Welcome to the Haswell Overclocking Thread. In here I will do my best to provide information regarding Haswell overclocking.
=================================Basic Haswell Info==================================
What is Haswell?
Haswell is the forth and latest-gen cpu line out for consumers. Of all the ones released the two of most interest are the 4670k and the 4770k (~$220, $330 respectively). These are the two unlocked versions of Intel's line of CPUs. This means we can adjust the core multiplier for overclocking later. It has IPC improvements over Ivy Bridge, meaning a 3.5ghz Haswell will typically beat a 3.5ghz Ivy Bridge. take that into consideration if you get a lower overclock. Its integrated graphics are of no concern for the PC enthusiast as you will end up buying a discrete graphics solution. Of course, the 4770k is an i7 with hyperthreading and the i5 does not have that. There is no shrink this time; the manufacturing process stays at 22nm. This is the tock in the tick tock model from Intel.
What is this USB 3 error I heard?
There were some issues with sleep mode and USB 3 slots on some motherboards at the very start. The problem was caught and as far as I can tell, no consumers have been complaining about USB 3 issues and the vendors assure us that the issue has been fixed already.
What socket is this chip?
This chip is a socket 1150 chip. You will not be able to use the same motherboard from any generation ago. You will be able to use the same cooler if it worked with the 1155, however. This includes the popular Coolermaster Hyper 212 Evo, Noctua D14, etc.
Why the motherboard change?
This is due to power changes in the CPU. Some parts relating to CPU power control have been moved from the motherboard to the CPU. This allows for more advanced power states for the CPU. Mean note that while CPU power draw as a whole did not decrease at peak by that much, that some elements of the motherboard have been moved to the CPU causing the overall power consumption of the system to decrease as a whole.
What about chipsets?
The chipset of concern is the new z87 chipset. It allows for more 6gb Sata connections... for a total of 6. Up to 6 USB 3.0 ports will be available with native support. Typically you will find PCI 3.0 slots. Please note that a x8 PCIE 3.0 is equivalent to a x16 PCIE 2.0 slot. Ram is typically supported up to 3000mhz DDR3. In addition to all of these benefits, the onboard audio has been updated to the new ALC 1150 standard. Although the specifications are superior, please note it's the total integration of the technology that dictates the end sound quality.
Any info on batches?
There are the Malaysian batches and the Costa Rica batches. I've listed all the results from other people down in the graph later in this thread. I noticed no major difference between batches however I have not heard of a seriously bad Costa Rica chip yet. It may be due to the fact those are still rare. It's hard to figure out how well a CPU will overclock if you don't overclock it, but one very dodgy way of doing it is to check stock VID.
What about delidding?
Unfortunately, Haswell comes with problems of its own that would be fixed by a proper delidding. Please note that delidding may make your CPU harder to sell again and it may be dangerous if it goes wrong. The glue used to hold everything together left some room under the IHS or integrated heat spreader. You can use the razor blade method but I recommend the vice method as it is much safer. Please watch videos on it and ask around for more info before attempting. If you do something stupid, you've just broken your CPU. It doesn't happen all that often, but do be careful. I recommend going over to the delidding page for specifics but the very quick rundown is to clamp the CPU onto a vice, and hit it with a wooden block so the top metal part of the CPU pops off. Remove the gunk, add your choice of applicant... Coollaboratory is well regarded for this task. Thin layer. You can expect a good 10C decrease in temps, maybe more. Go to http://www.overclock.net/t/1313179/official-delidded-club for a nice guide on delidding.
Is your heatsink/cooling solution up to par with what you wish to attempt? Expect hot temperatures ahead. Overclocking isn't horrible, but the heat will be biting. Anything under 80C is absolutely OK for 24/7 running even for the paranoid. 90C or higher is only acceptable during a stress test. Please avoid 100C, that is dangerous and the CPU will attempt to throttle. It's bad practice to have the CPU so hot, it throttles.
