The Reality of Haswell Overclocking - Results Poll

[Tweakin]

As I recently upgraded to Haswell from i7-920 platform, I spent numerous hours reading OC results from all the major forums, and with anything in life; some of those forums show results from way too much voltage and individuals who want a big e-penis.

One gets a sense that Haswell can obtain 4.6Ghz easily and this is just not the case. It's Intel's lottery again, go big or go home! I wanted to create a quick poll with screen shots from CPU-z showing voltage and clock. I'm hoping to help destroy the myths and show what you can "realistically" obtain from these chips for those who are ready to venture into 4th Gen Intel while keeping input voltage within Intel specs.

Please add screen shots if you post with your 24/7 settings and thermals (air/water) to validate please.

IBT voltage with CPU-z and AIDA64 Thermals

 

[Cerb]

Reality not on your poll: I decided to not bother even getting parts capable of OCing, this time around, after reading how mild the OCs generally were, on average, with single arbitrary samples, without pretty high dollar and/or voltage investments, and how much work was going into getting even those OCs.

I have yet to get a sense that anyone is easily hitting more than about 4.2GHz...which, granted, is not bad, if you're already going for a $100+ mobo and $30+ CPU cooler. But, not compelling enough to put extra money into OC-capable parts, IMO, if looking at it from more of a value-based perspective. OTOH, though SB, and maybe IB, getting OC-capable parts, if just to stretch the life of the PC out, would have been a no-brainer.

 

[Termie]

What load application are you using to demonstrate voltage? If using adapative voltage, the results will depend on whether AVX is being utilized.

I don't have screenshots right now, but my 24/7 setting for all the game benchmarking I do on my 4770K is 4.4GHz with a +0.05 adaptive. I've logged dozens of hours in my testing. That takes me to 1.248v under Intel Burn Test, and about 1.2V in games.

Take note - this voltage is TOO HIGH for sustained IBT runs on my midrange air cooler. It gets into the mid-90s Celsius. But I don't benchmark IBT, I benchmark games, and the peak temps I've seen are about 67C. This is almost entirely due to the much lower voltage required for a gaming load, and hence my original question to the OP.

Also note - the limits on my previous 4670K were much lower, around 4.2GHz at 1.28V.

 

[Conroe]

I had my son's at 4.5 1.225v IBT stable but it hit 95c. I had to turn off all the power saving options in BIOS too. So I just set it to 4.0 and left most everything else on auto for his 24/7 use.

 

[OCGuy]

I'm going to stick with the theory that the majority of people who cannot get 4.2-4.5ghz on air are not spending enough time with their BIOS. Core voltage is not the only setting these days, I learned that with my chip.

There are some Haswell OC duds that hit the voltage wall at 4.0, but there are statistics out there that give a general idea of OC capability.

 

[PliotronX]

My 4670k is stable at 4.6GHz on a Hyper 212 at 1.195v. Dunno if it would have been the case before delidding though as I took the razor to it seconds after opening the package :whiste:

 

[nwo]

Only 4.1GHz on my 4770k :\

Haven't played around with the manual OC too much. Just used the auto OC. When I tried manually OCing, I got it up to 4.3GHz but every now and then I would get a restart. At 4.4GHz it would not boot up no matter what.

 

[OCGuy]

Only 4.1GHz on my 4770k :\

Haven't played around with the manual OC too much. Just used the auto OC. When I tried manually OCing, I got it up to 4.3GHz but every now and then I would get a restart. At 4.4GHz it would not boot up no matter what.

And this proves my point.

Auto OC is not a representation of how a chip can overclock. I had to clear CMOS when trying to auto to 4ghz.

 

[Tweakin]

What load application are you using to demonstrate voltage? If using adapative voltage, the results will depend on whether AVX is being utilized.

I have always used IBT for voltage and thermals as it really pounds the cpu. Voltage for this was static as Gigabyte boards don't really have an adaptive setting. I also firmly believe in Intel's 10% rule for additional voltage, hence my lower 24/7 clock.

 

[Tweakin]

I'm going to stick with the theory that the majority of people who cannot get 4.2-4.5ghz on air are not spending enough time with their BIOS. Core voltage is not the only setting these days, I learned that with my chip.

There are some Haswell OC duds that hit the voltage wall at 4.0, but there are statistics out there that give a general idea of OC capability.

Agreed, many potential tweaks with this chip, but the vast majority of them will only bring stability to chip, core voltage is still (IMO) the major issue. As you pointed out, many are using the Auto OC software and calling this an overclock, which it is...just not from an enthusiasts perspective.

 

[Khato]

I've been running 4.5 GHz with a Noctua NH-D14 cooler, though I have a feeling that no one will really care about my result since I realized that I don't care in the least about being able to run AVX instructions. The only software I have that uses AVX would be video transcoding and I prefer to do that on my underclocked/undervolted media computer instead as it's far more efficient. (For comparison, this processor refused to run AVX workloads above 4.2 GHz.)

