Inside Intel's Secret Overclocking Lab: The Tools and Team Pushing CPUs to New Limits

(Image credit: Tom's Hardware)

All of Intel's first processors were overclockable, but they weren't specifically designed to offer much frequency headroom, or with features that improved overclockability. Instead, overclocking was truly opportunistic: Any overclocking headroom you received was down to the luck of the draw and your ability to work around the internal limits of the chip. Even with the limited tunable parameters available, enterprising enthusiasts found ways to push chips to higher speeds, thus wringing out more performance from cheaper models.

That eventually created a problem for the company, though. Intel tells us that back in 1999, counterfeit Pentium 3 processors began popping up in the grey market in Asia. The counterfeiters simply overclocked the cheaper models and then sanded off the product identifiers/frequency ratings, replacing them with the overclocked frequency and model numbers that matched the more expensive models. To put a stop to the practice, Intel locked all processors to the rated frequencies in an attempt to prevent counterfeiting.

Due to feedback from its customers (and media), Intel brought back overclockability in ~2003 and unveiled its new Extreme Edition processors. These chips allowed overclocking in exchange for a pricing premium, just like we see with Intel's K-Series processors today, but the remainder of Intel's processors remained locked.

And thus, Intel created a multi-billion dollar business centered on chip overclockability. However, because these early unlocked chips still weren't designed specifically for overclocking, there were shortfalls. For instance, Sandy Bridge chips could clock memory higher than allowed via the maximum programmed multiplier. As a result, Intel now assigns what could be considered ludicrous ratios when new tech rolls out, but years later, those ratios could come in handy. For instance, DDR4 ratios were assigned at a maximum of 8000 MHz years ago, but it wasn't until recently that the world record was set at over 6000 MHz, and due to foresight, there is still multiplier headroom left over.

Around the Haswell generation, Intel began adding in features and knobs to ease overclocking, which required a transition from 'opportunistic overclocking' to designing and architecting for it. The initial overclocking team consisted of engineers that volunteered for the task, like Francois Piednoel, a famed former Principal Engineer at the company. This loosely-banded team pushed the limits of silicon for the sake of furthering overclocking capabilities, but Intel didn't establish a formal team and lab until 2016, which now consists of eight team members that work on OC full time. There are still several volunteers throughout the company that help out, many simply because they are, like us, chip enthusiasts that enjoy overclocking.

Some of the notable advances include working with the Israel Design Center to design in BCLK overclocking from Skylake's initial design phases, along with addressing the Haswell cold bug issue (FIVR doesn't respond well to subzero temperatures) and designing in a workaround to ensure that future chips didn't suffer the same fate. Other advances include the introduction of the AVX offset, an incredibly useful tool for overclocking that the team patented. That offset also plays a crucial role in the auto-overclocking Intel Performance Maximizer (IPM) feature, which the team developed the algorithms for and worked with the software team to implement.

And Intel's forthcoming discrete Xe GPUs? We're told that the overclocking lab will play a role in those products, too, including the GPU overclocking software. There are a few 'organizational boundaries,' so the efforts won't be identical, but the team can define a wish list and also work with the GPU team to ensure those features make their way into the IPM software. That means the same one-button overclocking approach will come to Intel's GPUs.

We fully expect that overclocking Intel's discrete GPUs will be part of the lab's mission, and when asked, we were told with a smile that "some things" had been removed from the lab prior to our arrival. We suspect that 3D benchmarking of the new Xe cards is already underway. 

The OC team also helped pioneer per-core CPU overclocking to expose extra performance from the diverse range of CPU core capabilities, and that feature eventually evolved into Turbo Max 3.0.

As it does with several other groups inside Intel, the team also has an integral role in developing software, like key modules inside XTU, including developing the XTU benchmark and adding in application profile pairing. The team also assures that all of the necessary OC'ing knobs are exposed and working correctly in Intel's various software utilities.

As you can imagine, there are likely also other areas the team has developed, or is developing, for new tech for that can't be disclosed because they will worm their way into future products. As Ragland refers to it, the "OC pipeline of innovation" never stops.

Even with the focus on exposing features and assuring overclockability, the Silicon Lottery still applies, but this team's mission is to increase your odds of getting a tunable chip. In fact, the team now compiles overclockability reports for high-ranking Intel executives as a key checkpoint in the standard chip design flow, a process that started with the Devil's Canyon processors. To compile those reports, the team culls 50 randomly-selected processors from the hundreds of chips it tests for each new design, then charts out several key metrics, such as average OC frequencies, so the company can assure new chips are worthy of the "K-series" badge.

Enthusiast and Vendor Engagement

Legendary overclocker Allen "Splave" Golibersuch in action at the lab

Legendary overclocker Allen "Splave" Golibersuch in action at the lab (Image credit: Tom's Hardware)

One of the biggest steps forward came from merely engaging the community. In the past, Intel didn't directly promote overclocking with sub-zero cooling solutions, but that changed with the company's first subzero demo with cascade cooling, done by legendary overclocker Charles "Fugger" Wirth. 

Intel made its first live LN2 demonstration at IDF 2015 with overclockers Allen "Splave" Golibersuch, Fugger, and L0UD_SIL3NC3. The overclockers set an XTU world record live on stage in under two minutes, and now LN2 demonstrations at Intel's events are a common occurrence. 

Intel also regularly engages leading overclockers, like Der8auer (among many others). Intel says it works with ten to 15 of the top overclockers under NDA so they can visit its lab to test new chips. These overclockers test chips and give Intel feedback on several characteristics that are important for overclocking, like features and cooling methods. 

Intel team has also begun interfacing directly with the HWBot team that maintains the database of official world overclocking records, but that isn't strictly an "Intel initiative." Rather, the team does this to keep a healthy relationship with the enthusiast community. Case in point: The first-gen XTU benchmark had scalability issues beyond ten cores and could be 'hacked' to give out erroneous results, but the team sought feedback from the HWBot team to ensure that the second-gen XTU benchmark (which is already rolled into XTU) met their expectations. 

All of these actions are possible because of a shift in Intel's thinking in regards to interfacing with the overclocking community: In the past, the PR and marketing teams dealt with these 'external customers,' but Intel has opened up the process to allow the team cooking up the silicon to work directly with the community. 

But the work doesn't stop there. In fact, that's only the beginning. Intel's internal teams spend a vast amount of time overclocking chips themselves, often breaking world records that will never see the light of day, and work with motherboard vendors and other ecosystem partners to assure that not only the chips, but the platforms, too, are optimized for overclocking. 

Let's see what that looks like, and cover some of the gear they use in the lab. 


MORE: Intel & AMD Processor Hierarchy

MORE: All CPUs Content

Paul Alcorn
Managing Editor: News and Emerging Tech

Paul Alcorn is the Managing Editor: News and Emerging Tech for Tom's Hardware US. He also writes news and reviews on CPUs, storage, and enterprise hardware.

  • Dark Lord of Tech
    Can you get the AMD tour? Would love to see that.
  • PaulAlcorn
    Dark Lord of Tech said:
    Can you get the AMD tour? Would love to see that.

    I'll jump on a plane the second it is offered :)
  • bit_user
    @PaulAlcorn , thanks for the awesome piece!

    I'm still making my way through it, but wanted to draw special attention to this bit:
    the engineers told us they feel perfectly fine running thier Coffee Lake chips at home at 1.4V with conventional cooling, which is higher than the 1.35V we typically recommend as the 'safe' ceiling in our reviews. For Skylake-X, the team says they run their personal machines anywhere from 1.4V to 1.425V if they can keep it cool enough, with the latter portion of the statement being strongly emphasized.

    At home, the lab engineers consider a load temperature above 80C to be a red alert, meaning that's the no-fly zone, but temps that remain steady in the mid-70’s are considered safe. The team also strongly recommends using adaptive voltage targets for overclocking and leaving C-States enabled. Not to mention using AVX offsets to keep temperatures in check during AVX-heavy workloads.
    Thanks for that!
  • StewartHH
    Some one should comparison between different vendors die size like Intel 10nm vs AMD 7nm to see if there is actually performance gain. I would use per-core speed and not taking multiple cores into account.
  • bit_user
    @PaulAlcorn , uh oh. Now that I just finished heaping praise, I've got a gripe. In the penultimate paragraph:

    ... assures that the learnings lessons and advances made in the overclocking realm ...

    I was saddened to see the "learnings" virus infecting your otherwise admirable writing.

    I think "learnings" is one of those pseudo-jargon words that MBAs and other B-school types like to throw around, out of jealousy for practitioners of real professions. Everyone from auto mechanics to accountants, lawyers, and doctors needs jargon to adequately and efficiently express concepts and constructs central to their work. However, common sense pervades business to such a degree that I think they're embarrassed by how easily understandable it'd be, if they didn't inject some fake jargon to obscure the obvious. The resulting assault on the English language is disheartening, at best.

