The down and dirty on the hot and bothered ACX 3.0 cards

Subject: Graphics Cards | November 2, 2016 - 07:10 PM |
Tagged: pascal, nvidia, GTX1070, GTX1060, GTX 1080, fail, evga, ACX 3.0

View Full Size

Checklist time readers, do you have the following:

  • A GTX 1060/1070/1080
  • Which is from EVGA
  • With an ACX 3.0 cooler
  • With one of the model numbers above

If not, make like Bobby McFerrin.

If so, you have a reason to be concerned and EVGA offers their apologies and more importantly, a fix.  EVGA's tests, which emulate the ones performed at Tom's show that the thermal temperature of the PWM and memory was just marginally within spec.  That is a fancy way of saying that in certain circumstances the PWM was running just short of causing a critical thermal incident, also know as catching on fire and letting out the magic smoke.  They claim that this was because the testing focused on GPU temperature and the lowest acoustic levels possible and did not involve measuring the heat produced on memory or the VRM which is, as they say, a problem.

View Full Size

You have several choices of remedy from EVGA, please remember that you should reach out directly to their support, not NVIDIA's.  You can try requesting a refund from the store you purchased it at but your best bet is EVGA.

The first option is a cross-ship RMA.  Contact EVGA as a guest or with your account to set up an RMA and they will ship you a replacement card with a new VBIOS which will not have this issue and you won't need to send yours back until the replacement arrives.

You can flash to the new VBIOS which will adjust the fan-speed curve to ensure that your fans are running higher than 30% and will provide sufficient cooling to additional portions of the GPU.  Your card will be louder but it will also be less likely to commit suicide in a dramatic fashion.

Lastly you can request a thermal pad kit, which EVGA suggests is unnecessary but certainly sounds like a good idea especially as it is free although requires you sign up for an EVGA account.  Hopefully in the spare seconds currently available to the team we can get our hands on an ACX 3.0 cooled Pascal card with the VBIOS update and thermal pads so we can verify this for you.

This issue should not have happened and does reflect badly on certain factors of EVGA's testing.  Their response has been very appropriate on the other hand, if you are affected then you can get a replacement card with no issues or you can fix the issue yourself.  Any cards shipped, though not necessarily purchased, after Nov. 1st will have the new VBIOS so be careful if you are sticking with a new EVGA Pascal card.

Source: EVGA

November 2, 2016 | 07:40 PM - Posted by serpico (not verified)

"ensure that your fans are running higher than 30%" does that mean this failure is more probable at low graphical loads when the fan spins down? Cause otherwise isn't the fan usually higher than 30% anyways?

November 2, 2016 | 08:36 PM - Posted by Anonymous (not verified)

Hmmm.

In the video the card blew on a cold boot

November 2, 2016 | 08:54 PM - Posted by Anonymous (not verified)

Yes and no.

The card blew on an attempt to cold boot during troubleshooting, AFTER the card failed and the computer shut down completely. The damage had already been done when the computer shut down. While attempting to figure out what had caused his computer to just shut down, he tried to power it back up - with the PCI-E power un-plugged from the card, in a logical attempt to remove the card from the troubleshooting equation - and that's when the already-damaged VRM module popped. (Which says to me that the current that popped the module came from the PCI-E slot, which makes me hope that it didn't draw so much current to damage the motherboard, too.)

November 2, 2016 | 09:25 PM - Posted by Jeremy Hellstrom

VRMs don't gradually fail and they don't tend to go quietly either.

November 3, 2016 | 12:48 PM - Posted by Anonymous (not verified)

Talk about some flashy lighting, LCD's have nothing on that!
Flash, Crackle, and Pop! EVGA's new VRM mascots!

November 3, 2016 | 05:07 PM - Posted by Anonymous (not verified)

I wasn't suggesting that they do either of those things. You guys talked about it in the podcast last night, the guy came home to find his computer had shut down, and when attempting to start it up for troubleshooting, the VRM burst into flames and smoke.

November 3, 2016 | 08:32 AM - Posted by jabbadap (not verified)

And no power connectors connected. Ain't that reason to void warranty at least it's forbidden on many graphics card vendors installation guides.

I'm not saying that card might not have died already before that. EVGA made a very bad design flaw and there's no excuses of that. But you can brake even working card by just booting it up without all/none needed power connectors connected.

November 2, 2016 | 08:48 PM - Posted by Anonymous (not verified)

Upping the fan speed to "fix" inadequate cooling is not a solution. Anyway I thought Nvidia cards ran so much cooler so they didnt need much cooling...

Is EVGA liable if a fire started from your video card burns down the house?

November 2, 2016 | 09:20 PM - Posted by Anonymous (not verified)

Yes EVGA would be liable, if it can proven it was started by the graphics card catching fire.

November 2, 2016 | 09:23 PM - Posted by Jeremy Hellstrom

Consider Samsung for a recent example.

November 3, 2016 | 11:08 AM - Posted by Anonymous (not verified)

"Anyway I thought Nvidia cards ran so much cooler so they didnt need much cooling."

The GPU die is being cooled just fine. The problem is neglecting to cool the VRMs (which are not manufactured by Nvidia, and on EVGA's non-reference card are not even chosen by Nvidia).

When you go for an non-reference OEM design, you're rolling the dice on whether that OEM managed not to fuck things up.

November 2, 2016 | 09:24 PM - Posted by Anonymous (not verified)

Any miner worth their digital currency knows not to mine with an evga card (especially with the ACX cooler) or risk blowing your VRMs.

Why is this news now?

November 2, 2016 | 09:45 PM - Posted by Cory Brown (not verified)

EVGA has said that all new cards will have the thermal kit installed(new stock, not existing). The thermal kit should have been on there from the beginning, saving a couple of cents has came back to haunt them.

Props to EVGA for stepping up to the plate on this one and admitting a mistake.

EVGA has the best customer support out of all of the gpu vendors. Hence why I am STILL a loyal EVGA customer, IF they continue with the great customer support, I will continue to be a customer.

November 3, 2016 | 01:26 AM - Posted by pdjblum

Though rather not to have to call it, I am all about great customer service. Glad to know. Went from no way to I would consider getting an EVGA card after reading your post.

November 3, 2016 | 12:18 PM - Posted by Anonymous (not verified)

"EVGA has said that all new cards will have the thermal kit installed(new stock, not existing)."

Source? Nothing in their official statement on the issue has stated this. Their official statement on the issue stated that replacement cards would only have the new vbios.

November 3, 2016 | 04:18 AM - Posted by JohnGR

You can buy a 100mm x 100mm thermal pad on eBay for $1. I wonder how much it would cost EVGA to add those thermal pads from the beginning. Probably they would have filed for bankruptcy by now, if they where installing thermal pads in the first place. I would really want to see their financial report for the third quarter of 2017 and compare it with this year's third quarter's financial report.

As for the "fix" that puts the fans running at higher speeds, let's just remind that Nvidia invented first that fix with the bumpgate. That driver fix was meant to keep the chips cooler all the time, so they outlive at least the warranty period. If someone thinks that the firmware is enough, he would have to reconsider, except if thinking replacing his card before the warranty expires.

And that makes me think it will be a bad idea after a 12-24 months period from now, someone to choose to buy a second hand EVGA. You can't be sure how much the VRMs where stressed before the owner applying those pads.

November 3, 2016 | 04:35 AM - Posted by Jann5s

From looking at other thermal images in reviews it seems that the other brands are also pretty close to the limits on the vrms. Maybe EVGA is just the first in a series, and they are setting a good example. Let's hope that this won't happen again.

How stupid is it not to QA the temps of vrms and memory, somehow I'm not believing they didn't evaluate those.

November 3, 2016 | 07:19 AM - Posted by BlueHellSKC (not verified)

Started the RMA process and this is the response from EVGA.

If your card is working fine, there's no need to install the pads. We're offering a BIOS update for the cards, that will adjust them to a slightly more aggressive fan curve by default. It's recommended that you try the BIOS update first before requesting the thermal pad mod. The VRM thermal pads are purely optional. We can send you thermal pads to install yourself. To get these, please go to evga.com/thermalmod. The kits will contain detailed instructions on removing the backplates and cooler to install the pads, and thermal paste.

The kerfuffle came from a single review, where reviewers used Furmark. Furmark is a test known to cause damage to cards in the past. The tests were run for 2 hours, or near 8 times longer than would be recommended for that particular software.

Nvidia and AMD have both implemented limits in their drivers to throttle the performance of the cards when they detect Furmark running, due to the damage it causes. It doesn't have any bearing on the performance or durability of the cards in the real world, as it pushes cards to limits that no other application or game would ever do.

Please let us know if you have any further questions or concerns.

Regards,
EVGA

November 3, 2016 | 07:28 AM - Posted by JohnGR

Slightly? From 1600rpm to 2200rpm and from 38 decibels to 47, based on gamersnexus

http://www.gamersnexus.net/news-pc/2666-evga-heat-solution-thermal-imagi...

November 3, 2016 | 07:22 AM - Posted by RWGTROLL

I think it is weird that pcper is so happy to tell everyone that the RX 480 overdraws power from the pcie slot. Did the
480 have fire blowing out of the card? I think this is really fishy. Also not linking the site that you took that picture from is messed up also. I mean this should be a huge deal because parts are blowing up on a 600+ dollar card and everyone is like its ok EVGA you didn't do anything wrong. SO the real question is are you doing damage control for EVGA with same thing with Jayztwocents he seems to be doing damage control also.

