Nvidia GeForce GTX 1060 6GB Review

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.
Excellent card. I'm glad the GTX 1060 lives up to it's performance rumors.

The article was very well done, however I don't understand why the OPs are saying this is a 1080P card? The GTX 980 was designed for 2k gaming, and the gtx 1060 is pretty much equal to a 980.

I personally view the 1060 as a 1440P card, it's got the horsepower and vram to do it at high settings and ultra settings in lower end games.
 


I guess that you can call a card XXXXp gaming, when it can handle all games with maxed out settings at comfortable frame rate.
To be honest, this card can not do that for 1440p, not to mention 1440p@144Hz
 


True. However we have to think of other cards aswell. What about the GTX 750 ti? It is a 1080P card, but can't handle ultra settings in all the games of it's era. So it's a 900P card? I hope you see my point. :)

From what I read, a card is typically rated at a certain resolution if it can maintain an average of 50fps at high settings (not ultra).
 

cmi86

Distinguished
The GTX-1060 seems to be a good 1080/1440 performer, basically on par with the RX-480 in titles that display no bias (2 of the 9 titles in this suite) In titles that display an Nvidia bias (5 of the 9 titles) The GTX-1060 is somewhat faster. In titles that show an AMD bias (2 of the 9 titles) the GTX 1060 is somewhat slower. It is basically impossible to deny the fact that Toms test suite leans pretty strongly to the green side, but it's accurate as most studios tend to optimize towards Nvidia over AMD.

My only real gripe with the test suite selection is "Project Cars" This title is and always has been known to display a completely unrealistically positive performance bias towards Nvidia hardware. How this title has become a staple in a test suite that claims to be "objective" I do not know. However I can say the repetitive inclusion of this title is completely unacceptable.

This repetitive choice rivals Toms inclusion of "Metro Last Light" in the " 7990 vs. GTX-690 The Crowd Picks a Winner" where a group of users were gathered to participate in a blind hands on gaming comparison of the 2 GPU's. At this time Metro LL was known to be a strongly nvidia optimized title that showed very strong negative performance disparities towards AMD hardware( some people couldn't even play it before a patch was released), very similar to the now used "project cars". If the mere inclusion of this title in an opinion based performance analysis isn't enough, The event organizers at TomsHardware knew full well ahead of time that Metro Last Light would not even load on the dual GPU AMD board.. So being fair and objective as they are they obviously removed the title from the test suite to avoid a false negative that would obviously sway public opinion right ? Nope, they kept it and even awarded points to the GTX-690 for for the intentional ace in the hole.

Now a 1 time inclusion of a title like "project cars" may only be a little distasteful and is nowhere near as serious as the obvious stacking of the deck I previously described.. That said the repetitive inclusion of a title with such bias and unrealistic results does begin to rival the blatant and intentional nature of the 7990 v. 690 debacle I previously described.

Me personally, I will be choosing the RX-480. Comparably less expensive AIB boards will be neck and neck with similar GTX 1060 boards in the now and continually faster going forward. Historically AMD hardware has always been more powerful from a raw throughput standpoint. Where Nvidia makes up this ground is by heavily optimizing current gen hardware for current gen games. When an Nvidia hardware/game generation becomes dated and is no longer optimized for, it begins to fall off of the performance curve rather drastically (just look at the GTX-6/7XX family now..)

Compare this to AMD who while often slightly behind at launch due to less optimization, tend to significantly improve hardware/driver optimization globally across their architecture over time. Just look at how the GTX-760 used to strongly beat a 7950 and at times even challenge the flagship 7970. Now it lags behind behind even the 7870/R9-270X and cant touch the 7970/280X/380X that now even beats the faster GTX-960.

My point in all this ? I think the 1060 is a good little card; especially for the TDP, well done Nvidia. That said.. I have said for a long time and will continue to say that Toms has a way of shining nvidia cards in the best possible light at launch and that throughout time the GTX-1060 will slide solidly behind the RX-480 in terms of performance and longevity. Oh and get rid of project cars, trash.
 

IceMyth

Honorable
Dec 15, 2015
571
3
11,165


The thing is the NVidia is dropping the SLI option and seems mainly this one 1060 unless the game is implemented in away to run on 2 GPU's.

"Notice the lack of an SLI connector up top? Nvidia recommends a GeForce GTX 1070 or 1080 to gamers looking for more performance than a 1060 delivers (of course), and does not support SLI on the 1060. "

So you cant SLI 1060, and if you are looking for something will last or upgraded 1070, 1080, RX 480 as they can be SLI/CrossFire.

