EVGA offers fixes for overheating GTX 1070 and 1080 cards

Internet forums and rumor mill sites have been ablaze with chatter regarding overheating PWM circuits and memory chips on EVGA GeForce GTX 1070 and 1080 graphics cards fitted with the company's ACX 3.0 coolers. EVGA has issued a formal response regarding the problems, and offers two different remedies that can be administered by users without shipping their cards in for service.

EVGA says the PWM circuits and memory chips run "marginally within spec" while "under extreme circumstances," particularly when running stress-test programs like FurMark. The company says GPU temperatures are around 70° C when running the notoriously-demanding FurMark stress-testing tool. The cards' thermal management sets the fan to approximately 30% of its maximum speed during these conditions. EVGA will offer BIOS updates to raise fan speeds in order to combat potential overheating, and will cross-ship cards for users uncomfortable with flashing the BIOS themselves.

The manufacturer is also offering optional thermal pad kits for owners of affected cards. The kit is available free of charge, but users must remove the cooler from their card to install it, as the company hasn't offered to cross-ship cards for this purpose. The company will still "stand behind its customers with full warranty and cross-shipment" in the event of accidental damage during thermal pad installation. EVGA hasn't specified whether or not future cards will include the thermal pad kit from the factory.

The full list of affected cards, EVGA's press release, and the link to request a thermal pad kit can be found on the company's website.

Comments closed
    • Ultracer
    • 3 years ago

    In other news:” EVGA gpus explode!”

    Poor Note 7, it’s so annoying to hear people say sth explodes while the device only ignites itself…

    • ronch
    • 3 years ago

    Really reminds me of the Note 7 fiasco.

    • Dark Pulse
    • 3 years ago

    Call me crazy, but when you’re spending $600-800 on a GPU, the last three words you want to hear, when it comes to the card’s functionality, is “marginally within spec.”

    Thankfully I got an MSI GTX 1080 Gaming X 8G, but it’s pretty disappointing when people who shell out money for expensive cards can’t even have the satisfaction of knowing it’ll be stable for a few years.

      • ClickClick5
      • 3 years ago

      This is why I have ended up not trusting any company to keep things cool in stock form. Last three GPUs I have purchased, I also buy a new cooler. Then temp gun the board’s chips to make sure they stay cool, and well below max temp. My 980 tops out at about 53c now.

      I also repaste the southbridge with AS5.

      Heck, I repasted my PS4 the day I brought it home. Never even plugged it in. And as always, there was an enormous amount on the APU.

      I can’t stand the “Well the limit is 90c, so our coolers are designed to keep the card in the safe zone….of ~85c.”

      That is like driving to the grocery store with your car right at redline…all the time.

      EDIT: All my consoles have been repasted. At least the ones that give off heat. The N64 and back were ok.

      /rant

    • Bensam123
    • 3 years ago

    Weird, I have some of these in high density mining rigs and they don’t get anywhere near as hot as what’s being shown in the photo. I just checked them with a infrared gun and they aren’t anywhere close. EVGA coolers tend to cool better then all the other brands except Asus (which they cool just as well as). The backplates on the MSI cards I have are actually about 10c hotter.

    MSI cards I’ve had nothing but issues with, they run hot and I’ve had the VRM controllers literally burn up, close to a 50-60% failure rate on those (1070s). That is basically what’s being described in this article. Gigabyte cards, the fans will start failing between 6-12 months after they start mining. The fans on those cards have been garbage for about the last 3-4 years. They all uses the same half height sleeve bearing fans. Even if they cool well, the fans inevitably fail.

    They must be doing some really weird shit to get these sort of temps while keeping GPU usage down so the fan doesn’t ramp up… and this is coming from a miner.

    • xeridea
    • 3 years ago

    Couldn’t you just adjust the fan curve in software instead of flashing bios?

      • morphine
      • 3 years ago

      … that’s precisely what the BIOS flash does.

        • xeridea
        • 3 years ago

        I know, I am saying, couldn’t you use PrecisionX, Firestorm, Afterburner, etc to change it instead of risking flashing bios? Also easier for the masses. Changing fan behavior shouldn’t require a BIOS mod, I have changed fans on 1060 in software.

          • morphine
          • 3 years ago

          You can. But a BIOS fix is permanent and you don’t have to worry about anymore. Just applying a PrecisionX fix or something is a surefire way to forget about it later on.

          This can also be described from also protecting users from themselves (forgetting to re-run the fix). Besides, keep in mind it’s usually dead simple to update the BIOS, and there’s only [i<]one[/i<] unchangeable task to take care of. Anybody who's not in the know can easily screw up a fan curve and make it worse.

            • xeridea
            • 3 years ago

            Yeah it is more permanent, not saying bios mod is bad, just odd there is no mention of software fixes. I have never modded a GPU bios, though I have on motherboards for new CPU support. It’s probably not hard on modern GPUs, I have just never had a need.

    • Voldenuit
    • 3 years ago

    Every other story I’ve read on this has reported that evga f’ed up by forgetting to install the thermal pads in the first place, and that this is not a ‘typical’ configuration, but a massive cock-up at the factory.

    • psuedonymous
    • 3 years ago

    Another case of non-OEM cooling not being thought out beyond what the plastic shroud looks like. Though the previous worst offender was also EVGA (the ACX 2.0 design for the 970 having a heatpipe that didn’t actually contact the GPU, due to re-using old stock of heatsinks from the previous generation), so it may just be them not putting basic effort into thermal design.

      • Sabresiberian
      • 3 years ago

      Yeah I don’t buy EVGA for “marginally within spec” hardware. They reputedly have one of the best after-purchase support systems in the business, but I’d rather not have my video card fail and have to find out how good it is (or not).

    • wingless
    • 3 years ago

    This isn’t a surprise coming from EVGA. Watch ‘PCB Breakdown’ videos for various GPUs on YouTube. I only really trust ASUS and Gigabyte now.

    • Waco
    • 3 years ago

    I’m surprised subpar VRM cooling is still a thing.

    Heatsink manufacturers seem not to care though – I had a Swiftech waterblock (shortly) on my 4870X2 and I was told I was an idiot for caring that VRM temps were in the 120+ C range even with a stock clocked card running a not particularly Crossfire friendly game.

    • Skid
    • 3 years ago

    I disliked the fact the the fans on my EVGA 1080 SC ACX 3.0 card would idle until the GPU hit 60 C. Yikes. I used the EVGA Precision XOC software to change the fan profile to run the fans all of the time. GPU temp is now 35 C and hits only 65 C @ 1900-2000MHz under hard usage… well, gaming usage, not chronic, fanatic benchmark usage.

      • swaaye
      • 3 years ago

      I don’t see anything wrong with the stock behavior there. The idle fan feature is actually a modern benefit. 60C is not a problem whatsoever.

      • xeridea
      • 3 years ago

      Many, many new cards of all brands, and GPU vendors have this feature. 60C is nothing for a GPU, so it is good to increase the life of the fans, reduce power, dust and such. It is kind of pointless running fans at idle with a good cooler. The feature can be disabled of course for those that don’t want it.

    • Srsly_Bro
    • 3 years ago

    I wonder if this is an issue if running f@h 24/7.

      • general_tux
      • 3 years ago

      It hasn’t been so far for me with my 1070. Just stopped running f@h 24/7 for most of October with TRFrankenbot without any overheating issues. Still may have to take them up on the thermal pad kit though, just in case.

    • xeridea
    • 3 years ago

    Not surprising, I have an EVGA 1060 and it constantly throttles @ 82C. I don’t think it has any VRM or memory cooling either. Why are they using mediocre PWM on $600 card?

      • Srsly_Bro
      • 3 years ago

      Check to see if there is an air gap between the vrm and the hsf. It is speculated that this effects more cards with acx 3.0 design.

        • morphine
        • 3 years ago

        EVGA specifically states the issues are specific to the ACX 3.0 cooler design.

          • slowriot
          • 3 years ago

          And clearly EVGA is on top of things so we should just trust them…

      • swaaye
      • 3 years ago

      It throttles at only 82C?

      I think it’s hard to say if VRM and RAM cooling is necessary in a stock configuration if it has airflow from fans above. It’s not an uncommon thing, that’s for sure.

        • xeridea
        • 3 years ago

        82C is default thermal limit, which it hits no problem. I don’t overclock. If I lower power limit it helps a bit, but still bad. I have a Zotac card with same style cooler, runs 5-10C cooler and doesn’t throttle.

          • Voldenuit
          • 3 years ago

          Yeah, hitting 82C and subsequently throttling at stock clocks is not great cooling.

          My Gigabyte GTX 1070 Windforce stabilizes at 67C under load (overclocked @ 2012/9000 MHz), granted that cooler is something of a behemoth.

    • chuckula
    • 3 years ago

    Well snap.

    Haven’t had any problems myself, but I may be looking into this.

      • BoilerGamer
      • 3 years ago

      Same here, then again unlike some people I don’t buy a $600+ GPU to run Furmark 24/7.

        • K-L-Waster
        • 3 years ago

        Really? Then how do you get all that critical Furmark work done?

Pin It on Pinterest

Share This