KitGuru KitGuru.net – Tech News | Hardware News | Hardware Reviews | IOS | Mobile | Gaming | Graphics Cards
AMD's decision to launch its first next-gen 14nm GPU in the mid-range, the RX 480, has invoked an almost instantaneous response from its rival Nvidia. AMD released the RX 480 at the end of June into the crucial sub-$250 segment and not even a month later Nvidia is ready with its direct competition – the GTX 1060.
Unlike the GTX 1080 and GTX 1070 which are both based on GP104 , the GTX 1060 introduces the new Pascal GP106 GPU which, although smaller, shares many traits with GP104 including Simultaneous Multi-Projection, enhanced memory compression and support for new technologies, such as Ansel.
GP106 delivers up to 10 streaming multiprocessors compared to up to 20 on GP104, this means the total GTX 1060 CUDA core count is 1,280 rather than 2,560 (GTX 1080) or 1,920 (GTX 1070). That reduction in CUDA cores brings a respective drop in power consumption as the GTX 1060 has a 120-watt TDP, which is the same as the previous-gen GTX 960 but 30-watts lower than AMD's RX 480.
At $249 the Nvidia GTX 1060 goes head to head with the AMD RX 480 8GB, though AMD also offers a 4GB variation starting at $199. However, it should be noted that, like with the GTX 1080 and GTX 1070, the Founders Edition is more expensive than entry pricing – $299 in the case of the GTX 1060.
The decision to offer 6GB of VRAM puts the GTX 1060, perhaps deliberately so, right between AMD's two RX 480 models. That 6GB frame buffer seems like a sweet spot for covering the rare handful of games or applications that could creep beyond a 4GB frame buffer.
Nvidia hasn't detailed if the GTX 1060 is aimed a specific audience though it seems probable Nvidia has its sights on the mainstream 1080p and ever-growing 1440p markets given the GTX 1060's price point.
4K-leaning Gamers would benefit from the more potent GTX 1070 or higher, though it wasn't that long ago the GTX 980 was aimed towards 4K gamers and Nvidia claims the GTX 1060 offers GTX 980-like levels of performance, make of that what you will. Although, the omission of SLI capability limits the 4K potential of the GTX 1060.
Expect a capable Virtual Reality experience with the GTX 1060 too, Nvidia's own figures claim a 24 % advantage over AMD's RX 480. That being said we'd imagine given the significant expense of VR headsets that early adopters of this technology would lean towards a higher-performance offering, something circa $400+.
| GPU | AMD RX 480 | Nvidia GTX 960 | Nvidia GTX 970 | Nvidia GTX 1060 | Nvidia GTX 1070 | Nvidia GTX 1080 |
| Streaming Multiprocessors / Compute Units |
36 | 8 | 13 | 10 | 15 | 20 |
| GPU Cores | 2304 | 1024 | 1664 | 1280 | 1920 | 2560 |
| Base Clock | 1120 MHz | 1126 MHz | 1050 MHz | 1506 MHz | 1506 MHz | 1607 MHz |
| GPU Boost Clock | 1266 MHz | 1178 MHz | 1178 MHz | 1708 MHz | 1683 MHz | 1733 MHz |
| Total Video memory | 4096 or 8192 MB | 2048 or 4096 MB | 4096 MB | 6144 MB | 8192 MB | 8192 MB |
| Texture Units | 144 | 64 | 104 | 80 | 120 | 160 |
| Texture fill-rate | GT/s | 72 GT/s | 109.2 GT/s | 120.5 GT/s | 180.7 GT/s | 257.1 GT/s |
| Memory Clock | 7000 or 8000 MHz | 7000 MHz | 7000 MHz | 4006 MHz | 4006 MHz | 5005 MHz |
| Memory Bandwidth | 224 or 256 GB/s | 112 GB/s | 224 GB/s | 192 GB/s | 256 GB/s | 320 GB/s |
| Bus Width | 256-bit | 128-bit | 256-bit | 192-bit | 256-bit | 256-bit |
| ROPs | 32 | 32 | 56 | 48 | 64 | 64 |
| Manufacturing Process | 14nm | 28nm | 28nm | 16nm | 16nm | 16nm |
| TDP | 150 W | 120 W | 145 W | 120 W | 150 W | 180 W |
Considering the Asus RGB OC variant costs almost £330, why not include GTX 980 Ti results? (https://www.overclockers.co.uk/asus-geforce-gtx-1060-strix-oc-aura-rgb-6144mb-gddr5-pci-express-graphics-card-gx-40a-as.html and https://www.overclockers.co.uk/evga-geforce-gtx-980ti-superclocked-acx-2.0-6144mb-gddr5-pci-express-graphics-card-06g-p4-4993-kr-gx-288-ea.html)
No SLi support is just petty imo
Price is kind of fair in the current market. It’s more expensive than your standard mid range, but compared to AMD’s 480 or Nvidia’s 980/1070 you get decent value. As long as prices aren’t actually going to be more expensive because of initial shortages.
