Gigabyte overclockers crank the knobs on a Core i7 to 7.5 GHz

Gigabyte wasted no time getting to the business of overclocking after the introduction of Intel's Core-X CPUs and the accompanying X299 platform at Computex last week. The motherboard manufacturer's in-house overclocker HiCookie and his five-person team spent the hours of the show rocketing an Intel Core i7-7740K past 7.5 GHz and setting four new 3DMark records.

The team used liquid helium at -220° C, a Corsair AX 1500i power supply, and one of the Gigabyte's own X299-SOC Champion motherboards to set the records. The 7.5 GHz CPU clock speed record was achieved using a Corsair Vengeance LPX 16GB 4333 MT/s memory kit, while the 3DMark records were set using G.Skill's Trident Z 3600C17 memory modules and a pair of Gigabyte Aorus GeForce GTX 1080 Ti graphics cards.

The CPU clock speed record is the headline news, but the team also set records of 356,678 points in 3DMark03, 737,222 in Aquamark, 71,928 in 3DMark06 with one GeForce GTX 1080 Ti, and 71,176 in 3DMark06 with two GTX 1080 Ti cards.

Comments closed
    • LauRoman
    • 3 years ago

    Still doesn’t make x299 and or kaby-lake-x a good choice…

    • ptsant
    • 3 years ago

    Helium is in very short supply. I’d rather get the Ripper.

    • ronch
    • 3 years ago

    I haven’t been following overclockers but I’d like to know which processor holds the highest overclock record today? Is it still the FX at 8+ GHz?

    Edit – ok after some Google-fu I think the record holder is an FX-8350 at 8.79GHz. Not too shabby for a 32nm chipotle. See, the FX is still a nice piece of engineering, despite all its shortcomings.

      • DrDominodog51
      • 3 years ago

      [quote<]See, the FX is still a nice piece of engineering, despite all its shortcomings.[/quote<] [url=http://hwbot.org/submission/2440102_wytiwx_cpu_frequency_celeron_d_352_8543.71_mhz/<]Except Intel did the same five years prior.[/url<]

        • ronch
        • 3 years ago

        Amazing that a 65nm Celeron reached that clock. Do note though, that the FX is also a far more sophisticated chip.

          • Krogoth
          • 3 years ago

          Netburst was designed for massive clock speeds, but physics made it difficult to achieve though.

          The canceled Tejas was originally planned to scale up to ~10Ghz

            • tipoo
            • 3 years ago

            There are some things I wish I could see the results of in a what-if machine, the Tejas is one of them.

            Tejas is now a fighter jet at any rate 😛

          • maxxcool
          • 3 years ago

          And both are unusable at those speeds so it matters not.

            • ronch
            • 3 years ago

            Those extreme overclocking stunts remind me of big, unstable atoms they create in the lab that last only for a gazillionth of a second. Cool but not really all that useful in daily life.

    • arunphilip
    • 3 years ago

    [Quote]71,928 in 3DMark06 with one GeForce GTX 1080 Ti, and 71,176 in 3DMark06 with two GTX 1080 Ti cards[/quote]
    I don’t get this – so 2 cards are actually worse than one? Or was the test CPU bound even with that overclock?

      • DrDominodog51
      • 3 years ago

      The benchmark appears to be CPU bound. The two-card run only did better than the one card run in one subtest out of six, and in the rest performance suffered.

    • ImSpartacus
    • 3 years ago

    I’m beginning to think these silly sponsored overclocking stunts are for purposes of attaining “halo” product status and the mindshare that comes with it.

    Otherwise, I just don’t get it.

      • ronch
      • 3 years ago

      Of course it is.

      • DPete27
      • 3 years ago

      [url<]https://www.youtube.com/watch?v=9CS7j5I6aOc[/url<]

    • Krogoth
    • 3 years ago

    This is hardly new.

    Regular Socket 1155 Kaby-Lake chips can go just about as high with LN2 and volts. It is really about silicon lottery. I do suppose that Socket 2066 can allow more current to go through but Skylake-X chip are going to chips that need it more for extreme overclocking.