Stock -> Coolermaster Hyper 212 Evo -> Noctua D14/Thermalright Silver Arrow -> x60 Kraken -> Custom Loop
First, a little bit of a heads up on adaptive mode. You can take advantage of the new CPU power states. Whereas normally my computer would hammer my CPU with 1.385v of juice, it will now sip, say, 0.3v doing daily activities that I do but scale up when more power is required. This is very good for the longevity of your CPU as well as for power savings. To toggle it you first need to make sure your power settings for Windows are NOT set to high performance as you will not be able to take advantage of this feature. Second, you need to adjust the option in your motherboard BIOS. Enable C states and set the Core voltage mode to be adaptive instead of manual override. You can do the same for ring bus if you want. There are various modes or levels of C states. I recommend setting it to C7. Some people reported their power delivery being higher than estimated on idle because their power options in Windows is set to performance, not balanced. It does not affect me personally. To check if Cstates are working, close all major programs. Open HWMonitor or HWInfo and check the Vcore, not the CPUVid. The Vcore should jump around and increase as load increases.
NOTE: ONLY stress with adaptive OFF! The reason for this is because while adaptive is all fine and dandy, it can force the motherboard to give the CPU more volts than you set it to under very heavy loads. This primarily happens with synthetic benchmarks. Stressing with adaptive may overwhelm your cooling solution and cause sad things to happen. According to my tests, running adaptive with very heavy real life CPU loads MAY still cause the CPU to get more voltage than you set it to, but the difference is down to say, 0.03v. That isn't too bad but that is still significant, so be aware. Upon "stressing" via my chess program which loads all cores with 100% load, I hit 0.036v above my set limit. Gaming won't really use 100% all core at all times however. Gaming will be absolutely fine. Chess draws 100% from all cores 100% of the time without stop until you make it stop, so in a way it is like a semi-stress test. Encoding videos, for example, will not use 100% all cores or anywhere near that level. I know most people don't make their CPU run chess all night, but it's something to be aware of. On gaming I saw my CPU voltage go an insignificant 0.01 over max, and only for small periods of time.
So why do I stress with chess? For one, it does not elevate voltages by a ridiculous amount in adaptive. It's not a dealbreaker by any means for other stress tests but it's nice to know. It also simulates a maximum real life work load. Expect the heat to be a notch or two higher than your most intensive computer game, and the same with heat. It's a fine stability test, and can also be altered to become a benchmark to test various uncore/core settings. If you're interesting, try downloading Arena 3.0 GUI for free. The Stockfish 4 engine is free and IMO the hardest to pass. A guide can be found here:
Update your UEFI to its latest version! This is still a new launch and bugs and updates are expected! Update your stress test and your monitoring programs! Note: New stress tests might be more intensive.
First and foremost: Expect overclocking results to differ. We should all know by now that not every chip is created equal. With Haswell, this is more true than ever. Expect a very wide variance in end results. It's a silicon lottery, folks.
1. Set Uncore (AKA Ring Bus) to core ratio to manual. Set it to stock multiplier manually. With ring bus running on stock and locked at stock for now, no need to fiddle with ring bus voltage. That goes to auto. Some motherboards MAY increase Vring to unsafe levels if you didn't manually set ring bus to stock because the motherboard will try to auto-overclock the ring bus if left on auto. Avoid future headaches by following step 1.
2. Set any XMP profile OFF for ram. If your ram is above 1600, set it to 1600, no higher while we're testing overclocks. Heck, if your ram is XMP'ed for 1600, lower it to non XMP. Doesn't hurt.
3. Start ramping up the core multiplier and voltage until you think you've found your sweet spot. Do stress test and if you pass, go to step 4.
NOTE: Yes, you up the core multiplier by 1 each time. You can probably get away with starting at 4ghz though. Because the core clock is 100, 100 x 40 = 4000mhz or 4ghz.
4. Now we fiddle with ram, higher ring bus/ring bus voltage in effort to get marginally better performance. Stress test. Do not raise ring bus or ram if it means lowering core clock for stability.
5. If you are hardcore or have a lot of time on your hands, you can try adding clockstrap to the mix but I'd stay away from it unless you're super duper picky. Nobody has reported very good results with this but you're welcome to experiment.
