2016/10/19 19:52:10
Scarlet-Tech
Salem13
It looks like (selfish bastard) people using an EK block are OK?

I had a fun time placing that TIM on, the blue backing drove me nuts lol... now I'm so glad it's there!


The ek block will remove the hest sufficiently. Ek has always done well focusing on the vrm section...

On the other hand, they have recently changed to a new thermal tape, and I do not like it specifically because of the protective tape on it.. I like the old stuff way better.
2016/10/19 20:18:30
Greene MaChine
GamerSX
restocking fee in a warranty case? how? anyways i am in the EU so we do not have the concept of a restocking fee here even during the first 14 or 30 days return period, all they did was give me about 80% of what i originally paid because i have used the product for one year, so they deducted that, still a pretty good deal for having the freedom of buying any other product i wanted instead of being stuck with a problematic SKU ... 

 
No a restocking fee from the retailer.  It's virtually common practice in North America.  Alot of smaller businesses will also tack on a %, if your in store buying with a credit card.
It's great that you could return it (the 780).  Over here, you can usually do it,  but alot of places will fight you for it, unless you spend more for a higher end product.
 
 
Good Explanation on the heatsinks Scarlet-Tech.
 
 
libneon
 Ya, I think Furmark results can to some degree be disregarded but the Metro Last Light loop was still hitting 97.7 C which still seems way too high for that part of the card. That was right in the area where my card died as well. It seemed fine for the couple of months I had it before that.

 
Unfortunately, you have to be cautious with alot of software nowadays.
OCCT was another with  functions that can hammer the GPU.  Probably a few others.
There was malware a couple of years ago that would take over your card to mine bitcoin, if i recall correctly.
The original Stalker had a bug where you could set vsync on, and it worked in game, but at the menu it would draw a fast as the card could push.  Dunno if that was fixed..
I used a Usenet client coded in Java from Giganews a few years back, and somehow it interacted and pushed the desktop (2D) at full-speed, for no reason i could see.
 
Use a Frame-limiter when needed, set a Custom Fan Curve in Precision/Afterburner and keep your case clean and the fans going.
 
 
Greene
2016/10/19 20:56:23
HeavyHemi
Scarlet-Tech
libneon
 
Why do the other cards not show such hot spots then?
 
The founder's edition shows no hot spot: http://www.tomshardware.de/nvidia-geforce-gtx-1080-gtx-1070-grafikkarten-roundup,testberichte-242137-10.html
 
EVGA makes great products, I've been buying them for years, but there is definitely a hot spot over the VRM/memory area that seems to exceed the safe range for those components...and I can't even imagine what it's like if someone's case isn't ventilated right or kept clean.

 
 
I do not have a way to measure these hot spots, nor am I an electrical engineer that could figure out why adding twice the amount of VRMs would cause a hot spot.  
 
The SeaHawk (MSI) has a relatively hot VRM section compared to the FounderEdition, and they utilize nearly the same VRM section with a few minor differences. http://www.tomshardware.de/nvidia-geforce-gtx-1080-gtx-1070-grafikkarten-roundup,testberichte-242137-2.html
 
The gigabyte G1 Gaming has less VRM with no full coverage heatplate, and has a hotspot in a completely different section: http://www.tomshardware.de/nvidia-geforce-gtx-1080-gtx-1070-grafikkarten-roundup,testberichte-242137-4.html
 
If I had to go with best guess, the heat plate is not dissipating enough heat due to the low fan speeds and strive for silent operation.  
 
The Galax HOF would be a good example, as the PCB contains extra VRM kind of like the FTW: http://www.tomshardware.de/nvidia-geforce-gtx-1080-gtx-1070-grafikkarten-roundup,testberichte-242137-6.html  Notice that the heat plate has fins to assist with dissipating the heat, and the cooler utilizes more fans.  
 
I also notice that they utilize furmark, which NVidia black listed because it was killing GPUs. I mean, if they are going to torture test a GPU, why not use a program known to overheat and kill GPU's?


I noticed that too. And to get the power draw they are claiming they had to disable the driver throttling for Furmark to get it. That isn't anything you see in the real world except under extreme benching scenarios.
The only thing I used Furmark for was testing different case and GPU fan settings for noise versus worst case temps.
2016/10/20 06:19:51
Arsenic13
Anyone here with a post BS100% RMA GPU have the chops to disassemble the card to see what's changed if anything?
2016/10/20 06:26:47
NetQvist
HeavyHemi
Scarlet-Tech
libneon
 
Why do the other cards not show such hot spots then?
 