    Yes, if you've ever heard of her, you probably guessed I'm a fan of Lucy Kellaway, former journalist of the Financial Times and BBC. Worth a read:

    The 8 Lucy Kellaway rules for claptrap and the fundamental theorem of corporate BS
    Lucy Kellaway’s dictionary of business jargon and corporate nonsense
  • Gurg
    AMD CTO Mark Papermaster: "you can't rely on that frequency bump from every new semiconductor node." AMD's future outlook of very limited frequency bumps, performance increases only from more cores and expensive software modifications to use more cores.
    VersusIntel Ragland: "People who think this the end of the world for overclocking because our competitors' 7nm has very little headroom, that's not true. Intel is all about rock-solid reliability; our parts aren't going to can count on your part running at spec, so there's so much inherent margin that we will always have overclocking headroom...I think users will be happy with the margin we can offer in the future."

    Ouch! Intel's Ragland really "punked" AMD's negative outlook.

    PS Great fascinating article
  • jiang-v
    Anyone knows how to made contact with them? cause I fould a big bug on 10th corex chip about adaptive mode overclocking
  • nofanneeded
    In the past OC gave a huge difference , today we can easy hit 4.4 all cores without OC and this is more than enough for me.

    for me OCing is dead. and I dont care about missing 5 fps.

    I put the price difference in a better GPU ...
  • CompuTronix
    Outstanding article! Thank you, Paul! I would love to have been there. I have a few dozen questions that the Team may or may not have been allowed answered.

    However, like bit_user, I found it of particular interest that the Team was forthcoming regarding specific voltage and temperature values they're comfortable with running on their personal home rigs, which max out at 1.425 and 80°C. With respect to electomigation and longevity, every day in the forums we see many overclockers express their concerns over these very issues.

    On their website, Silicon Lottery shows Historical Binning Statistics that include the Core voltages used to validate their overclocked 14 and 22nm processors. For 22nm the maximum is 1.360. For 14nm the maximum is 1.456. While Intel's warranty is 3 years, Silicon Lottery's warranty is 1 year, which suggests at least one reason for the voltage difference between Intel's Team and Silicon Lottery.

    Here's a forgotten link to a revealing Tom's Hardware video interview of July, 2016, with Intel's Principal Engineer (Client Computing Group), Paul Zagacki, where BGTnJkuqlbo']Intel Discusses i7-4790K Core Temperatures and Overclocking. The video coincides with the formation of Intel's Overclocking Lab, also in 2016. In the video, Intel points out that overclocking abilities begin to "roll off" above 80°C, which agrees with the value the Team revealed in your article.

    While Core temperatures, overclocking and Vcore are often highly controversial and hotly debated topics in at least the overclocking forums, the term "electromigration" is closely related to a much less known term, which is "Vt (Voltage threshold) Shift". With respect to voltage and temperature, the two terms describe the causes and effects of processor and transistor "degradation" at the atomic level.

    In the Intel Temperature Guide, in Section 8 - Overclocking and Voltage, I created a table for Maximum Recommended Vcore per microarchitecture from 2006 to the present. For 22 and 14nm, those values are 1.300 and 1.400 respectively. I also created a graph showing the Degradation Curves for 22 and 14nm processors. The table and graph helps overclockers get a better perspective of the degradation and longevity issue:

    Sparing our members and visiting readers the deep dive, Vt Shift basically represents the potential for permanent loss of normal transistor performance. Excessively high Core voltage drives excessively high current, power consumption and Core temperatures, all of which contribute to gradual Vt Shift over time. Core voltages that impose high Vt Shift values are not recommended. The 14nm curve suggest 1.425'ish is the practical limit, which also agrees with the value the Team revealed in your article. The curve also suggests that Silicon Lottery might be pushing the edge of the envelope a bit.

    The concern here is that when novice overclockers casually glance around the computer tech forums, where conflicting and misleading numbers get flung around like gorilla poo in a cage, many don't realize through the fog of all the confusion that one size Vcore does not fit all. Aside from high Core temperatures, Vcore that might be reasonable for one microarchitecture can degrade another. So 22nm Haswell users now wanting to overclock their aging processors to keep up with today's games need to heed the degradation curves, which applies as well to 14nm Skylake and Kaby Lake users.

    CT :sol:
  • JamesSneed
    Paul this was a wonderful article. Seriously, nice job.