6. Set it to adaptive mode instead of manual now. Do not stress with adaptive.
If you simply raise the multiplier on the core and change the voltage, you'll probably run into a bad overclock because the overclocked ring bus will hinder the core overclock. And they say it but it's true: Core is king. You'd generally rather have stock ring bus if that means getting 100mhz faster core clock. Same for ram of course... it's a tall order to hit DDR3 3000 with higher overclocks. I'm not 100% sure if you need to stop XMP profile for the ram speed if it's 1600. Personally I'd lower it to non-XMP. You can turn it back on when you've got you core clock set. If by some small chance the XMP is hindering the overclock however, you won't know if it's ram and probably won't think of ram as a possible cause. Box timings should be fine.
I've posted benchmarks under this section somewhere, which shows graphs proving ring bus settings to be of little consequence in benchmarks and applications. Keep in mind if you are adamant in overclocking the ring bus you also need to provide extra voltage to it. Your core clock should always be equal to or higher than your ring bus.
"1:1 Cache Ratio"
In a perfect world we'd all be running 1:1 cache ratio, but we'd also be running 500 ghz overclocks and sipping iced tea on clouds. We don't live in that world. If you got a cherry picked unit, fine, you can hit 1:1. For the rest of us, you cannot, pure and simple. Say the highest core overclock you can get is 4.6ghz. If you try to bring your uncore also to 4.6ghz, very likely you've either 1) Crashed your system because the uncore OC makes the core OC unstable 2) Crashed because you lack sufficient Vring 3) Applied unsafe Vring. You can't get past the first issue. You'd rather have 4.6ghz core and stock uncore (3.4ghz is stock core and uncore for 4670k, 3.5ghz for 4770k) than 4.5 core and 4.5 uncore. So what the heck is this 1:1 Cache Ratio nonsense?
It's the idea that your ring bus helps your performance up until it is the same speed as your processor. But you should know by now that ring bus helps performance... by a super small margin. The amount is negligible. It's basically saying, if your ring bus is higher than your core speed, that extra ring bus isn't doing anything supposidly. But the entire point is useless as pretty much nobody can hit 1:1 in the first place, let alone get above 1:1. Let me make this crystal clear: 1:1 doesn't make your CPU magically faster. You don't get an extra boost for doing 1:1. It's the same boost from 1:0.9 to 1:0.95 as 1:0.95 to 1:1. The amount of performance gain from uncore is roughly the same no matter how close your uncore is to your core. All that jabber about "keeping uncore 200-300mhz below core" is simply misleading. There is no such bottleneck that occurs if it's lower which those people seem to imply, and I have hardcore benchmark after benchmark to back up my statement. You overclock core with uncore set to stock so it doesn't lower your max core overclock. Then you overclock uncore without ever lowering core to get a higher uncore. If it happens to be 200-300 mhz under your core, awesome. If not or you don't want to push an unsafe Vring, that's fine too. Overclocking Haswell is complicated as is, last thing we need is to mislead and confuse people with 1:1 ratios.
You will pretty much always be fine at 1.2v for core voltage provided you're not stressing on synthetics. Synthetics are things like Linpack, Prime95, Aida64. They are just that, synthetic tests, as are not actual real-world loads. Non-synthetic stress tests would be like chess or encoding a video with CPU only.
If you do want to stress I recommend Noctua D14 as a starting cooler. What happens if you have a 212 Evo from Coolermaster and you want better cooling? You now are stuck with a lower-end cooler. I get reports of 80C on Prime when using 1.2v to 1.25v using D14.
By the time you hit 1.3v you need to already know what you want to stress with. Linpack will implode your CPU if you're on air. To achieve good temperatures with Prime and air at this point I recommend a delid. By 1.35v stressing with Prime is very dicey and high end closed loop is required.
At any rate, 1.4v is not recommended unless you want to go my route with my configuration and you really want to push it and you need the voltage. Otherwise only high end closed loop or a custom loop should play at this setting. Haswell temperatures are very reliant on voltages, not frequency. Note that while I said, for example, that 1.35v requires a high end closed loop solution or better for stressing, that does not mean that's required for gaming/any other non-synthetic application. Only you can decide what sort of stability is acceptable to you.