November 3, 2016 | 07:39 AM - Posted by Anonymous (not verified)

I kinda have to agree with this... can't really argue against it.

November 3, 2016 | 01:20 PM - Posted by Jeremy Hellstrom

Nor can I.  I am limited to discussing what I actually posted and can't really debate against what the voices in your head tell you we said.

November 3, 2016 | 03:41 PM - Posted by Anonymous (not verified)

I suspect Ryan doesnt want to upset EVGA. Just about every Nvidia release EVGA gives him cards to giveaway.

WHY YOU SAY BAD THINGS ABOUT US, NO CARDS FO U!!!

November 3, 2016 | 10:34 PM - Posted by Anonymous (not verified)

Yeah, you don't bite the hand that feeds you.

November 3, 2016 | 07:49 PM - Posted by JohnGR

No, no, that's not what the voices said.
(probably the only good line/scene in the latest Suicide Squad movie)

November 4, 2016 | 02:49 PM - Posted by Anonymous (not verified)

Your name should be Jeremy Shillstrom,still waiting for that 2-3 pages of test done by you all,like you did to the rx 480 not gonna happen i bet.

November 3, 2016 | 10:12 AM - Posted by Anonymous (not verified)

Lol leave it to the comment section for the mouthbreathing conspiracy theorists to come out of the woodwork. Yes yes, its all a huge conspiracy against AMD and PcPer is in EVGA's pocket. Maybe if you post enough of this crap you too can be a Red Team Plus member!

November 3, 2016 | 02:10 PM - Posted by Anonymous (not verified)

I rather not be, and love me my 1070. But it's really green here I have to say, I don't dislike it, I come here for the green news.

November 3, 2016 | 10:03 PM - Posted by RWGTROLL

I Think that Amd Red Team Plus is a joke no one on that team does anything for the community, let alone that they do a pay it forward build for someone out side the family / friends. I just think it is weird that everyone takes EVGAs mess up so lightly when it is much worse then other company in recent times. I like pc per and think they do awesome reviews it's just they didn't pick this apart like they would have other company's and it is fishy

November 5, 2016 | 07:03 PM - Posted by Anonymous (not verified)

You're projecting again. Pcper didn't say much about the rx480 power draw until they verified for themselves. Do you expect them to fly off the handle without data? Is cautious and factual reporting by biased all of a sudden?

November 5, 2016 | 07:04 PM - Posted by Anonymous (not verified)

At least we can agree that RTP is a cancer... You know where to go to commiserate

November 6, 2016 | 08:13 AM - Posted by Anonymous (not verified)

Gameworks is a cancer, those RTP are just irrelevant.

November 3, 2016 | 09:08 AM - Posted by Baldrick's Trousers (not verified)

Also check out JayzTwoCent's video on thing subject on Youtube:

https://www.youtube.com/watch?v=URyG1OP8p8I

Jay has a close relationship with EVGA and was invited for a visit regarding this very issue.

November 3, 2016 | 09:11 AM - Posted by Baldrick's Trousers (not verified)

Jay also demonstrates how to fit EVGA's thermal pads.

November 3, 2016 | 03:28 PM - Posted by molson

Sent a request to EVGA this morning and below is their response. So... Should I be worried or not?

"I want to clear up any confusion or concerns about the misconception about the thermal pads and the vram chips. Basically, there is no issue with overheating and the benchmark that was used to test the temperature of the video card far exceeds what a regular game can run. Any articles that reported on this issue did not do any actual testing and all the information was speculation at the time. There are now articles online with actual testing of our video cards and I highly recommend looking at the results.

Furthermore, we will be releasing a bio update that lowers the temperature of your video card by 15'c-18'c which will resolve this issue without the need of extra thermal pads but you are more than welcome to request them for your video card. There is no risk of using your video card without the vbios update or the thermal pads. The card that was posted with a burnt vram chip was an isolated incident unrelated to this. Any replacement through the RMA process will come with thermal pads installed.

Here is the new review of our video card >> http://www.gamersnexus.net/news-pc/2666-evga-heat-solution-thermal-imagi...
Here is a video from actual testing of our video cards >> https://www.youtube.com/watch?v=URyG1OP8p8I
Here is the full information about this concern/Request thermal pads >> http://www.evga.com/thermalmod/

Either way, if you prefer to RMA/cross-ship RMA (shipping is paid) or if you have further questions, let me know and I'll be more than happy to help you."

November 4, 2016 | 02:01 PM - Posted by Anonymous (not verified)

I am not shocked at all coming from a brand that uses paper capacitors and the cheapest ct parts they can get their hands on.

November 4, 2016 | 02:01 PM - Posted by Anonymous (not verified)

I am not shocked at all coming from a brand that uses paper capacitors and the cheapest ct parts they can get their hands on.

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

By submitting this form, you accept the Mollom privacy policy.