Soon as you take AVX instructions out of the equation Haswell becomes quite reasonable for overclocking on both frequency and temperature. It's really kinda sad that so many enthusiasts declare their maximum OC based on AVX stress tests despite the fact that said stress test is the only application installed on the system that uses AVX instructions.

 

[Kougar]

I'm going to stick with the theory that the majority of people who cannot get 4.2-4.5ghz on air are not spending enough time with their BIOS. Core voltage is not the only setting these days, I learned that with my chip.

There are some Haswell OC duds that hit the voltage wall at 4.0, but there are statistics out there that give a general idea of OC capability.

My chip got up to 4.2Ghz stable but even that required a triple 140mm radiator to keep the core temps under 80c during 24/7 loads. The problem was, it didn't remain stable and it got to the point the CPU could not even run at 4Ghz at stock settings anymore.

I had to apply extra voltages just to keep it stable at 4.0Ghz which wasn't the case when I first began OCing the chip. I never went above 1.3v for any setting and kept it under 80c for most of time it was in my system. It wasn't worth the extra heat/power consumption for a paltry 4Ghz so I swapped it to a 4771 that has VTd capability. Unless I'm buying a high-end soldered chip I'm never going to pay extra for an unlocked consumer-grade chip again.

 

[Lepton87]

Unless I'm buying a high-end soldered chip I'm never going to pay extra for an unlocked consumer-grade chip again.

What if a chip like 2500k/2600k comes out again? Even 50% OC are possible with those chips, almost like good old times.

 

[Madpacket]

This is why overclocking with Haswell is kind of pointless.

 

[Tweakin]

...I don't care in the least about being able to run AVX instructions.

I don't either...

 

[Kougar]

What if a chip like 2500k/2600k comes out again? Even 50% OC are possible with those chips, almost like good old times.

Nope, those OC'd well but they still had features disabled on the unlocked parts. I'm also pretty sure Intel will continue to use thermal goop instead of solder on consumer-grade chips for the foreseeable future anyway.

'Good old times' for me was the "Core" era... Took my E8300 and gave it a 100% overclock from 1.86Ghz to 3.8Ghz, and it was good for 24/7 use without too much heat. At this point I'm going to hold onto this 4771 for a few years until I see some $500 eight-core chips with solder and intact feature sets come out. It's bad enough Intel jipped launch day Haswell buyers with its 8-series C1 stepping chipset bug.

 

[Tweakin]

This is why overclocking with Haswell is kind of pointless.

Point taken...

For me, as an enthusiast; I'll always take the extra "breathing" room offered by a mild overclock. Half the fun is trying to see how far you can go!

 

[OCGuy]

This is why overclocking with Haswell is kind of pointless.

???

 

[OCGuy]

My chip got up to 4.2Ghz stable but even that required a triple 140mm radiator to keep the core temps under 80c during 24/7 loads. The problem was, it didn't remain stable and it got to the point the CPU could not even run at 4Ghz at stock settings anymore.

I had to apply extra voltages just to keep it stable at 4.0Ghz which wasn't the case when I first began OCing the chip. I never went above 1.3v for any setting and kept it under 80c for most of time it was in my system. It wasn't worth the extra heat/power consumption for a paltry 4Ghz so I swapped it to a 4771 that has VTd capability. Unless I'm buying a high-end soldered chip I'm never going to pay extra for an unlocked consumer-grade chip again.

Strange...I can do 4.0 with the stock cooler and rarely see 80c.

Haswell is a solid step forward for Intel’s microarchitecture design. Its improvement over the Sandy-to-Ivy Bridge processor shrink is palpable and impressive.

It overclocks well on ambient cooling – especially our sample – but that overclocking level will depend heavily on your particular CPU. Thanks to speaking with the kind folks at ASUS (whose efforts helped make the on-release overclocking guide with excellent tips and guidance possible), we know that the average CPU overclock on Haswell might go down a little bit. There are dogs (4.3 GHz chips) and there are great chips (4.8 GHz+), all at the same voltage. On average, you can expect general overclocks in the 4.5-4.8 GHz range. The dogs are rare, but they do exist.

http://www.overclockers.com/intel-i7-4770k-haswell-cpu-review

Now, I’ll share a secret imparted by the folks at ASUS who gave several reviewers some tips on overclocking the retail stepping Haswell chips: Set Vcore to 1.20 V. Set all cores to 46x (which would be a 4.6 GHz overclock), save & reboot. If the system boots past the UEFI and either begins to load or, ideally, makes it into the OS and is stable, you have a 50th percentile or greater chip on the Haswell overclocking-ability bell curve. If it won’t at least boot there and make it into the UEFI, you probably have less than a 50th percentile chip. You can expect chips in the lower 50th percentile to top out in the 4.4-4.5 GHz range at 1.25 V.

