Nvidia GeForce GTX 1000 Series (Pascal) MegaThread: FAQ and Resources

Page 113 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Perhaps is websites would actually look at VRM temps unlike the less important GPU temps companies like EVGA would actually care more about VRM cooling. But all websites only look at the GPU temps, until that Toms article where the RX 470 Strix had overheating memory modules which was later edited out of the article and all related things wiped out of history and existence, such as comments related to the matter.

Yes, I will never forget that.
 
I didn't see that and no moderator ever deleted any comments because editorial asked. Not unless the comments were abusive. That's the kind of thing that makes Community ( mods + Kevin + Joe etc. ) angry. I'm not saying it didn't happen but it didn't happen due to the mod team.

A retracted mistake perhaps?
 
It 100℅ happened and I won't back. I don't believe the mod force was involved it had to have been from higher up upon the request of Asus. Math Geek remembers it, too. I 100℅ remember it.

There is even a post from RedJaron who writes articles saying "I won't recommend the Strix cards this generation due to their poor VRM and memory cooling".
 
boy this EVGA vrm issue sure is gonna need a lot of explaining to keep the haters away. it is ONLY ON THE FTW EDITION OF THE 1070 AND 1080 CARDS that a problem exists. the vrms are not directly cooled and have no thermal pads which was noted by a number of reviews i read. under certain conditions, a vrm may fail do to overheating. the video with it actually flaming up is not a good example since the card had already failed and he started up it again multiple times after it failed. the integrity of the board was already compromised and when multiple times power was supplied, something shorted and gave a nice little light show.

so again, only the 1070 and 1080 FTW edition of cards is effected and it takes a poor airflow case with high internal temps, long term near max gpu loads and a smiting by the pc gods for the vrms to even fail. until tom's exposed the issue, there have been no reports of cards failing. but now that the issue is known, sites are stressing the crap out of the cards to see what happens. there were no reports of the cards failing from normal gaming sessions EVER that i have seen reported anywhere. if you have one, request the thermal pads or return the card for a replacement and problem solved. obviously a poor decision by EVGA how they laid out the cooling but they'll fix it and eat all the costs they encounter without batting an eye.

give it a couple months and new FTW cards will come already updated with thermal pads and this will drift out of everyone's mind until the next time amd has a problem and folks bring this back up again to defend amd (as if that somehow makes the next issue "ok" or something)
 
That's good news for me. I have an EVGA GTX1070 Gaming SC Black arriving tomorrow and I wasn't looking forward to the hassle of installing pads. Now If I just got lucky and got the Samsung memory instead of the Micron memory, all will be right with the world.
 
that fix is a simple bios update anyway. so quick and easy and something i tend to do anyway when getting new hardware. a bios update is not always needed but rarely hurts since you never know how long the part has been on a shelf and how many updates there may be for it. gpu's and mobo's in general i always update bios on when putting them in a system.


overall, this nvidia generation seems like it had a lot of major hickups. mostly a paper launch, crazy high price gauging and the memory issues on the 1070 cards. throw in a couple specific brand issues and seems like they were more interested in the hype rather than delivering a new awesome gaming experience.

wonder what is the next issue we'll see for them? how long did it take to identify the .5 gb missing ram from the gtx 970 cards??
 
also wanted to note that on the EVGA thermal pad site http://www.evga.com/thermalmod/ there is a list of cards eligible for the free pads.

these include some 1060 6gb and 3gb models if you wish to add the pads and most of the 1070 and 1080 models. there also appears to be a bios update that ups the rpm of the fans to help them cool down. i'd get that anyway unless you already have a custom curve in place to compensate.
 


Nothing so far, but mine is technically a reference design and not the FTW.

I also have to say I haven't used my card much, been very busy lately and haven't had time to run anything more time consuming than the occasional round of Unreal Tournament.

Still haven't re-installed my water cooling loop. Though absolutely no complaints with the ACX 3.0 cooler.
 
Going on a slightly different topic...