 


I see your point. And will agree to some extent.
About 750ti. i can call it an entry level for the 1080p.
May be they should add levels to the rating. Like xxxxp@nnMhz - Entry, Good, Excellent.
For example I consider 1070 an excellent 1080p card as it can handle maxed out games with minimum 60FPS :)
Though after having it for a couple of weeks, made me to consider upgrading 1920x1200 monitor to 34" 21:9 1440 monitor with minimum 75Hz.
Now I have to figure out what will be the difference in gaming experience between 75Hz and over 100Hz monitor. There is a big price difference :(
 
Couple of things worth noting ...

-The AIB cards are here for the 1060, not so for the 480.

-First time nVidia's x60 beats an AMD x80

-There is no such thing as a $199 480.... cheapest on PCpartpicker is $270, only 1060 I found was an Asus Strix at $329

-We finally have a current generation reference card that doesn't throttle, the reference 1070, 1080 and 480 all exceed their throttling temps at load @ stock settings

-With 9xx versus 3xx last generation, despite all evidence to the contrary, there was a flood of posters still claiming an imagined need > 4 GB at 1440p / 1080p ? What do we compare with the 1060 6GB, the 480 4 GB or the 8GB ?

-Glaring omission here, no OC results, so we have to go to Techpowerup
GTX 1060 FE - increase in frame rates from Overclocking = 14.0%
Non-refernce GTX 1060 - increase in frame rates from Overclocking = 18.2%
RX 480 - increase in frame rates from Overclocking = 5.0%

-nVidia's decision to drop support for SLI is puzzling
Yes, the 960 in SLI was a failure
Yes, obviously they wanted to curb the 2 x X70 option being as much as 40% faster than their top card.
Yes, with console ports being every more popular, game support for SLI is becoming an after thought for some developers
Yes, SLI performance for the 10xx series is not what it was for the 9xx and previous generations, the question is why ? Well, while two 970s was 40% faster than a 980 for roughly the same cost, two 960s which cost more couldn't catch the 970... same thing here. Two 480s cost more and can't catch the 1070.

Relative Performance @ 1080p in TPU Game Test Suite:
GTX 1060 FE - 97% x 1.14% = 110.6%
Non-refernce GTX 1060 - 100% x 1.151 = 115,1%
RX 480 - 90.0% x 1.050 = 90% x 1.05 = 94.5%

The reference 1060 is 17% faster than the reference 480 in TPUs test suite
The non-reference 1060 is 22% faster than the reference 480 in TPUs test suite

TPU Card Ratings w/ 8GB MSRP:

-$239 AMD 480 Reference - Load temp @ stock settings = 84C / 41 dbA @ load / Rating = 8.9
https://www.techpowerup.com/reviews/AMD/RX_480/29.html

- $249 GTX 1060 Reference - Load temp @ stock settings = 77C / 34 dbA @ load / Rating = 9.5
https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_1060/31.html

- $289 GTX 1060 Non-reference - Load temp @ stock settings = 67C / 28 dbA @ load / Rating = 9.8
https://www.techpowerup.com/reviews/MSI/GTX_1060_Gaming_X/29.html

nVidia for the win with the reference cards .... 17% faster overclocked, half as loud, doesn't throttle. CF support is a plus for the 480, but 2 x 480 cost more than the 1070 which gives the CF option the same failing as the 960 had with the 970 .... it only makes sense when the two lower tier cards are less expensive and / or provide more performance than the higher tier card.

Neither of those happen here. Not to mention the wide power disparity (164 watts per card x 2 for the 480 versus 148 for the 1070)... which means you need and have to pay for a 200 watt bigger PSU and add extra case fan(s) to handle the extra heat

As for the non-reference comparison, too early to comment as AMD hasn't brought anything to the table as yet. Why ? I would guess that AMD, like nVidia was looking to pull a bit more cash in early days when supply was low and they could rake in a bit more cash from the folks who wanna be the 1st on on their block to get the latest thing. Perhaps AIB partners wanted a bit more time to address the power / cooling issue. But unless we see the AIB cards solve the throttling issue and they provide an overclock ability that exceeds 10%, AMD has tough days ahead.