Now why exactly did I list my recommendations about voltages this way? What matters in the end are two things when it comes to voltage safety: A) You do not hit above 95C under whatever you wish to stress and B) You do not exceed 1.45-1.5v no matter what. The thing is, most people run into the first problem long before they hit the second, because by 1.35v if you want to run Linpack, you're already getting dangerous temperatures on air. If you are using a custom loop with a delid though, the second problem might hit you first. Personally I am running 1.425v at 4.6ghz on D14 and the only reason why I can do so is because I'm not stressing with synthetics. This of course has drawbacks, in that I can't ever use synthetics to stress at this speed. However if my settings are stable then I can squeeze in that extra 100mhz because the max real-world loads will not anywhere near Prime95. Please note this is only for people who are willing to deal with Bsods and know what they are doing. Only you can decide what stability is stable enough for you.
Quick note on auto-overclocking: It will not be as efficient as manual. Do it manually. I wrote a guide. Use it. If you have MSI motherboard, OCGene will block manual overclocking. You need to click on the OCGenie button in the BIOS to stop that from happening. Lower end MSI boards may be voltage locked above a set amount, say 1.3v! Beware!
About Ring Bus aka Uncore aka Cache Ratio Tweaking:
The naming used differs between motherboard manufacturers. Keep in mind that Uncore is the same as Ring Bus, and is sometimes known as 'cache ratio'. Some boards have 'minimum' and 'maximum' cache ratio. Just set them to the same. Obviously, 'cache voltage' is 'ring bus voltage' or 'uncore voltage'. For ring bus voltage, 1.2 is considered pretty safe. But read the voltages others used in my chart first! Ring bus takes less voltage, don't just replicate it as if it were core clock. You are going to need to raise ring bus voltage if you plan on overclocking the ring bus significantly. If your ring bus is manually set to the default value, meaning it's not overclocked for sure, leave it at auto is typically fine. Try not to exceed 1.35v. I try to keep it at 1.3v or under personally. If you do not set ring bus to stock multiplier manually, some motherboards will try to overclock it on its own, which might not only crash your system, it could also damage your CPU because the dumb motherboard is setting an unsafe voltage!
One last thing: In a test where I went from stock to 4.1ghz uncore @ 1.27 Vring and I upped SA/Io voltage by +0.135v, the temperature went up 4-5C in 5 minutes compared to without those voltage bumps. Expect a small temperature increase with overclocking Uncore.
About Other Settings:
PLL settings generally don't do too much but if you want to fiddle around with it, go for it. Just be careful about the voltage ranges. For LLC or Load Line Calibration, this is to help combat Vdroop but I don't find it particularly useful because Haswell isn't really vulnerable to Vdroop. The CPU Input Voltage on some Asus motherboards is actually the VRIN aka VCCIN. The VRIN can be thought of as the entire amount of voltage drawn by the CPU and all of its components. try not to go over 2.0v for this. My opinion is this: The higher your core voltage, the higher the chance you need to change this value. I don't feel you need to play around with this setting at 1.2v core voltage. Maybe at 1.35v? Maybe higher.
You are of course, free to experiment with these settings to see if it nets some stability for you. In most cases it won't, but hey, you might be the exception to the rule.
When your Vcore is really going up, at least 1.35 probably 1.4v or above, you may need to change other settings. For one, keep your Vccin or total CPU voltage to 0.5v above Vcore. You can try 1.9 or 2.0v. 2.2 is dangerous. Vccin is also known as Vrin. In Asus ROG boards, try tweaking the "eventual input voltage" instead.
Also, you may need to alter the voltages for SA, IO Analog, IO digital as well. Try adding 0.1v to them. For me, it fixed an odd 9C bsod when the vcore became too high, and adding more voltage made the CPU more unstable. Please note: It is unclear at this point what voltages are dangerous. Be careful with these voltages. JJ from Asus said these voltages help stabilize a higher Dram divider but I got 9c errors at as low as 1600 DDR3.
CPU VID vs Vcore
There is a difference between CPU VID and CPU Vcore when I mention both of them together. I repeat: Only when I am talking about VID and Vcore in the same sentence does my definition of Vcore change.
Normally when I say Vcore I mean what you think I mean. But when I mention VID vs Vcore, VID is the amount of core voltage you set in the BIOS yourself. You should know it, you're the one that set the voltage in there. The Vcore is the number measured by Hwinfo or HWmonitor on your CPU when it is under max load.
