings on the description is because they do vary from CPU to CPU. whilst it is a premium to pay for something you still have to put in the legwork for, we get countless requests from people asking for speedbinned CPUs.

I must get around three emails a day asking me how a certain CPU is overclocking from various forum members/customers who just want a CPU that clocks well or above average.
there was previously no way to offer an above average CPU to a customer but I think this is the best way.

a lot of enthusiasts want a better chip and are willing to pay premiums for that. the premium we charge is merely for the time of the technician who speedbins the cpu

if you can think of any room for improvement or suggestions then they would be more than welcome :)
 
not really.
the guaranteed 4.8GHz CPU gives the end user piece of mind that it has hit those speeds before they buy. we do state the hardware that the overclock is achieved on for the customers own reference. if they choose to use different hardware then that's obviously their own decision. the chip has been proven to work on the hardware stated. it is made quite clear that overclocking results have to be obtained by manual overclocking.
the main thing is that 4.8GHz can be a difficult clock to achieve on low volts so it is a much better idea having an advantageous starting point. i.e. a cherry picked CPU.

I agree
Its all about paying that bit more money for the peace of mind that you will be getting a decent CPU out of it instead of playing the 'Sandy Bridge Lottery'

I'd rather know that I'd be getting a 4.8Ghz chip that runs with safe voltage than the risk of ending up with a bad one that needs 1.4v+ just to get 4.6Ghz stable
 
How will ocuk know if there is a cherry chip can do 4.8GHz so easy ? By checking out batch number (the best batch number get a very low vcore!) if the customer willing to pay extra for those premium chip!

Two years ago rjk picked the cherry chip for my i7 920 bloomfield which it rather crazy did 4.00GHz at very low vcore at 1.200v and it stable for a whole 2 years! I guess it just my lucky day in 2009 when I brought the ocuk bundle.
 
Its been proven that batch numbers mean next to nothing with these new Cpu's
Two identical batch numbers can be a world apart when it comes to voltages & max frequency/multiplier
 
I agree
Its all about paying that bit more money for the peace of mind that you will be getting a decent CPU out of it instead of playing the 'Sandy Bridge Lottery'

I'd rather know that I'd be getting a 4.8Ghz chip that runs with safe voltage than the risk of ending up with a bad one that needs 1.4v+ just to get 4.6Ghz stable

agreed if i had had the spare money (and i could have found them on ocuks shop) then i would probably of paid the extra. mine seems to be stuck at 4.6GHz with 1.35v or below, havent tried to up the vcore for 4.8GHz as temps dont allow it. with all the contradicting max 'safe' figures i have stuck with 4.4GHz so i know im below the 1.35v that was originally quoted.
but i would also be very miffed if i already had the rest of the hardware (like i did as i was running a G620 before) and paid for a 4.8GHz chip that just didnt like my setup.
 
How will ocuk know if there is a cherry chip can do 4.8GHz so easy ? By checking out batch number (the best batch number get a very low vcore!) if the customer willing to pay extra for those premium chip!

no, all of the speedbinned CPUs are tested at that speed in the hardware stated for 8 hours+ before they can ship

Its been proven that batch numbers mean next to nothing with these new Cpu's
Two identical batch numbers can be a world apart when it comes to voltages & max frequency/multiplier

correct. batch number means nothing for SB chips. with bloomfield, it was always a fair indication of projected performance but 32nm cpus tend to have no batch trend.

but i would also be very miffed if i already had the rest of the hardware (like i did as i was running a G620 before) and paid for a 4.8GHz chip that just didnt like my setup.
agreed which is why we do what we can to help. I do a lot of work via our support forums to assist with issues.
However, the customer should only purchase a product like this when they are confident that they have sufficient knowledge to achieve the overclock themselves.
 
agreed which is why we do what we can to help. I do a lot of work via our support forums to assist with issues.
However, the customer should only purchase a product like this when they are confident that they have sufficient knowledge to achieve the overclock themselves.

yeh, i know that whenever i have had an issue with anything the ocuk staff have been more than helpful. the fact that the chips are only tested on one motherboard does put me off though. i know its all about 'time is money' so i completly understand, but you must have several test builds there, so maybe testing the chip on the most popular board of each manufacturer would then allow people to have confidence that the chip 'should' be guaranteed to work on pretty much most things. but like i said, 8hours on @least 4-5 motherboards would take an age so the prices would bump up to compensate this.
 
Sound like there is no point to buy many i7 2600K chips to find the cherry one! Unless it cost alots of money and waste all money too !

I leave my i7 2600K @ 4.4GHz for now as it stable at 1.31V (under load) 1.33V at the bios. And max temp doesn't go over 62C. That's the only safer vcore and temp I can get up to. But 4.6GHz seem a massive jump up to 1.38V at the bios!

Is it normal for sandybridge set itself at tjmax @ 98 while my bloomfield is 100 ?
 
Last edited:
I finally found the most stable now. No more cold boot issues (it now disappear) because I had changed the settings correctly in the bios.