I have to agree with TPUs conclusions ...

https://www.techpowerup.com/reviews/AMD/RX_480/29.html

The weakest point of AMD's reference design is certainly the thermal solution. It doesn't use any heatpipes or other high-tech means to keep the card cool. .... As a result we are seeing temperatures of up to 84°C, which has the card clock down further to keep cool..... What's even worse than the heat is the terrible fan noise. ... in gaming, the fan ramps up a lot, emitting 41 dBA during gaming (not Furmark). This makes the RX 480 the loudest card launched in recent history, much noisier than, for example, the GTX 1080 (which is almost twice as fast). AMD has mentioned to us that the reference design is deliberately weak to leave room for partners to improve on their custom designs. To me, this sounds a bit like "let the partners deal with the problem".

All we managed without causing stability issues was an increase of the GPU clock from 1266 MHz to 1335 MHz, which is a lousy 5%; again, the worst I've seen for years on a reference board. This is further complicated by the fact that the card will often clock down during OC because it a) exceeds the board power limit or b) runs too hot. If you increase the power limit using WattMan, you'll run into the thermal limit quicker. It does seem as though there is a huge spread between GPUs on review samples. I've heard of reviewers who see stock temperatures well below 80°C, while others reach up to 89°C. Assuming AMD selected the best cards for press review, retail cards might even be worse, which means higher temps, more noise, and lower performance.

https://www.techpowerup.com/reviews/MSI/GTX_1060_Gaming_X/29.html

On average, in games, the card runs at a clock speed of 1930 MHz, which is 89 MHz higher than the reference design, resulting in a 3% performance increase, which is less than what we are used to seeing from the reference to custom transition. The underlying reason for that is that the reference design does not suffer from thermal throttling, a first for NVIDIA in many years. It will be interesting to see where other custom designs will end up performance-wise.

We saw temperatures of only 67°C during heavy gaming, and pretty much perfect noise levels. Even with demanding games, it will be difficult to notice the two fans, which are spinning slowly, emitting only 28 dBA, which is quieter than any other card in this performance class that we ever tested. This also qualifies the card for a low-noise media PC, which has enough performance to enable full-details gaming at 1080p, on your big screen. MSI has also included the idle-fan-off feature we love so much since it provides a perfect noise-free experience during desktop work, Internet browsing, and even light gaming.

Either way, MSI's card is awesome and should be on top of your GTX 1060 shopping list. Compared to Radeon RX 480, you get everything improved at the cost of $60, which means 10% worse price-performance, something that I'd go for any day.

And there you have it .... yes, with the $199 MSRP AMD will be able to claim a the price / gaming performance crown. The 7% performance difference between the stock reference 1060 and 480 is easy to give up for a wide enough price gap ... the 22% difference between the 1060 AIB OC'd and the reference 480 OC'd is not. By comparison, the 980 was only 10% faster than the 970 and it costs twice as much.

The 960 had a price performance advantage over the 970, but certainly did not impact 970 sales. When you account for the power, heat, noise and potentially larger PSU and case cooling costs, I'd have to agree with the above quote and say for 22% more OC'd performance, lower noise, no throttling, lower heat .... "most users will go for the price premium any day".

 

Myrmidonas

Distinguished
Mar 13, 2013
128
3
18,715
The reviewers lack 3rd party RX 480 models to test. nVidia drivers will improve GTX 1060 performance that may be 20% to 60% more expensive (right now) but what u give in price you take back in power bill costs, as it seems.

To those that are disappointed with the lack of SLI, I tell you this : DX12 mutli-adapter configurations. We buy a GPU to play future titles not only old/current ones.

Rsxbj
 

reclusiveorc

Distinguished
Oct 27, 2011
55
0
18,660
How come there are no NvENC x265 video encoding benchmarks for any of the 10 series? Need to know if it is worth upgrading from my GTX 970 primary encoding computer or my GTX 950 secondary encoding computer.
 


How 249 vs 259 is 20% price difference ?
and 30watt difference will take 3000 hours to run the cards at full load to compensate 10$. That will take about 2 years of 4 hours per day.
 

FormatC

Distinguished
Apr 4, 2011
981
1
18,990


ID said, that they optimzed the render path ONLY for AMD yet and that the Nvidia implemention will follow later. How we can test a hald-done game without the optimized driver?

Currently asynchronous compute is only supported on AMD GPUs and requires DOOM Vulkan supported drivers to run. We are working with NVIDIA to enable asynchronous compute in Vulkan on NVIDIA GPUs. We hope to have an update soon.

Any question? All current tests are useless ;)

 

FormatC

Distinguished
Apr 4, 2011
981
1
18,990

SLI won't run (no connector) with DX11 but DX12 allows it in some cases. All what we need are working games ;)

 

sixgunbuddyguy

Reputable
Apr 23, 2015
8
0
4,510
I keep seeing sites say that this price range doesn't get a lot of SLI, but I would have thought that's where a lot happens. People that can't afford a 1070/1080, but have $250 now and another $250 in a year or two.
 
