ROG Rampage VI Apex shows simple tweaks can tame X299 VRMs

Roman "der8auer" Hartung recently cast shade on a handful of motherboards based on Intel's X299 chipset. According to the extreme overclocker, the boards' voltage-regulation circuits can hit temperatures as high as 105° C with a Core i9-7900X overclocked to a relatively modest 4.6 GHz. 105° C is the range in which many such circuits will begin to throttle the system for safety. Hartung also criticized the single eight-pin EPS +12V power connectors and current VRM heatsink designs on many X299 boards.

Now Hartung is back in a new video discussing how some simple changes on Asus' ROG Rampage VI Apex motherboard avoid these temperature and power limits. While his comments reflect his feelings about Asus' overclocking-focused board specifically, the broader takeaway is that relatively simple improvements could be incorporated into X299 boards from all manufacturers to improve VRM temperatures and performance headroom.

In particular, Hartung appreciates the dual eight-pin EPS +12V connectors, high-quality inductors, and the beefy heatsink attached to the VRM components on the Apex. The VRM heatsink is of particular interest. Hartung says Asus has revised the cooler since the board was first shown off at Computex, and it now features a finned primary heatsink over the board's inductors and MOSFETs. That primary heatsink connects to another, larger hunk of aluminum perched over the rear I/O ports using a single heatpipe. The VRM cooler also has threaded holes for attaching a fan to provide additional airflow to the heat-producing components. Hartung is somewhat less enthused about a small slab of aluminum strapped to a trio of capacitors on the reverse side of the motherboard.

The professional overclocker presents test results that show his i9-7900X-plus-Rampage-VI-Apex test system stable with a 4.9 GHz overclock. The CPU was pulling a hair-raising 340W from the VRMs in this setup, but the system wasn't throttling due to the temperatures of that circuitry. The processor was limited instead by the thermal dissipation capacity of Hartung's liquid cooler. Attaching a small fan to the VRM heatsink reduced VRM temperatures from a hot-but-within-component-specification 103° C to a much-less-alarming 87° C under these conditions.

This older photo of the Apex lacks the heatsink fins that will presumably be found in the retail model.

The Asus ROG Rampage VI Apex is sure to be an overclocker's delight when it is released to the public, but those looking to build a more productivity-focused X299 system might not appreciate the board's four DIMM slots instead of the usual eight. We can only hope that all motherboard manufacturers incorporate similar design changes into the VRM heatsinks of their X299 products soon.

Comments closed
    • Chrispy_
    • 2 years ago

    HahahahahhhahhhahahaaahahhahHahahahhahhhahhahahhahhahahh
    Hahhhahhahh…..

    <deep breath>

    Ahaahahhahhahahahahahahhahhahhahah!

    I’ve been criticising these firms for [i<]years[/i<] about their stupid, low-surface-area aesthetic motherboard heatsinks and then I watch this video. Asus solved the heatsink problem by.... (wait for it) [url=https://d.justpo.st/media/images/2014/03/f03c0766621a2dbc67fee991b9be7376.jpg<][b<]PUTTING FINS ON THEIR HEATSINKS[/b<][/url<].

      • Waco
      • 2 years ago

      My thoughts exactly. The insanity of form over function on high end boards is ludicrous.

      • Wirko
      • 2 years ago

      What went wrong here, don’t finFETs have enough fins to keep your whole house cool?

      • derFunkenstein
      • 2 years ago

      BRILLIANT!

        • BurntMyBacon
        • 2 years ago

        [b<]BRILLIANT![/b<] In light of the double facepalm link, I thought it deserved another:

          • UberGerbil
          • 2 years ago

          One better: [url=http://i.imgur.com/Lwij0Cv.jpg<]triple[/url<]

      • cynan
      • 2 years ago

      I don’t get it.

      • Srsly_Bro
      • 2 years ago

      Is this the part where you reference a post from months ago and then claim that you and only you could have possessed the foresight to define the mistakes the aforementioned firms made over the years?

      Oh wait…

        • BurntMyBacon
        • 2 years ago

        Nah.

