Here's the link:
 
https://www.reddit.com/r/nvidia/comments/58sf6q/thermal_imaging_of_evga_gtx_1070_ftw_under_load/d96jcxn/
 
Credit goes to megarows, for having the effort of trying to clarify all this.
All in all, temps will never be a issue in games, but if needed you should just adjust the fan curve on Precision XOC and everything will be relatively cool, even on high TDP.
I'm running the aggressive preset atm, since I really don't care about the noise and my case fans are already running at max speed either way. 

Finally someone with some confirmation and testing!!! This should of come from EVGA at the very beginning and all this speculation would be non existent.

shannonjpower
Finally someone with some confirmation and testing!!! This should of come from EVGA at the very beginning and all this speculation would be non existent.

Yep. And it was done 7 days ago. I'm suprised at how nobody posted it yet.

That's cool.
 
Under normal gaming scenario, VRM temps are hotter than other brands, but not a real issue.
But then, overclock is something to consider. With hotter temperatures, less overclock.
 
MY own conclusion is, VRM operates under safe temperature under normal gaming scenario with fan speed on AUTO. But, overclocking on those cards should be really bad due to high temperatures.
I would not recommend cards with ACX 3.0 for overclocking.

Thanks for the post Broken.

Shannon, it's hard for EVGA to make a statement when everyone ignores the statements they make. Yes, it happens. Yes, evga already made a statement saying that the card operate within the designed temp limits. Yes evga already stated (as well as nvidia) that furmark should not be used. Yes, EVGA said "we know the customers want lower temps, so we will offer optional thermal pads to help reduce temperatures further".

The rest is user speculation. As I stated, the design of the heat plate and ACX has been used for years. Only one thing changed, and that was the fan speed to reduce noise. I set my card to run the slave bios (ftw) and fan went from 30% to 50% and temps dropped considerably.

If users want silent system, expect higher thermal output at the same tdp. If users want lower temps, they are required to make a little more noise or invest into some form of liquid cooling.

for the ftw'vrm to run at rated spec (35a each), keep them under 100c is fine.
but what we concern is that the heat from vrm will be spread out the near components and those ir images show clearly that this is a serious issue. 
only vrm is design for 125c operation. that micron vram is only 95c~100c (i forget). and those Multi-Layer Ceramic Capacitors...

Scarlet-Tech
Thanks for the post Broken.

Shannon, it's hard for EVGA to make a statement when everyone ignores the statements they make. Yes, it happens. Yes, evga already made a statement saying that the card operate within the designed temp limits. Yes evga already stated (as well as nvidia) that furmark should not be used. Yes, EVGA said "we know the customers want lower temps, so we will offer optional thermal pads to help reduce temperatures further".

The rest is user speculation. As I stated, the design of the heat plate and ACX has been used for years. Only one thing changed, and that was the fan speed to reduce noise. I set my card to run the slave bios (ftw) and fan went from 30% to 50% and temps dropped considerably.

If users want silent system, expect higher thermal output at the same tdp. If users want lower temps, they are required to make a little more noise or invest into some form of liquid cooling.

we dont know what is the meaning of run in specs. the datasheet of ftw'vrm only stated 35a at 100c. is evga referring this as the specs or do evga simply takes the 125c or even 150c (the shut down temp) as specs. there is not doubt that ftw can operate correctly even vrm temp reaches 110c. i dont think the caught fire problem is really related to vrm itself as the vrm has some protection built in.
 
some ppl buy ftw cards for the card pcb quality, mainly the vrm quality. so if evga is saying ftw can run in specs but the vrm is derated below 35a each than this is not acceptable to me.

1080x1200 is not a very demanding resolution. His Furmark runs are indicative of what someone playing normal games at higher resolutions / high refresh rates might experience during normal gameplay though. I'm regularly hitting the power limit(112%) on some games such as DOOM or Shadow Warrior 2 where I also see the highest core temps.

Scarlet-Tech
Thanks for the post Broken.

Shannon, it's hard for EVGA to make a statement when everyone ignores the statements they make. Yes, it happens. Yes, evga already made a statement saying that the card operate within the designed temp limits. Yes evga already stated (as well as nvidia) that furmark should not be used.

Sorry, but EVGAs statements were very ignorable; little more than EVGA saying all was well.  Don't just say it, and expect everyone to take that as gospel.  Demonstrate that it's true.
 
See how that simple little table posted on reddit did that?  EVGA should have done that.

Scarlet-Tech
Thanks for the post Broken.

Shannon, it's hard for EVGA to make a statement when everyone ignores the statements they make. Yes, it happens. Yes, evga already made a statement saying that the card operate within the designed temp limits. Yes evga already stated (as well as nvidia) that furmark should not be used. Yes, EVGA said "we know the customers want lower temps, so we will offer optional thermal pads to help reduce temperatures further".

The rest is user speculation. As I stated, the design of the heat plate and ACX has been used for years. Only one thing changed, and that was the fan speed to reduce noise. I set my card to run the slave bios (ftw) and fan went from 30% to 50% and temps dropped considerably.

If users want silent system, expect higher thermal output at the same tdp. If users want lower temps, they are required to make a little more noise or invest into some form of liquid cooling.

Making a statement is one thing, providing actual tests results with numbers smack bang in your face is another. EVGA had the opportunity to do these tests and share their results when they had the card on the test bed. Why would you bother running tests on a card WITH the optional thermal pad applied when that itself was not in question or concern, it was always about the pads being optional and if temps were within spec without the pad. EVGA have stated all along there is nothing to worry about but failed to provide any test results or nothing to backup their claims. We had to wait for a consumer to give us what we've all been begging and waiting for.
 
I'm in no way calling anyone a liar here and EVGA has been right all along, but clearly you can see from the continual uproar and responses that this could have been handled much better.