    There’s no reason to get Kaby-Lake-X unless you want to invest into the Socket 2066 platform and plan on getting a future drop-in upgrade with a “cheap/used” Skylake-X chip or its successor.

    You cannot utilize most of the platform features. Two DIMM channels are disabled since memory controller Kaby Lake-X is still dual-channel DDR4. It possible that some of the 16x physical PCIe slots are disabled as well since there’s not enough PCIe lanes on with Kaby-Lake-X chips and I doubt motherboard vendors are going throw in an extra PLX to make up for this. The intergrated GPU is obviously useless since Socket 2066 has no pins allocated for it. The worst part is that you need to get a special DLC key to unlock RAID 1/5 on the PCH for SSD M.2 media and you can only use Intel-brand SSD M.2 media. Stupid market segmentation at its finest.

    The only reason I suspect that Intel even decided to bother to make “Kaby Lake-X” is because larger Skylake-X chips have significant yielding issues. There’s not enough supply for lower-end HEDT market for motherboard vendors to make Socket 2066 boards in volume. Intel was forced to convert regular Skylake silicon into this platform to make up the difference. It has nothing to do with Ryzen recent market disruption either since this was planned well in advanced before Ryzen hit retail channels.

      • DrDominodog51
      • 3 years ago

      Yeah. The only really interesting thing about this OC is the lack BCLK overclocking involved.

      This only set world records in (mostly) irrelevant single threaded benchmarks. Elsewhere, the 6950X and 1800X dominate.

      • ImSpartacus
      • 3 years ago

      Is there at least an option for a RAID 1/5 season pass?

      • ptsant
      • 3 years ago

      I don’t understand why people downvoted you.

      The only reason for Kaby on the 2066 platform is as a placeholder for people who can’t afford to get everything at once and plan to upgrade at some point. I mean, the PCI lanes are not there, the memory channels are not there. Half the motherboard is simply useless. So, you get to pay $300-400 for the motherboard so that you can’t use it? Who thought of that?

        • Krogoth
        • 3 years ago

        It is because Intel-fans don’t want to read between the lines. Regular Kaby Lake on 2066 socket is only happening because it is likely there Intel is having problems producing Skylake-X chips in quantity for HEDT market. i7-7720 and i7-7740 are essentially “Celeron” of the HEDT family.

        They never needed to convert normal desktop silicon on their previous HEDT platform (Socket 2011). If the yields were good on Skylake-X chips then the 7800-7820 could easily hold the low-end part of the platform. Intel wouldn’t bother spending the extra money and effort in converting Kaby Lake silicon.

    • siberx
    • 3 years ago

    That must have been quite the exotic cooling setup if they were using liquid helium at nearly 50c above its atmospheric boiling point.

      • Mr Bill
      • 3 years ago

      Yes and helium is in such short (and expensive) supply; what a waste if there is any leakage. We get all our helium from natural gas reservoirs; which are the only natural traps sufficiently tight to accumulate it in quantity from radioactive decay in the underlying strata. Liquid Nitrogen would be so much cheaper at -195C and loss to evaporation is not a loss of the resource.

        • Stargazer
        • 3 years ago

        But what if the atmosphere is contaminated by nitrogen!
        <.<

        • ptsant
        • 3 years ago

        LOL, I wrote the same thing before I read your comment 🙂

        • Krogoth
        • 3 years ago

        Helium is only rare on Earth because almost all of the primordial helium escaped the Earth during the Hadean Eon. The helium supply comes entirely from Alpha decay of massive radioactive elements.

          • Mr Bill
          • 3 years ago

          +3 Krogoth is correct; no need to ding him.

      • Krogoth
      • 3 years ago

      Liquid Helium a very crappy coolant though (it has poor heat capacity per molar). It is only good for infrared telescopes and superconductors.

      Liquid Nitrogen is actually a better coolant for extreme overclocking cooling despite having higher melting/boiling points.

Pin It on Pinterest

Share This