I love how people crap their pants over power consumption of the 1060 when they install wimpy transformer coolers into these cards which causes throttling. And this card throttles at around 75C.

Power consumption is blown out of proportion since the primary benefits, less heat output, are overshadowed by budget cooling systems to save the manufacturers more money.

Oh, and let's give it a good 2 months until actual availability. Enjoy your inflated prices!
 
@17Seconds, the problem with your conclusion is it leaves out a few choice facts:
- The RX480 outpaces the 1060 in all the DX12 and Vulkan benchmarks (only ATOS included here but other sites have more, see Hardocp review).

- Power savings here aren't what they were last gen. Getting a 160W GTX 970 vs a similar performing 280W R9 390 was pretty significant (a lot of heat, noise and additional power draw). This gen we are looking at 160W (RX480) vs 120W (1060). That difference is going to make diddly squat difference to a setup- a decent 500W psu is ample for either, and whilst AMD's ref cooler isn't as good as nVidias, the card isn't very loud or especially hot running. AIB cards will improve on this as well.

- Freesync vs gsync is another issue worth considering. An RX480 + freesync screen will be significantly cheaper than a 1060 + gsync monitor. Worth thinking on for those interested in pushing either card to 1440p.

So my summary is: If you mainly play DX11 titles, the GTX 1060 (once cheaper AIB cards are available) is probably the one to get. If you are interested in DX12 / Vulkan games, the RX480 has a pretty consistent lead *and* costs less.
 

esLeg

Commendable
Jul 19, 2016
1
0
1,510
Looks to me like the RX480 actually performs quite well so if you can pick one up at a lower price like 200$ then it would be better overall, especially with Vulkan and the improvements for the future. If you're upgrading the RX480 will perform well on older games too which NVidia seems to excel at more, and it's only like a few FPS difference when talking about heavy DX11 or DX9 games. I would honestly just get an RX480 at this point now that the official results are out.
 

Myrmidonas

Distinguished
Mar 13, 2013
128
3
18,715



Do the math yourself. It is 300$ GTX 1060 Founders and RX 480 250$ (8 GB) , you know. Where does your numbers come from? Apples to apples please. Market prices.....
 

Vogner16

Honorable
Jan 27, 2014
598
1
11,160
This review of the 1060 is exactly what I expected out of NVidia for the 1060. the Hype of it can beat the 980 is false but the price still makes it a valid component.

Still I cant stop myself from getting this impression that in 2 years we will see the Rx 480 beating it in every benchmark... AMD beats NVidia in DX12 and Vulcan almost everytime. it may be 12% faster now, but I'm putting my money with amd and buying an xfx DD RX 480
 

edward_23

Reputable
Jan 5, 2016
10
0
4,510
How about gsynch/freesynch? At $250 it doesn't make sense to buy monitor with $200 gsynch module, and all new monitor from LG and Samsung seem to be equipped with freesynch without noticeable price difference.
 

techy1966

Reputable
Jul 31, 2015
149
3
4,685
"ID said, that they optimzed the render path ONLY for AMD yet and that the Nvidia implemention will follow later. How we can test a hald-done game without the optimized driver?"

Not trying to be a ahole but we test what is in front of us when a product is released just like other sites just did. It does not matter if it paints either product in a bad light or not. It shows the state of the drivers for any given game as well as the performance anyone can expect from a product at release date and most likely farther down the road as well. From the other reviews it clearly shows the 1060 for the most part gets its butt kicked when DX12 is involved when benched against the RX 480 not just in Doom with vulkan. This may change over time as new drivers come out for both parties involved but right now this is the paint set we got and the canvas needs to be painted in such a way it shows us what to expect when we go out & buy a new product today not 2 weeks from now or 2 months but right now.

I don't even like the RX 480 very much I think AMD should have released it with 48 ROP's instead of the weaker 32 ROP's it would have balanced the card out a lot better. It is still a pretty good buy for someone into 1080p gaming as is the GTX 1060 anything higher than 1080p for these cards is a pretty big no go if you do not want to turn the dials down to much to get the needed speed. The 1060 is not a 980 replacement right now I do not think but in the future with driver refinements it probably will get a lot better than it is now. I would tell a friend to get either card right now once the third party ones are here.
 
Status
Not open for further replies.