What do you all think about Pascal overclocking? Is it even worth it?

I've been running some calculations and I've realized that the biggest problem with Pascal is that while you get an insanely higher frequency clock than any other GPU around right now, you can't overclock by the same percentage.

So with a 2000mhz stock boosting frequency, you can still only get 120mhz OC on the core clock which comes out as only an 8% performance increase compared to cards with lower frequencies that are in the 20% range.

I personally have gotten to the point that I'm not going to overclock my pascal 1060 because the extra performance is too small.
 
if you look at it as out of the box clock speeds then the % is much higher :)

1600 or so mhz oc'ed to 2100 is a rather large number. 31%ish

of course we know that boost 3.0 makes this number nonsense as every card boosts to 2 ghz or so anyway. reviews and such show such a small increase in fps, that it is jut not worth it to me to do it. a lot of extra power draw plus possible lessening of lifespan for a 1-2 fps increase just seems not worth it to me.
 


Yeah that's what I thought. I'm talking with GPU boost 3.0 doing all the work. :)
 


interesting. i noted they were handing out thermal pads for some 1060 models as well (i assume ftw models but did not look up part numbers). i wonder if the whole ACX 3.0 turns out to be a bust. can't say i have seen something like that before where a whole cooling set-up is faulty for all models. kind of odd since the 2.0 was good enough for the much higher powered 900 series cards. how they managed to take lower powered cards and update the cooling for less results is a question to ask the engineers over there. (i'm sure they will point at marketing like they always do)
 
The ACX cooler is great. I had a GTX 780 ACX and now I've got a 980 Ti ACX. They just missed the VRM this time around.
 
i think it is good to go as well but just noting that the older models you speak of are ACX 2.0 while this is a new 3.0 version of the cooler. gamer's nexus did some testing with the thermal pads and new bios fan profile EVGA released. dropped the temps a lot http://www.gamersnexus.net/news-pc/2666-evga-heat-solution-thermal-imaging-of-vrm-1080-ftw down from 109 degrees to 85 degrees with thermal pads and 81 degrees with pads and higher fan speeds.

so clearly the cooler is adequate as 81-85 degrees is not that bad for vrm's from reviews of other cards, it is rather average.
 
because that card is based on the reference pcb which only has a 5 vrm's on it vs 12 for the ftw card. the ftw cards are very over-engineered and have a ton of power phases and other pieces. they also take a lot of extra power. 2 x 8-pin power vs a single 8-pin for the lesser models.

all this extra stuff creates a ton more heat for the FTW card. without the thermal pads helping get the heat away from the vrm's it can allow the temps to get way out of hand.

i am speaking of the 1070/80 models here. the 1060 adds some phases on the ftw but not near as many since they really are not needed. so your sc card uses less power and has less heat producing parts so the 2.0 cooler is more than enough for it.
 
That's great Math Geek. Thanks for the prompt and informative answer. It's the single fan version if anyone is interested in such details and want peace of mind.
 


Lately I've kept my 1080 @ stock because i get close to 1800 mhz on a regular basis and the PC is very quiet...
Overclocking the card to 2000 can yield approximately 5-10 fps+ more at the cost of noise...

Titan X Pascal will benefit more from Overclocking
 
I still dont understand why 1 1080 will overclock higher than 2 will in sli. This is the first generation i've seen this occur. With a sli disabled i can get to 2156mhz but with 2 enabled the highest i can get is 2088mhz but it usually settles out at 2050mhz after a bit of gaming.
 
perhaps some thermal issues (well not really an issue but simply higher temps in sli) are coming into play? 2 cards so close together produce more heat. maybe the few extra degrees of heat is keeping the cards from boosting as high. the reality of the 100 mhz more is 1 fps if you're lucky so not exactly something to worry about. at least for a single card that extra 100 mhx is not even noticable so i'd assume it makes little difference in sli as well.
 


both card can get that high? also when fully loaded was is temp for both card?
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom