Intel Core i5 & i7 Sandy Bridge Overclocking and Feedback

[cadaveca]

4.2 GHz should be no problem at all. The G1.Sniper2 I reviewed HERE will set 4.2 GHz with the push of a button on ANY 2600K CPU.

The button that does that is the "Quick Boost" button found on the 5.25-inch bay device pictured here:

On the G1.Sniper2, that button works in real-time, allowing you to run "stock", and getting an immediate overclock up to 4.2 GHz whenever you like. I tested it quite a bit in different apps and such, and I never ran into any issues.

 

[Crap Daddy]

That's cool. Almost can't belive it. On the other hand 4.2Ghz is what I have on an ASRock P67 Extreme 4 (not much of an overclocker myself but I like reading what you guys do) with a 2500K just by setting the multi at 42. Everything else is on auto and default. 4.2 is the max turbo frequency, idle at 1.6. Works like a charm and I found out that it's the sweet spot for performance (gaming really don't need more) with minimum hassle.

 

[Nordic]

I'm back with a better overclock. Msi released a newer bios for my motherboard so I decided to try to get a better overclock. I did.
Last time I could not overclock with turbo boost. Now I can overclock to 4.7 with turbo boost. At least on one core. Multipliers per core run 44 45 46 47 starting from 4 cores. If I try to make any of those higher windows freezes up in boot.

 

[johnnyfiive]

My evga p67 ftw will do 4.3 with everything default, just change it to 43 and boot. Pretty insane how easy these sb chips are to oc. I recall spending a week getting 4.3 stable with my i7 920, worked damn hard for that oc. Now 4.3 takes 30 seconds, literally.

I'm currently at 4.9 on my 2500k, its humming along. Commmme onnnn ivy!

 

[Chicken Patty]

Joining a little late, but better than never



BTW, are these temps okay? I heard these CPU's can go up to about 80ºc safely, is this true?

 

[cadaveca]

yes, it's true. mine spend most of their time @ 85c loaded. Just don't use PLL overvolt, and keep everything else relatively decent.

Good temps, too, BTW. I wish my chips ran that cool!

 

[Chicken Patty]

yes, it's true. mine spend most of their time @ 85c loaded. Just don't use PLL overvolt, and keep everything else relatively decent.

Good temps, too, BTW. I wish my chips ran that cool!

All voltages except vcore all default, too easy to overclock these things. These temps are on air too, not too bad.

 

[technicks]

Brought the cpu voltage back to 1.29v. But my temps seem high? I haven't touched the other settings, just set it to 4.6 in the bios and lowered the voltage.
Can i reduce the cpu PLL voltage? Would that help to decrease the temperature?

 

[cadaveca]

Yes, it can decrease temperature slightly.

Your temps are fine, however. If they aren't breaking 75c after 8 hours of prime, you are 100% good to go, with room to spare. Been almost a year now I've been running my chip are far higher temps, and I have ZERO issues to report so far.

Default CPU PLL is 1.8v, and you've got 1.83v according to the AsRock tool. You may try going down to 1.75 or so, but that may not work. Generally it's accepted to lower CPU PLL voltage to 1.65v or so when running PLL Overvoltage, but if PLL Overvoltage is not enabled, I cannot gurantee that lowering CPU PLL will serve any benefit.

 

[technicks]

Next week i'm gonna redo/clean the waterloop to take out the nb blok. Will buy some decent thermal paste. I used the paste i had laying around from an old air cooler. Maybe that will help a bit to.

 

[cadaveca]

I found my CPUs weren't much cooler under water either. Was confused by it, even. But then I started to investigate power consumption, and found that when these chips push out like maybe 125w @ 4.5 GHz, there's very little room to improve temps, no matter the cooling solution. I suppose it's just too hard to remove so little heat from the IHS.

I mean after all, I ahve my AMD chips to compare with here. I saw SUBSTANTIAL cooling gains with water on my 1100T, could clock further and all that, but these SB chips...usually can max out on air, anyway.

 

[technicks]

At least now i know it's ok to run these temps. Would be nice if they could drop a bit but i don't fancy cranking up all the fans.

 

[cadaveca]

Yeah, temps are fine. Do let us know how it works out after you redo the loop, as I am quite curious how it'll work out for ya.

I'm currently using a Corsair H70, with the low-speed fan adapters on, even, because I see little reason for more. Temps will drop a little bit for me with max fan speed, and my Noctua NH-C14cooler does better too, but I prefer to use the Noctua in my review rig as the mount is so easy to use. The mounts on these Corsair coolers don't withstand multiple mountings very well.

 

[technicks]

Think i'm going to sell both Swiftech blocks and get socket 1155 block so i can mount it correct.
I use the 775 mounting holes and the block is mounted a bit crooked. Not that it will improve the cooling but it looks weird.

 

[Chicken Patty]

So far, I think water cooling will be pointless on my setup as well. These things run super cool.

 

[Darkleoco]

Couldn't pull 4.5 Ghz with even 1.29vcore does this mean I have a crappy chip for OC'ing?

 

[cadaveca]

Nope. My 49( no PLL OV) multi chip needs 1.32v for 4.5GHz

My 46 multi chip needs less voltage.

TOtally random situation..it's not about voltage it seems..it's mroe about overall power consumption, and votlage won't tell you that. Some lower-volt chips consume more current and still maintian the same overal lwattage thanks to the lower VID.


Max multi that boots into windows with PLL OV turned off tells you how good your chip is. Just diabled all but one core(2 cores by windows becuase of HT), set votlage to 1.525v, and do the multi test, one by one, to see where it doesn't make it into windows.

 

[LAN_deRf_HA]

Got my AP15 and R4 on my H70 down to around 1000 rpm. I cranked it up for stress testing but I probably didn't need to. These chips are great on just about anything but a stock cooler. Though on stock if you leave the thermal protection on they'll throttle down if they get too hot. In theory you could just toss one up to 4.2 GHz for a casual gamer and it would keep itself safe.

Speaking of, what's a good voltage for 4 GHz? 1.2v?

 

[cadaveca]

Yeah, 1.2v or so. might need a touch more, depends. Might even only need 1.175v. That's good for 4.0 for me.

 

[Darkleoco]

Nope. My 49( no PLL OV) multi chip needs 1.32v for 4.5GHz

My 46 multi chip needs less voltage.

TOtally random situation..it's not about voltage it seems..it's mroe about overall power consumption, and votlage won't tell you that. Some lower-volt chips consume more current and still maintian the same overal lwattage thanks to the lower VID.


Max multi that boots into windows with PLL OV turned off tells you how good your chip is. Just diabled all but one core(2 cores by windows becuase of HT), set votlage to 1.525v, and do the multi test, one by one, to see where it doesn't make it into windows.

Good to know that mine isn't bad at least, been running at 4.2 Ghz but need to clock my memory back up to 1600, hopefully I can get 4.5 Ghz without much trouble. Is 1.525v safe doing that test? I'm sure you know what your talking about I'm just overly nervous since this is my first time overclocking.

 

[Chicken Patty]

Nope. My 49( no PLL OV) multi chip needs 1.32v for 4.5GHz

My 46 multi chip needs less voltage.

TOtally random situation..it's not about voltage it seems..it's mroe about overall power consumption, and votlage won't tell you that. Some lower-volt chips consume more current and still maintian the same overal lwattage thanks to the lower VID.


Max multi that boots into windows with PLL OV turned off tells you how good your chip is. Just diabled all but one core(2 cores by windows becuase of HT), set votlage to 1.525v, and do the multi test, one by one, to see where it doesn't make it into windows.

Mine is at 1.368v for 4.5 GHz, but I do crunch so it's at a constant 100% for it's whole lifetime pretty much I might try to see if I can go lower, just had a BSOD, but I had a whole bunch of voltages and stuff on AUTO. I just dialed them in so we'll see if that was the cause of the BSOD.

 

[cadaveca]

Good to know that mine isn't bad at least, been running at 4.2 Ghz but need to clock my memory back up to 1600, hopefully I can get 4.5 Ghz without much trouble. Is 1.525v safe doing that test? I'm sure you know what your talking about I'm just overly nervous since this is my first time overclocking.

Yes, it is OK for quick testing for multi. It is NOT OK to stress test in any way. Just try to get into windows(3 cores disabled, of course), and if it gets to desktop, then shut off. Rinse and repeat, etc, etc.

You always take your chances when overclocking. If you're not confortable with it, don't do it. No big deal. Up to you to balance the risk/reward.

Mine is at 1.368v for 4.5 GHz, but I do crunch so it's at a constant 100% for it's whole lifetime pretty much

Mine has run Linpack for the majority of it's operating hours. I don't leave the PC on at night, and obviously I shut Linpack off when gaming. I like to tweak memory timings, and stability is of the utmost importance, so I'm always testing different things. Quite different usage scenarios, you and I, but I do try to account for all usage scenarios in my reviews, too. That's why my OC testing is done with relatively modest clocks...clocks that nearly any similar chip should be capable of.

 

[Chicken Patty]

Yes, it is OK for quick testing for multi. It is NOT OK to stress test in any way. Just try to get into windows(3 cores disabled, of course), and if it gets to desktop, then shut off. Rinse and repeat, etc, etc.

You always take your chances when overclocking. If you're not confortable with it, don't do it. No big deal. Up to you to balance the risk/reward.



Mine has run Linpack for the majority of it's operating hours. I don't leave the PC on at night, and obviously I shut Linpack off when gaming. I like to tweak memory timings, and stability is of the utmost importance, so I'm always testing different things. Quite different usage scenarios, you and I, but I do try to account for all usage scenarios in my reviews, too. That's why my OC testing is done with relatively modest clocks...clocks that nearly any similar chip should be capable of.

That's a very good approach you take and indeed you do at great job at that. I haven't tried tweaking memory yet because the sticks I have now are not the greatest. The one in my system specs should be here tomorrow, these should clock nicely and are already CL 8 so it's already an improvement there. This is what I'm running now, no need to mess with them in the meantime, like I said. New RAM should be in tomorrow.

 

[cadaveca]

Anything over 1600 MHz gives only modest gains, really. You'll go from 18k to 22k to 24k - 1600 to 1866 to 2133. Assuming reasonable timings. The biggest thing is the latency drop, from about 52ns down to 36ns or so with the good sticks.

LinPack load is pretty serious, so if I haven't killed a chip yet, I think that's pretty good. Broken power switches twice, and that gave me a scare,twice.

 

[Chicken Patty]

Anything over 1600 MHz gives only modest gains, really. You'll go from 18k to 22k to 24k - 1600 to 1866 to 2133. Assuming reasonable timings. The biggest thing is the latency drop, from about 52ns down to 36ns or so with the good sticks.

LinPack load is pretty serious, so if I haven't killed a chip yet, I think that's pretty good. Broken power switches twice, and that gave me a scare,twice.

Geez! These sticks I got are 1600, but at least they can run tighter timings and 1T with only 1.5v, these don't do too well, I need at least 1.65v to run 9-9-9-24 1T, and I know that's not the best option for the SB's. Not sure why the CPU'z screenshot says it can do that at 1.5v, because it doesn't, trust me on that.