        Chrispy_ has definitely been vocal (contemptuous even) about the ever decreasing thickness of products in general at the expense of features, performance, and durability for no other reason than some misguided sense of aesthetics. Chrispy_ certainly isn’t the only one, but may possibly be the most vocal on TR.

        • Chrispy_
        • 2 years ago

        Heh, I don’t think Chucky’s here to take the bait.

        But no, there’s usually a large number of gerbils agreeing with me when I moan about silly heatsinks. They either look stupid, don’t work, get in the way of other things, or a combination of those things.

    • ptsant
    • 2 years ago

    I’d rather have 2x the RAM instead of 10% more perf at the cost of 50% more heat.
    Workstation board, please.

      • Kougar
      • 2 years ago

      You know they have a half-dozen other X299 models right? This is the extreme overclocking model…

    • Anomymous Gerbil
    • 2 years ago

    What are the additional slots next to each pair DIMM slots? The linked Rampage VI Apex page doesn’t have much info.

      • chuckula
      • 2 years ago

      Those slots are for M.2 NVME drives that stick out from the motherboard for better cooling.

        • kuraegomon
        • 2 years ago

        I went with a more effective – and admittedly more expensive – option: [url<]https://www.angelbird.com/prod/wings-px1-1117/.[/url<] But, this is for a Samsung 960 pro 1 TB on my workstation build where I managed to hit throttling issues. I imagine that those slots on the Apex would work very well with a tower-style CPU air cooler, or a top-mounted radiator. You'd get much cleaner airflow with the later setup, actually.

          • UberGerbil
          • 2 years ago

          Does that angelbird thing have thermal pads or paste to link the exterior with the chips on the M.2 gumstick? Trying to understand how that shell, even if it’s copper, makes the thing cooler than leaving it naked to the airflow.

            • thecoldanddarkone
            • 2 years ago

            It’s aluminum and it does use thermal pads. I owned one and it worked well (not 70 dollars well though). I ended up getting an intel 1.2 TB 750 instead of using the 950pro.

            This was when the 960pro had just been released and wasn’t in stock by any retailers.

        • derFunkenstein
        • 2 years ago

        Wait, really? I thought they were just more DIMM slots so you could put 4 DIMMs on each of the 4 channels, but thought it was weird that it was four-way asymmetrical.

          • chuckula
          • 2 years ago

          They are using the DIMM physical form factor but if you look carefully you’ll see a metal insert near the top of the slot that prevents you from accidentally putting a memory DIMM into the slot.

            • derFunkenstein
            • 2 years ago

            Hey that’s a cool idea. I had no idea this was a thing now. Thanks!

    • UberGerbil
    • 2 years ago

    [quote<]The professional overclocker[/quote<]Wait, [i<]professional overclocker[/i<]? So he makes most or all of his living doing this? I've never really given any thought to this niche. What are the sources of his revenue? I know there are contests with prize money, and presumably he has a Youtube channel he can monetize. But does he have sponsors? Does he get the gear gratis from the manufacturers? I'm not trying to be one of those "LOLBIAS!" dumbass conspiracy theorists; I can fully appreciate that sites like TR can be given free review samples without having their objectivity compromised (though I also need a track record of unbiased reviews before I'm willing to not take it with a grain of salt). I'm just trying to understand a subculture and corner of the industry I've never given much thought.

      • Kougar
      • 2 years ago

      Some actually do, yes. While there are multiple annual overclocking events with free hardware provided to use and abuse in order to set records with said vendor’s hardware, those events and prize pools are simply how many have gotten started in the scene. Ones that stand out across multiple events tend to gather a following and that becomes marketable.

      Vince “K|NGP|N” Lucido is one, for several years now he has been making his money by working directly with EVGA to test their products and design customized premium “K|NGP|N” edition cards with his tweaks baked into the hardware.

      Other graphics & motherboard makers have hired well-known successful overclockers, either just to make PR events, create custom hardware sold under their alias, actual mainstream product testing, or any combination of the above.

        • UberGerbil
        • 2 years ago

        So in some cases their expertise enables them to draw consulting fees (or at least free hardware) from the mfrs. That makes sense. Thanks for the illumination.