GTX 1060 SLI supp can destroy AMD… AMD in this moment have only a RX480, and nVidia currently holds almost 80% of the market. Nvidia is realistically able to eject GTX1060 with the same price as RX480 and with SLI supp. That can kill AMD. AMD is currently very economically vulnerable, but USA will never let AMD to bankrupt because they make a very important part in USA total monopolu and total domination in the world with PC (CPU/GPU chips) tehnology…
I get the “why didn’t you include x product in the results” quite a lot, the answer is just I didn’t have one available to test. I agree, the more products the merrier but there’s a limit to how much testing can be done and is feasible.
LOL this article is SO BS, why is it [H] puts the 480 as slightly to much better in a few games, with ~20w power difference, and yet you come here and it shows the 1060 WAY WAY faster then 480, and some 50w or so less power, what is going on here?
[H] is very pro Nvidia in vast majority of cases, but they do test very well, sorry but I simply do not believe this review, either using fine tuned drivers, declocked 480 and overclocked 1060 or something like this, try harder next time,
All the reviews I check out either some are like [H] and quite close to the numbers they give for performance and power usage or like this one(majority are) and show the 1060 are ZOMG amazing compared to 980 or RX 480, seems to me Nvidia is giving cookies out, suggesting specific settings that help them and hurt competition or something like this(wouldnt put it past Nvidia to showcase a max clocked 1060 vs a “bad” 480)
Either way price Nvidia wants to charge for FE vs AIB prices is just whack, as well, less transistors etc leads me to believe there simply is not much left then shaders running insane clocks and likely meaning limited ability to do other more advanced industry standard things, possibly even burning cards out sooner rather then later.
http://www.hardocp.com/article/2016/07/19/nvidia_geforce_gtx_1060_founders_edition_review/1
http://www.polygon.com/2016/7/19/12217078/geforce-gtx-1060-benchmarks
Look at legitreviews. Tweaktown. TpU. Anandtech. Sweclockers. Computerbase.de Hardware.fr
Most reviews have them roughly equal in DX12. Some have one outperforming the other, often RX480 is on top but not always. Look around the web before sprouting bull.
I just spent last 3 hours doing so thank you so much
Your 2 links up against a lot of others saying the 1060 is pretty good. And 1 of the sites you listed even had a fair bit of praise for it.
Dunno, I think Ryan Martin did a fair review. Do you have any more evidence of your claim?
A good Hold-me-over card while we wait for 7nm gpu’s, dx12vulkan adoption and maturity, as well as VR and 4k. I have little doubt this will be the hottest selling card of the year. Well, this or the 3gb model.
AMD’s cards are very closely matched, so it’s gona come down to what features you want or what you think of either company. Myself? heh, I see why 75% of gamers on steam use an NV card, I never should have got this 7870HD back in the day, I just new I’d regret it!
Your forgetting that AMD currently powers 75% of the gaming market with Nintendo and Apple joining the fold soon.
So really Nvidia have more to worry about than AMD as they only have PC desktop GPUs, so with so much hardware running GCN it puts Nvidia into a niche. What incentive is there to optimise for Nvidia ? As it only effects small percentage of overall gaming market.
Sorry to disappoint but the GTX 1060 was stock, the RX 480s were stock, the drivers were the latest versions from both manufacturers (16.7.2 AMD and 368.64). Each website is using a different methodology, a lot of sites it seem have focused on DX12 whereas many of our tests are DX11, perhaps this explains why AMD isn’t as close in our tests as they seem to fare better at DX12 and at higher resolutions. I will take this onboard to include more DX12 tests going forward.