What does this mean? Your Vcore could be above your VID. If you set 1.3v in the BIOS that's 1.3v VID. If you are also under adaptive voltage and you're running Prime95, your Vcore could be a whopping 1.5v, way above your set 1.3v.
Finally please note, there are multiple reports of people having a higher Vcore than VID even under non-synthetic loads but the extra voltage is relatively small. Just be careful and monitor voltages closely. As your VID increases the extra voltage drawn in from a regular non-synthetic load increases.
Credits to Maxforces for the second part of the benchmarks. From my personal benchmarks, I found the drop of 0.7ghz for the ring bus to be an equal performance hit of 0.05ghz decrease in core clock and this difference shows in a very CPU reliant benchmark like chess.
NOTE: Do not stress on adaptive!
You should have your method of stressing picked out. Once again, we all have different opinions on what we should stress things with and for how long. With Haswell, the AVX2 instruction set is introduced. This is very rarely used and if you're sitting here going HUH?, then you won't use it in real life usages. The problem is, when you stress via Linpack your temperatures will be scorching hot because the instruction set just nukes the temperatures. So here you have to decide how stable is stable enough. Some purists feel they must pass all tests at all times. Version 27.9 of Prime95 is the normal Prime version, the 28.1 version is scorching hot. If you insist on passing all tests at all times, you need to be realistic about your goals as your overclocks will be severely hampered. In my opinion if you're going to pass the latest Prime you might as well have gone for IBT or Linpack.
The two other popular options are Aida64 and Prime95. For Prime95 I recommend using blend for 6 hours. If your average Aida64 run passes after 12-24 hours and Prime95 is fine after 6, I would call that rock solid stable. Please keep in mind the temperatures on Haswell really do shoot up during a stress test. You will not hit those temperatures under normal operation. If you hit 95C you might be fine outside of stress testing. It's very unlikely your games will seriously stress your CPU across multiple cores. Aida64 is considered to be less stressful compared to most other tests, so running it for longer might help. Despite JJ from Asus saying that Prime95 is not 'certified' for Haswell, all of my data shows it's a fine stresser for Haswell. Although Asus has spent a long time looking at Haswell overclocking, I feel videos of JJ from Asus are a nice starter but not the end-all source of information.
A new program, Intel's XTU, is now a popular test for Haswell stability, since Aida64 is not a free product and may only be downloaded as a trial.
My personal standards are a bit out there but it works for me. See, my most CPU intensive application in real world usage scenario is 24/7 chess calculation. That uses 100% all cores at all times nonstop. If I can run that without ever getting a single crash, there is no way I will crash in games. So far from what I can tell, I'm correct. But some will burn me at the stake for doing so. But hey, it let me shoehorn in an extra 200-300mhz and it's my computer! I guess my point is, just find what works for you. I've survived 1.38v on air which is generally unfathomable, because I do not do synthetic stress test anymore. If you decide to ditch synthetic stress test you will be able to wiggle out MUCH more voltage but there are obviously drawbacks.
Try using coretemp or Realtemp or HWMonitor to monitor your temperatures. HWInfo is the best if you want all the info in one place and you want tons of info. To measure and check your adaptive settings you can only use HWMonitor or HWInfo. For me, the VCore is under my motherboard which may be confusing. It should be changing a lot. You need to use the latest version of the software! First check to see you have the latest version!
If you look at the FFT, it changes from large to small. Small FFT is good for producing heat while large FFT is better for finding instability in your RAM. If you fail on large, but pass on small then it probably means your RAM overclock is unstable. If you fail on small, it means your CPU is unstable.
If you fail almost instantly (within 30 minutes), then your vcore is too low. Try raising vcore in increments of 0.005V until you can pass at least 2 hours of blend, although it's up to you how long you want to be stable. I find that 2 hours of blend stability is equal to pretty much stable unless you are doing very intensive tasks such as folding, autocad, etc.
This game is known for being easily unstable under an overclock. Some people consider this a better test of stability than even Linpack!
Haswell is still a new launch. That means programs associated with it will constantly get updates. Older versions of Prime are easier to pass than newer ones. Many programs have glitches. There are idiosyncrasies. The only way to iron out what's what is with your cooperation, and lots of communication.
===================Google Doc Containing Settings of Various People's Overclocks============