          • Kougar
          • 2 years ago

          Well at the top they tend to be directly hired by said companies, will get a desk and lab for playing with the hardware. I’m not sure if Kingpin works directly for EVGA or it’s just a licensing/brand partnership, but I know ASUS and Gigabyte have had famous overclockers in their employ for the last decade.

          As for the overclocking events, all the hardware tends to be provided on loan for that event. There’ll be a tray of processors in particular that tend to get reused for events, so they don’t get to keep anything except prizes and winnings. Or at least that’s how it was when I attended one. Considering the condition of the motherboards / GPUs after they get done most people wouldn’t want to keep them anyway….

            • jihadjoe
            • 2 years ago

            Yeah they’re basically like test drivers for automobile companies. The Valentino Balbonis of the tech industry.

      • derFunkenstein
      • 2 years ago

      If someone gives you $50 to do it in your spare time, are you a professional?

    • Voldenuit
    • 2 years ago

    Anyone remember when chipsets* were hot enough to require tiny, noisy, heatsink fans?

    I had a Gigabyte NForce4 mobo with a noisy fan on it that I replaced with a Thermalright northbridge heatsink (yes, it was a time when pc cooling companies made aftermarket [i<]chipset heatsinks[/i<]). Hopefully we're not going to go back to those days. * yeah yeah, those are *VRM*s on the X299 boards.

      • Kougar
      • 2 years ago

      I doubt it, there’s no standard to design around. Those northbridge coolers had a standardized pinout spacing on two opposing corners that they could use to mount through the board. Those little chipset fans were always the first thing to wear out in a system, ugh…

      We’ll probably see an uptick in watercooled VRMs again though, already seen a few block designs for the X299’s.

      • dyrdak
      • 2 years ago

      “Anyone remember when chipsets* were hot enough to require tiny, noisy, heatsink fans?” – you’d still need those fans if manufacturers maintained the size of heatsinks.
      BTW, with the northbridge and numerous other features integrated into the CPU, power requirements of the supporting chipset(s) had to drop.

        • jihadjoe
        • 2 years ago

        I remember just ripping that crap straight off and going passive with a bigger heatsink. Often cooler and no chance of the dinky fan going wrong. Only drawback was the middle group of PCIe slots couldn’t take in long cards anymore, but most of the time the only full length card would be the GPU so no biggie.

      • UberGerbil
      • 2 years ago

      Dude, I remember when CPUs didn’t need fans. Or heatsinks. They were just another chip on the motherboard. Of course they were only running at 1 – 8 MHz…

        • Voldenuit
        • 2 years ago

        Yeah, also the days when graphics cards didn’t need fans, and just had extruded solid metal fin heatsinks.

          • BurntMyBacon
          • 2 years ago

          Or the days when they were also just another chip on the card. No heatsink required. Of course heatsinkless CPUs take you back even farther.

          I think the most modern CPU designed to run with no heatsink was a 40MHz AMD 80386 compatible processor. The most modern socketless CPU with no heatsink was … 8086?

          Should have read one more post:
          [quote=”Wirko”<]Even the AMD Am386DX-40 ran bare.[/quote<]

        • Wirko
        • 2 years ago

        Even the AMD Am386DX-40 ran bare.

          • jihadjoe
          • 2 years ago

          The first CPU I had that needed a heatsink was a Pentium 100, and there wasn’t even any thermal compound, just sort of plonked on top of the CPU.

      • Captain Ned
      • 2 years ago

      Most anyone with a VIA board quickly replaced that 40mm chipmunk caller with a Zalman ZM-NB47 passive cooler.

        • Chrispy_
        • 2 years ago

        I certainly did. I don’t think we had any choice back in those days – the Abit and Asus 40mm fans lasted for approximately two weeks and the Zalman was the only aftermarket chipset cooler that was readily available.

    • USAFTW
    • 2 years ago

    You know you’re desperate when you have to force your engineers to literally cut the corners off your product to pander to “L33T G4M3R5”. Who in their right mind thought that that was a good idea?