If your chip will boot at 4.6 GHz and 1.25 V, that’s very good. It means you have at least an average chip. If it will boot at 4.6 GHz and is stable there, then you may have an above average chip. The best chips will be able to do 4.8 GHz stable at 1.25 V. Our sample did 4.8 GHz, but at 1.3 V and on a custom water loop. Using 1.3 V will likely put a chip out of the air cooling / AIO water cooling thermal envelope. Temperatures in all of these scenarios, from the dog 4.3 GHz chips up to the good 4.8 GHz chips, will always be in the ~90°C range. That’s just the nature of Haswell. With the VRM on-die, think of Haswell as Ivy Bridge plus 10° C.....

http://www.overclockers.com/3step-guide-to-overclock-intel-haswell - This is a good guide that helped me when I was stuck hanging @ 4.0 with more vcore than I use @ 4.2 now



If all you do is put everything on "Auto" and maybe manually set the Vcore, you aren't going to be happy. This isn't Penryn, I had to learn it the hard way as well.

 

[Morbus]

My Asus auto overclock suite will set my 5670 to 4.5GHZ on all 4 course at 1.26V. With a stock cooler. I imagine a decent cooler will get it up to 4.6 or 4.7.

I guess I have a pretty good chip, and I may buy an aftermarket cooler in a year or so, just to play around with it. It's a shame to leave it there at stock levels.

Strange...I can do 4.0 with the stock cooler and rarely see 80c.

Depends on what you're doing. I was running Assassin's Creed 4 on 4.2GHZ (edit, it was actually 4.2 for 1 and 2 cores, 4.1 for 3 cores and 4.1 for 4 cores), 1.2V and it never went above 75, and was usually around 65/70. With a stock cooler.

 

[Tweakin]

My chip got up to 4.2Ghz stable but even that required a triple 140mm radiator to keep the core temps under 80c during 24/7 loads. The problem was, it didn't remain stable and it got to the point the CPU could not even run at 4Ghz at stock settings anymore.

I had to apply extra voltages just to keep it stable at 4.0Ghz which wasn't the case when I first began OCing the chip. I never went above 1.3v for any setting and kept it under 80c for most of time it was in my system. It wasn't worth the extra heat/power consumption for a paltry 4Ghz so I swapped it to a 4771 that has VTd capability. Unless I'm buying a high-end soldered chip I'm never going to pay extra for an unlocked consumer-grade chip again.

I could do 3.9 on stock voltage with thermals in the high 50's on a dual 120 rad while running IBT...something seems off.

 

[nwo]

My Asus auto overclock suite will set my 5670 to 4.5GHZ on all 4 course at 1.26V. With a stock cooler. I imagine a decent cooler will get it up to 4.6 or 4.7.

I guess I have a pretty good chip, and I may buy an aftermarket cooler in a year or so, just to play around with it. It's a shame to leave it there at stock levels.

You definitely have a good chip there. My Asus z87-A auto OC only took my 4670k up to 4.3 stable. I bricked it when I manually tried 4.5... :$

 

[BonzaiDuck]

What if a chip like 2500k/2600k comes out again? Even 50% OC are possible with those chips, almost like good old times.

I am very pessimistic.

Ten years ago, visiting my cousin in WA, I met an Australian engineer who attended cuz's church. he worked for Intel, apparently involved in processor design. Couple years later, I visited again and had opportunity to see the same guy at the church. He'd quit; was trying to start a new career. He said:

"It's getting down to the mole-eck-ular level!" For some reason, he didn't want to do that work anymore. And I'm probably naïve and unknowledgeable about the business he was in. But this thread reminds me.

I think the limitations that are there are the result of the 22nm lithography. And I thought it curious that there wasn't a die-shrink between IB and Haswell.

I'm building an IB-E system this year -- the chipset is moving toward end of its lifecycle. I don't want to wait for Haswell E. IF it turns out better than I think, I'll still have a great computer, even if I'm disappointed for missing an opportunity by waiting. But I don't think it will be that great.

You can probably build a great computer with a Haswell, and maybe the enthusiast parts-buying population are driven by a lemming impulse. I don't mean that to be unflattering, but we're always sitting on the edge of our chairs, hoping that new is better.

I've got friends with six-year-old dual core systems who believe the technology has far exceeded their needs or perception of progress.

Intel is being driven by a different agenda based on mobile markets. At least that's what I perceive as these new i7 generations are released.

 

[Tristor]

I'm IBT stable at 4.6@1.37vCore, but I am considering dropping back down to 4.4, since I got stable at 1.265vCore. I hit a bit of a wall at 4.4, and it took a big jump to go from 4.4 to 4.5, and then not so much to get to 4.6. I never pushed it to 4.8, but I think I can probably hit 4.8 around 1.43vCore based on my current scaling. This is with a delidded i5-4670k on a lapped H110. Temps were fantastic the entire time, it's not thermally limited, just voltage limited, which is concerning to some degree. (I had a VID of 1V though, so my chip was binned well from the factory.)

 

[Jacky60]

I'm at 4.7 with a 4770k at 1.256 vcore, 1.199 ring voltage, no screen shot as I really can't be bothered trying to prove this to anyone other than myself. I followed this guide:
http://www.overclockersclub.com/guides/overclock_intel_4770k_guide/