Before 4.60GHz @ 1.38Vcore Bios (1.35Vcore in windows) LLC Level 5 (cold boot issues and BSD) Memory 1333MHz 8GB (4 slots of 2GB) @ 1.50V
Now 4.60GHz @ AutoVcore Bios (1.32Vcore in windows) LLC Level 6 (No more cold boot issues) Memory 1600MHz 4GB (2 slots of 2GB) @ 1.50V

So, it big difference and I think LLC Level 6 is the best option while leave the Vcore set at Auto but under load it only do 1.320V but at idle it went down to 0.984V (4.4GHz same setting but under load it only did 1.28Vcore and idle at 0.984V - very impressed!) going to try to push 4.8GHz and 5.0GHz now.
 
Got BSD again !!!! Totally fed up with it. I might go for OCUK bundle instead to see if it was stable for 4.60GHz, if not, I will ask rjk to test it for 24 hours plus to get it stable or fully refunded if it fail stable!
 
As soon as I worked out that IBT, Lynx and Prime Small FFTS were next to useless for measuring stability all of my BSOD's went away. @bulldog, I honestly doubt it's the Gigabyte board or the CPU if you are indeed 8 hours Prime Blend stable. Have you tried running mem test?

Hey dude, try Linx with AVX support.. it'll run your cpu red and finally is on par with Prime blend for stability testing. Theres a modified IBT as well with AVX. Finally got 5c hotter than blend and small FFT, though still only topping out at 82C in 30C ambient with my Noctua. (2600k 4.8GHz 1.385v).
 
Hey dude, try Linx with AVX support.. it'll run your cpu red and finally is on par with Prime blend for stability testing. Theres a modified IBT as well with AVX. Finally got 5c hotter than blend and small FFT, though still only topping out at 82C in 30C ambient with my Noctua. (2600k 4.8GHz 1.385v).

What board you got Justintime. I was very scary tonight as my pc didnt boot up at all, not loading to the bios at all. PC kept going on and off all the times. I point my screwdriver at the 2 pins by the F Panel, to clear CMOS but it won't work. So, I took the battery off, for 30 mins. Put the battery back in, and try clear CMOS again, failed to boot, I lost my temper and almost punch the door. I went downstair to have a cold drink and watch tv for an hour to calm me down. Then I went back to my pc and try once more, switch on, bingo it work !
That's the worse board I ever had in my experience, my other board EX58-UD5 are far the best and had cold boot issues in the past and got fixed with bios updates but this board Z68X UD4 B3 is really annoyed me. I don't know what to do with cold boot issues and I was so angry with my shop owner refuse to replaced the board under my 3 years warranty. I was suspicious that gigabyte had a new board again, it Z68XP UD4 to replaced Z68X UD4 - make me wondering why gigabyte had so many model numbers!

I just give up overclocking now. The memtest is no difference as I did try three different memory rams, 1333MHz, 1600MHz and 2000MHz and all was fine passed memtest.

Could be my power supply doesn't like my board at all ? Because when the pc was going on and off all times, I heard annoying noise come from power supply - really loud then go quiet! Never happen to my other board EX58 UD5. Still cold boot issues on my 3.40GHz stock speed. I knew from day 1 - it not fit for purpose! The board are deffo faulty.
 
Last edited:
I was getting the odd BSOD here and there at 1.33v, so I upped my voltage to 1.35v, leaving LLC at 6 still. Plus I upped my memory to 1.55v. Not had any BSOD errors since.

You might want to try those settings to see if it helps. It might save you from trying to punch any more doors in the future. ;)
 
Using a cheap and cheerful Asrock P67 PRO 3 Rev 3 (B3), quite stable. Your Psu seems more than adequate. Have you got another graphics card to try? Is your BIOS the latest? I've used that board with no problems, run it barebones (RAM, CPU, GPU and one hdd) and see if its doing the same.
 
1) my bios is latest from gigabyte F7
2) I will try 1.55 on memory ram even thought my memory spec is 9-9-9-24 1.50V
3) I had no spare graphic card sadly. Wish the board had onboard graphic!
4) My board is now tested at the nearby area shop and see if it any faulty and will let me know later today.
 
I was getting the odd BSOD here and there at 1.33v, so I upped my voltage to 1.35v, leaving LLC at 6 still. Plus I upped my memory to 1.55v. Not had any BSOD errors since.

You might want to try those settings to see if it helps. It might save you from trying to punch any more doors in the future. ;)

what is your memory timing please ? Is it 1600MHz at 9-9-9-24-2T @ 1.55V ?
 
Cheers. I just had a phone call from the shop and it now confirmed that the board is faulty and will give me a new board replacement today. :)

Updated: They had rang me and say my new board did 4.6GHz overclocked with vcore of 1.31V and still ongoing testing all day and overnight as I won't have it until end of the day tomorrow :)
 
Last edited:
Updated today: My pc is now back home with me as the new motherboard had passed 4.6GHz for 28 hours from the shop and I checked the bios just now, all setting are very different to the one I set before. It was 1.34V LLC 4 and Memory set at 1.54V and CPU PLL set at 1.84V for 4.60GHz. I just try test blend test (prime) for 30 minutes for same setting 4.60GHz and the vcore drop to between 1.31V and 1.33V and idle at 1.34V, before I set it all on Auto and 1.35 LLC 4 and it crashed with BSD straight away. So, I must be set the wrong setting at the bios.
 
Last edited:
Back
Top Bottom