      • TwistedKestrel
      • 2 years ago

      Considering the location of the cutouts, I would assume it is for cable routing/hiding purposes

    • chuckula
    • 2 years ago

    Internal though process of motherboard OEMS: Heatsinks were invented to take heat away from components that get ho—- OOH LETS ADD MOAR RGB LEDS!!!

    Wait what were we doing?

      • BurntMyBacon
      • 2 years ago

      [b<]Engineering:[/b<] These heatsinks were designed specifically to remove heat from heat producing components and prevent them from overheating. We've designed some VRM heatsinks that will keep the VRMs at a nice comfortable temperature even under heavy overclocking loads. [b<]Marketing:[/b<] Those heatsinks look ugly. Nobody wants your ugly heatsink. Make them look better so people will buy them. [b<]Management:[/b<] True, we do need people to buy our product. Furthermore, that heatsink is 25% the cost of the board. Can you make them cheaper and better looking without sacrificing any performance. [b<]Engineering:[/b<] No. Modifying the shape and materials will have some detrimental effect on performance. You'll need to decide what the right balance is. [b<]Management:[/b<] We don't like compromises. We'll put more people on the job. [b<]Engineering:[/b<] We've come up with a design that should cool the VRMs adequately under normal loads and allow reasonable overclocking without hitting the thermal limit. Oh, and it is cheaper and looks like .30-06 bullets. [b<]Marketing:[/b<] I still can't help but think something is missing. Out studies show us that ever since we've put RGB LEDs on all of our gaming products, 100% of our gaming sales have had RGB LEDs. If you want to sell to the gaming crowd, you HAVE to put RGB LEDs on it. [b<]Engineering:[/b<] LEDs don't make sense on a bullet. Also, if we remove any more material, the board won't be able to handle overclocking loads. Furthermore, adding LEDs would make it cost as much as our original heatsink, defeating the purpose of reducing the materials used. [b<]Management:[/b<] Well, it is hard to argue with 100% adoption rates and we don't warranty our board against overclocking anyways. Add the RGB LEDs. [b<]Engineering:[/b<] The board is marketed for overclocking. Do you really think people will be happy when the VRM temperatures won't allow it to be overclocked. [b<]Marketing:[/b<] That's just marketing. Besides, they can overclock it. They just can't do it for very long. That's why we put a warning that overclocking will void your warranty. [b<]Engineering:[/b<] We've modified the heatsink to your specifications. Your VRM heatsinks look like rainbow colored tracer rounds that you can see firing directly into the sun. The VRMs operate barely within their thermal limits guarantying that overclocking will be an exercise in frustration and normal operations will cause the board to fail about a month after the warranty is up. [b<]Marketing:[/b<] Looks good. Overclocking is supposed to be difficult. Otherwise everyone would do it. [b<]Management:[/b<] Yes, and it has the added benefit of generating future sales. Once the warranty is up, the board will be hopelessly obsolete anyways. [b<]Engineering:[/b<] You don't think that they'll move on to a company that has easier overclocking and allows them to use their processors as long as they are useful? I'd like to direct your attention to the number of Sandy Bridge processors still in use. [b<]Management:[/b<] What does bridge processing have to do with motherboards? Anyways, those other manufacturers are either doing the same thing, or they won't have RGB LEDs and our statistics show a 100% RGB LED adoption rate. [b<]Engineering:[/b<] <Collective Face-Palm>

        • Chrispy_
        • 2 years ago

        If management and marketing ever even bothered having meetings with engineers in the first place, I’m pretty sure your scenario would be accurate.

        Based on my experience of real design meetings, the engineers are only given the initial brief and introduction to the project/product/design AFTER the final decision has already been made by management + marketing idiots.

          • BurntMyBacon
          • 2 years ago

          [quote=”Chrispy_”<]Based on my experience of real design meetings, the engineers are only given the initial brief and introduction to the project/product/design AFTER the final decision has already been made by management + marketing idiots.[/quote<] This is my experience as well. The scenario above is more of a Dilbert style parody. It's more laughs even if it relaxes the accuracy a bit. In reality, the most likely interaction between engineering and marketing might be an occasional passing word in the hall.

Pin It on Pinterest

Share This