OMG, AMD fan… RX480CF not outperforms 1080, only 1070 in some games… CF and SLI are not good solution for gaming, too much bugs! Best solution for games is single card solution… AMD have 63% for bankruptcy, nVidia only 1%, so in your mind AMD is maybe the best, but in reality and on the market is quite different atmosphere…
http://www.guru3d.com/articles_pages/geforce_gtx_1060_review,30.html
Guru3d has a much larger suite of games and a much more balanced conclusion. I’ve held my tongue as long as I can but Kitgurus game list is way out of date and I’ve mentioned it many times ( this will probably get me banned which will kill me as I’ve been around since DH). Its no good saying we don’t have time to bench more games when everyone else manages it, thank god for Leo Waldock he’s the true hero of Kitguru, that’s a man who gives 110%. If you don’t have that level of commitment you might as well just be another news aggregator with semi, largely unwitty comments.
Must Have – if you have games museum 🙂
Why would you be banned? You dont like our gpu reviews and have asked many times for me to include the witcher 3 and i didnt. Ryan hasnt either. You say games like tomb raider and gta5 arent relevant and we dont agree. Readers ask for these results all the time. Not really much more i can say on this. I have been in hospital for weeks and ryan was kind enough to help out with this one. Its good to hear you like leos work, i help produce his videos and work closely with him on all his content like i do with all my team.
I think in your case its best as you say to use the other websites because we continually let you down here. Might save you posting the same negative comments every review. You are however always welcome around here, you have been around a long time.
Thanks mike i think he did a good job.
SLI means you need the extra PSU overhead to supply the watts for a whole new card, and it never gives more than like a 65% increase. Usually getting a new upgrade every few years (at current rate) is better unless you want to spend more. SLI or crossfire are generally useless for the people who want a quick 1-card upgrade. SLI is usually just used as a gimmick or benchmarking, and there were only a few times when it was worth it in cost versus getting a direct upgrade.
All the reviews I’ve see showed GTX 1060 is faster than 480 except hardocp. Maybe [H] is showing the wrong result.
You folks are welcome.
Rabid fanboys who gota scream conspiracy the moment they hear something even slightly against their holy chip maker is just silly. Keep up the great reviews and don’t let them get you down.
Look at the DX12 mGPU results on Timespy the 480s beat the 1080 by quite a margin. Now Nvidia would love for us to stay in a serial based DX11 ecosystem but sadly it is now end of life.
Multigpu under the new apis work completely different to traditional CF/SLI AFR, so nearly all the issues will be gone.
This also puts full control and optimisation into the devs hands as opposed to the vendor driver teams, which fits in nicely with my original statement above ^^
The future of GPUs is multigpu, you will see a switch from large single dies to multiple smaller dies on a single package, this will bring the cost of gpu production down.
If you look at Explicit linked mGPU under DX12 you we see this allows the GPUs to appear as a single to both dev and end user, the gpus can also pool resources.
The issues we have had in the past is all software.
MS have released a mGPU API for DX12 to aid devs and AMD have released a CF SLI into GPUopen for DX11 so the support is,already being provided.
Nvidias desktop gpu share is always skewed too due to the amount of OEM cards they ship, the high end gpu is still quite small.
So like I stated originally Nvidia have alot more to worry about and is possibly why they are investing heavily into industrial products.
The slow decline of the desktop market is effecting all the companies. Intel have cut 12000 jobs and are finding other avenues to invest in.
The desktop market does not effect AMD as much as it’s only part of their business.
one benchmark does not put the rx480 on 1080 or even 1070s level , you are fanboying here , other wise all MAJOR review sites would have done the same .
Doom developers specifically mentioned they are still working with nvidia to push proper vulkan support for nvidia .
Hi Lelisevis,
I’m always trying to take feedback onboard. Do you have any suggestions for games you would like to see? I am currently in the process of replacing Dirt Rally with Rise of the Tomb Raider (2016) in DX12 mode.
Thanks, Ryan
Nvidia have alot more to worry? