• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

EVGA Pascal cards hotspot problem

Associate
Joined
12 Aug 2016
Posts
6
My Hyrbid just went 100% fan, then went black. Did a hard restart then the GPU is still running, (LED's and whatnot) but not putting out an image. My iGPU is fine but can't even get a POST screen from the Hybrid.

I'm gathering I'll be needing to do an RMA? There wasn't any sparks or burning. It was just idling on Windows desktop.
 
Soldato
Joined
7 Aug 2012
Posts
2,640
This worries me. I got my 1080 SC in July I think and as yet no problems, ut given I have no way of monitoring temps of that part of card, it could be basically a ticking timebomb.

My 1 small glimmer of hope is the fact I got a 1080 Hybrid kit from them and have since replaced the ACX cooler with that, so hoping to hear that this unit does not have the same issue. I know the proper Hybrid 1080 appears to, but that uses a different pcb (FTW) and a different frontplate/ fan assembly so is a different cooler.
 
Soldato
Joined
31 Oct 2002
Posts
9,851
Shocking - I'd heard that EVGA were the best NVIDIA partner out there, now it seems they are one of the worst, judging by that Reddit thread.
 
Associate
Joined
26 Aug 2016
Posts
561
This worries me too. Recently bought a second 1080 FTW, so now I have two of these potential fire hazards in my case.

I hope EVGA's warranty is going to cover it if these things blow. I'd feel happier if they just did a recall.
 
Soldato
Joined
26 May 2014
Posts
2,944
I have been having issues with my computer resetting on boot since installing the 1070. Wondering if this could be related?
Nah, as the card would need to be under heavy load for an extended period to warm up to the point where this could be an issue, which won't happen during startup.

These hot spots are nothing new either. Plenty of cards from all different manufacturers have had them over the years. Here's a Titan X for example:

01-pcb_w_600ays4s.jpg


Of course, you may expect better from a custom cooler, especially on a card that draws a lot less power than a Titan X, but I'm sceptical as to whether it's actually causing any issues.
 
Caporegime
Joined
20 Jan 2005
Posts
45,621
Location
Co Durham
Nah, as the card would need to be under heavy load for an extended period to warm up to the point where this could be an issue, which won't happen during startup.

These hot spots are nothing new either. Plenty of cards from all different manufacturers have had them over the years. Here's a Titan X for example:

01-pcb_w_600ays4s.jpg


Of course, you may expect better from a custom cooler, especially on a card that draws a lot less power than a Titan X, but I'm sceptical as to whether it's actually causing any issues.

Could it be different memory chips back then that could handle higher heat?

On pascal cards the ram can only handle 95C
 
Associate
Joined
13 Oct 2009
Posts
778
This reminds me of my Inno3D GTX 980 Ti that failed a week after I sold it. I had used a water block on it myself, but with the regular cooler mounted the VRM temps rose to over 110c; it was a design flaw that Inno3D fixed in newer revisions as it lacked heatpads for the VRMs, which caused the VRMs to overheat and fail.

http://forums.guru3d.com/showthread.php?t=403905
 
Caporegime
Joined
18 Oct 2002
Posts
33,188
Could it be different memory chips back then that could handle higher heat?

On pascal cards the ram can only handle 95C

Yes, temps aren't absolute, just because one card is fine at 100C doesn't mean another is. If a chip is designed to run at 70c then 75C is too much, if a chip is designed to run at 145C, then 120C is fine.
 
Soldato
Joined
26 May 2014
Posts
2,944
Could it be different memory chips back then that could handle higher heat?

On pascal cards the ram can only handle 95C
Pascal cards use a variety of different memory chips, incorporating both GDDR5 and GDDR5X as well as both Micron and Samsung, and even different specific parts within all those parameters. For example, the Samsung GDDR5 on Pascal cards incorporates at least the parts K4G80325FB and K4G41325FE, both of which are rated for 8GHz. Unfortunately, Samsung don't seem to publish operating temperatures for their GDDR5.

However, the Titan X shown above used Hynix memory, specifically H5GQ4H24MFR-R2C, which is only rated for 85°C, even lower than the Micron GDDR5X.
 
Soldato
Joined
7 Aug 2012
Posts
2,640
It is not the VRAM that is the issue here guys, it is the VRM's - the voltage regulator modules.

It is confusing in that both acronym's are similar. But the issue is to do with the row of components towards rear of card which, as the name suggests, regulate the voltage to various other parts of the card.

EDIT: actually closer to centre of card than rear, and some of this heats is dissipated towards some the bank of vram chips that are closest to the VRM's, so could actually be having the effect of running those VRAM modules out of spec.
 
Last edited:
Soldato
Joined
26 May 2014
Posts
2,944
It is not the VRAM that is the issue here guys, it is the VRM's - the voltage regulator modules.
The VRMs would never be an issue for temperatures in the 90s though, because they're always rated much higher than that (usually around 120-125°C). The implication here is that heat generated by the VRMs is spreading to the VRAM and affecting its operation.
 
Soldato
Joined
16 Aug 2009
Posts
7,728
According to that article EVGA state cards built before August 31st have this issue, those after do not. My 1080 FTW was ordered on... August 4th. Oh joy.

I havn't had any black screen issues though its coil whine drives me nuts. Is it just me or does it not have any temp sensors on the VRM's? GPU z nor HWMonitor doesn't report anything but cards I've had in the past certainly did. Hmm.
 
Soldato
Joined
7 Aug 2012
Posts
2,640
The VRMs would never be an issue for temperatures in the 90s though, because they're always rated much higher than that (usually around 120-125°C). The implication here is that heat generated by the VRMs is spreading to the VRAM and affecting its operation.

I see. Yeah, I did see that the heat was being dissipated in all directions (including towards a bank of VRAM chips) instead of dissipated in the direction of the lump of a cooler and heat-pipes.

I have checked my ACX from 1080 and it certainly looks to exhibit same problem as indicated in that reddit - no piece running horizontally across the fins as the Asus card has.

I do not know if this has been rectified on my Hybrid and I don't really want to take shroud etc. to inspect as tbh it was pain in the ass to fit and get all cables managed properly and lying flat enough to get theh shroud seated and screwed on. I guess I will just play the waiting game and see if EVGA issue a notice of recall, or sit tight until problems start to show.
 
Soldato
Joined
7 Aug 2012
Posts
2,640
According to that article EVGA state cards built before August 31st have this issue, those after do not. My 1080 FTW was ordered on... August 4th. Oh joy.

I havn't had any black screen issues though its coil whine drives me nuts. Is it just me or does it not have any temp sensors on the VRM's? GPU z nor HWMonitor doesn't report anything but cards I've had in the past certainly did. Hmm.

Only card I have had in recent memory that could monitor VRM temps was a 7970. I am pretty sure that all my recent Nvidia cards (780Ti's, 970, Titan X and 1080) do not give this temp reading.
 
Caporegime
Joined
20 Jan 2005
Posts
45,621
Location
Co Durham
Only card I have had in recent memory that could monitor VRM temps was a 7970. I am pretty sure that all my recent Nvidia cards (780Ti's, 970, Titan X and 1080) do not give this temp reading.

Shame, even my last 290x had vrm temps and they could get very toasty on a 290x!!!!!!!!
 
Back
Top Bottom