The founder's edition shows no hot spot: http://www.tomshardware.de/nvidia-geforce-gtx-1080-gtx-1070-grafikkarten-roundup,testberichte-242137-10.html
 
EVGA makes great products, I've been buying them for years, but there is definitely a hot spot over the VRM/memory area that seems to exceed the safe range for those components...and I can't even imagine what it's like if someone's case isn't ventilated right or kept clean.

 
 
I do not have a way to measure these hot spots, nor am I an electrical engineer that could figure out why adding twice the amount of VRMs would cause a hot spot.  
 
The SeaHawk (MSI) has a relatively hot VRM section compared to the FounderEdition, and they utilize nearly the same VRM section with a few minor differences. http://www.tomshardware.de/nvidia-geforce-gtx-1080-gtx-1070-grafikkarten-roundup,testberichte-242137-2.html
 
The gigabyte G1 Gaming has less VRM with no full coverage heatplate, and has a hotspot in a completely different section: http://www.tomshardware.de/nvidia-geforce-gtx-1080-gtx-1070-grafikkarten-roundup,testberichte-242137-4.html
 
If I had to go with best guess, the heat plate is not dissipating enough heat due to the low fan speeds and strive for silent operation.  
 
The Galax HOF would be a good example, as the PCB contains extra VRM kind of like the FTW: http://www.tomshardware.de/nvidia-geforce-gtx-1080-gtx-1070-grafikkarten-roundup,testberichte-242137-6.html  Notice that the heat plate has fins to assist with dissipating the heat, and the cooler utilizes more fans.  
 
I also notice that they utilize furmark, which NVidia black listed because it was killing GPUs. I mean, if they are going to torture test a GPU, why not use a program known to overheat and kill GPU's?


I noticed that too. And to get the power draw they are claiming they had to disable the driver throttling for Furmark to get it. That isn't anything you see in the real world except under extreme benching scenarios.
The only thing I used Furmark for was testing different case and GPU fan settings for noise versus worst case temps.




The problem is that even the Metro Last Light test is getting too hot on the EVGA if the spec sheet that has been posted is correct. If the max temp according to the makers is 95 degrees and EVGA is at 94,7 degrees when the GPU is only at 73.... well that's bad! My gpu pretty much ran stable at 73 in all games and I've heard people having core temps past 80 when gaming, I don't even want to know how the vram feels then.
2016/10/20 07:24:26
acxcoolerssuck
guru3d also has vrm temp tests on some 1070s including ftw. same conclusion too.
 
 
one of the main reasons why nvidia has blacklisted furmark should be the poor cooling performance of nvidia stock coolers.
 
2016/10/20 07:44:15
NetQvist
acxcoolerssuck
guru3d also has vrm temp tests on some 1070s including ftw. same conclusion too.
 
 
one of the main reasons why nvidia has blacklisted furmark should be the poor cooling performance of nvidia stock cooler.
 




Ugh thankfully my reseller just sent me a message that I'll get a Asus Strix 1080 OC as a replacement instead of the EVGA, sent my card in on Monday before these thermal images and now I'm really glad I won't have to try another FTW.
 
The cards are literally fire hazards if you happen to get a bad batch, bad components or whatever the source of this problem is.
2016/10/20 08:20:30
mannitu78
interesting to hear that so many people have problems with the FTW. For me it would be ok, because i always build a Raijintek Morpheus on my Cards, with far superior memory and VRM cooling. So i would probably never run into problems.  I guess that would be the best way to handle it. Or just limit the Power % to a certain point. 2100mhz does not tell anything. It can pull 160Watts or 200+ watts on 2100mhz. Especially high resolution pulls a lot of power.
 
I have to say i would be so excited about a 1080 Classy-thermal analysis...couldnt find anything so long. On the 980 it has to be superior, cause you can even ov the VRAM to 1,7+ Volts on the VRAM on the long term, even with stock cooler. The PCB is just awesome. It would be very disappointing if the mighty Classy is not as good as it used to be anymore. But i guess the Classified is still fine...but its got nearly the same cooling plate for VRM and memory....which seems to be very ineffective.
 
 
btw I HAVE TO SAY something the wrote in the test which you cannot understand, as its german.
 
"You can bring the problem down to a certain point, if you put the fan on very high rpm, but thats only achievable with an annoying level Fan-Sound"...
sounds like dont get 106 C on the PCB if you turn the fan well above the stock profile.
2016/10/20 08:22:18
jompija
Quick test. 1070 FTW @ idle

 
Full load for 10 minutes, GPU temp @ ~65c

 

 
Observations?
2016/10/20 08:35:20
jaju123
jompija
Quick test. 1070 FTW @ idle

 
Full load for 10 minutes, GPU temp @ ~65c

 

 
Observations?


Well 56 should be fine right.

Use My Existing Forum Account

Use My Social Media Account