EVGA responds to hot VRM area on GTX 10 series

[zMeul]

sources:

• http://forums.evga.com/FindPost/2567129
• http://www.tomshardware.de/nvidia-geforce-gtx-1080-gtx-1070-grafikkarten-roundup,testberichte-242137-2.html

here you can request the thermal pads: http://www.evga.com/thermalmod/

 

 

the problem: the GTX1070 and apparently GTX1080 custom PCB designs from EVGA have their VRM area very hot - under heavy load

 

 

the VRM can opperate at hi temperatures, the VRAM does not - according to Micron GDDR5X spec sheet, the chips can safely operate between 0°C to +95°C

there are couple of people reporting card getting damaged with "burns" in the VRM area, while others report black screen with fans revving up to 100%

 

EVGA investigated and issued a response to the problem:

Quote

“The test used in the referenced review from Toms Hardware (Germany) is running under Furmark, an extreme usage case, as most overclockers know. We believe this is a good approach to have some idea about the graphics card limit, and the thermal performance under the worst case scenario. EVGA has performed a similar qualification test during the design process, at a higher ambient temperature (30C in chamber) with a thermal coupler probe directly contacting the key components and after the Toms Hardware (Germany) review, we have retested this again. The results in both tests show the temperature of PWM and memory is within the spec tolerance under the same stress test, and is working as originally designed with no issues. 
 
With this being said, EVGA understands that lower temperatures are preferred by reviewers and customers.
 
During our recent testing, we have applied additional thermal pads between the backplate and the PCB and between the baseplate and the heatsink fins, with the results shown below. We will offer these optional thermal pads free of charge to EVGA owners who want to have a lower temperature. These thermal pads will be ready soon; and customers can request them on Monday, October 24th, 2016.  Also, we will work with Toms Hardware to do a retest.”
 
Thanks,
EVGA
 
EVGA GeForce GTX 1080 FTW running Furmark with thermal pad mod (30C Ambient in Chamber) – October 21st, 2016

 

so, EVGA will be issuing, free of charge, two thermal pads to be applied to the VRM area between the baseplate and the heatsink, and between the back on PCB and the backplate

 

---

 

the problem with EVGA's designs is that they don't have active cooling for neither VRM nor VRAM:

 

 

everything (except for the GPU) is covered by a (base) plate

 

---

 

update Nov 02: http://www.evga.com/thermalmod/

 

EVGA has done some further investigation and decided that all the cards with ACX 3.0 cooler will be issued vBIOS updates to adjust the fan curve

apparently EVGA prioritized acoustics over thermals

full EVGA statement:

Quote

Recently, it was reported from several sources, that the EVGA GeForce GTX 1080 FTW PWM and memory temperature is running warmer than expected during Furmark (an extreme stress utility).

 

EVGA has investigated these reports and after extensive testing, below are our findings:

• On ACX 3.0, EVGA focused on GPU temperature and the lowest acoustic levels possible. Running Furmark, the GPU is around 70C +/- and the fan speed is running approximately 30% duty cycle or lower.
• However, during recent testing, the thermal temperature of the PWM and memory, in extreme circumstances, was marginally within spec and needed to be addressed.

Conclusion: EVGA offers full warranty support on its products, with cross-ship RMA*, and stands behind its products and commitment to our customers.

 

To resolve this, EVGA will be offering a VBIOS update, which adjusts the fan-speed curve to ensure sufficient cooling of all components across all operating temperatures. This VBIOS will be released in the next few days and users can download it and update their cards directly. This update resolves the potential thermal issues that have been reported, and ensures the card maintains safe operating temperatures.

 

For those users who want additional cooling beyond the VBIOS update, EVGA has optional thermal pads available. This update is not required, however; EVGA will make it available free of charge to any customer who is interested. To request the thermal pad kit, please visit www.evga.com/thermalmod.

 

Any customer who is not comfortable performing the recommended VBIOS update, may request a warranty cross-shipment* to exchange the product to EVGA for an updated replacement.

 

All graphics cards shipped from EVGA after 11/1/2016 will have the VBIOS update applied.

 

cards affected:

 

cards not affected: Founders Edition, Blower type, HYBRID, and HYDRO COPPER

 

----

 

update  05 Nov: http://forums.evga.com/gaps-between-vram-chips-and-the-stock-thermal-pads-on-the-midplate-m2570619-p3.aspx#2576924

EVGA will add VRAM thermal pads and thermal paste in the VRM package 

Quote

Starting next week we will ship thermal grease, memory thermal pads, along with PWM thermal pads in the package. It is recommended to remove the existing grease on GPU and memory pads and apply the new ones. For any customers that did not receive memory thermal pads, please contact us so we can arrange it.

 

Edited November 5, 2016 by zMeul

[potoooooooo]

Why do people like EVGA so much? There is shit like this every generation

[AresKrieger]

Just now, potoooooooo said:

Why do people like EVGA so much? There is shit like this every generation

Warranty support is really good and they do free upgrades if they no longer have replacement stock, also this is the first time I've heard of it.

[226477_1454181668]

Just now, potoooooooo said:

Why do people like EVGA so much? There is shit like this every generation

Any evidence, or just a general statement of "fact".

[CptLameJokes]

So is just the FTW series right? So the normal 1080/70s aren't affected?

[Mr.Meerkat]

1 minute ago, CptLameJokes said:

So is just the FTW series right? So the normal 1080/70s aren't affected?

Quote

the GTX1070 and apparently GTX1080 custom PCB designs from EVGA have their VRM area very hot - under heavy load

 

[zMeul]

9 minutes ago, CptLameJokes said:

So is just the FTW series right? So the normal 1080/70s aren't affected?

if I recall, the FEs have factory applied thermal pads on the VRM chips

 

here's a GTX1080:

Edited October 21, 2016 by zMeul

[Mr.Meerkat]

Hold on, there's a big difference between THW test and EVGA's...I highly doubt it made a 20C+ difference but it should still make a difference, THW didn't have a backplate while EVGA's did...

[zMeul]

3 minutes ago, Mr.Meerkat said:

Hold on, there's a big difference between THW test and EVGA's...I highly doubt it made a 20C+ difference but it should still make a difference, THW didn't have a backplate while EVGA's did...

EVGA put the backplate on because they actively use it as a heatsink

and yes, a 20deg diff can be achieved with proper heat management (applying thermal pads)

[Rohime]

16 minutes ago, potoooooooo said:

Why do people like EVGA so much? There is shit like this every generation

Hey... they listened... they retested ... they offered solution for those that want it ... they work with the people that reported it.

 

Whats not to love?    Many many many companies don't do anywhere near as much.

[Mr.Meerkat]

1 minute ago, zMeul said:

a 20deg diff can be achieved with proper heat management (applying thermal pads)

I was talking about just the backplate...but the point still remains, it's not exactly an apples to apples comparison plus you know...the backplate might be able to radiate heat away from the heat sensor  

 

Ah well, it's still good to see them to actively tackle an issue with their cards  

 

On a side note, I know how much good VRM cooling can help. The 290X that was cooled by the AIO at the very start didn't have the gelid VRM heatsinks so they were running up to 88C~ underload stock clocks and slapping on the gelid HS resulted in 60C stock and 75C with my 1200Mhz OC  

[Delicieuxz]

17 minutes ago, CptLameJokes said:

So is just the FTW series right? So the normal 1080/70s aren't affected?

 

14 minutes ago, Mr.Meerkat said:

Quote

the GTX1070 and apparently GTX1080 custom PCB designs from EVGA have their VRM area very hot - under heavy load

 

 

So, FTW and Classified? I think the ACX3 and SC use the same PCB as the FE.

[Mr.Meerkat]

Just now, Delicieuxz said:

I think the ACX3 and SC use the same PCB as the FE.

Thinking about it, I wouldn't be suprised if it was all ACX/custom cooled 1070/1080 but at the same time, they may have already applied the HS on em...

[TorqueS]

While I agree these kind of "errors" in manufacturing should not exist, EVGA is always prompt and opened to suggestion so that the problem gets solved as fast as possible. Kudos to them for that.

[Dabombinable]

This is actually similar to my GTX 970 G1 Gaming with its vRAM. Although I removed the backplate on mine, the vRAM runs a lot cooler and more stable if it has thermal pads installed between the rear chips and the backplate.

[RagnarokDel]

I'm sure none of them have their power limit to +50%.

 

Like what do you expect?

[Misanthrope]

1 hour ago, potoooooooo said:

Why do people like EVGA so much? There is shit like this every generation

This very gen by MSI If I'm not mistaken

[Nowak]

1 hour ago, Delicieuxz said:

So, FTW and Classified? I think the ACX3 and SC use the same PCB as the FE.

Yes, the regular and SC models use the same PCB as the FE.

[Techicolors]

doesn't seem like the SC model is that affected. here's guru3d's thermal imaging 

 

 

i certainly haven't run into any issues (mild overclock for a few months) 

[zMeul]

1 minute ago, Technicolors said:

doesn't seem like the SC model is that affected. here's guru3d's thermal imaging 

 

 

i certainly haven't run into any issues (mild overclock for a few months) 

add 10deg to everything - EVGA and Tom's did the testing at 30deg room temp

 

that M3 hottest point will be 91.3

[Delicieuxz]

3 hours ago, zMeul said:

add 10deg to everything - EVGA and Tom's did the testing at 30deg room temp

 

that M3 hottest point will be 91.3

It looks like the room temperature was 22 C.

 

"... man bedenke, dass diese Werte im klimatisierten Raum bei 22°C Umgebungstemperatur zustande kamen."

 

 

 

[Sintezza]

And what is concered as "hot vrm".

Which temperatures are we talking about at which parts?

What are the operating temperatures of the powerstages exactly?

 

[zMeul]

Just now, Sintezza said:

And what is concered as "hot vrm".

Which temperatures are we talking about at which parts?

not enough!?

[Sintezza]

4 minutes ago, zMeul said:

not enough!?

 

108°C on what?

On every single powerstage? or all powerstages combined?

If this card has IR powestages, (which it most likely does) then each stage can operate on a safe temp upto 125°C

Technically they should measure the temperature of each individual powerstage, to determine is there is any chance of serious damage.

 

 

[zMeul]

2 minutes ago, Sintezza said:

108°C on what?

On every single powerstage? or all powerstages combined?

If this card has IR powestages, (which it most likely does) then each stage can operate on a safe temp upto 125°C

do you even read ?!?!?!

 

1st: the 106.8deg is at the measuring point highlighted on the image

2ndly, the VRM isn't the problem it's the source of the problem - the actual problem is the VRAM that gets too hot from being in close proximity to the VRM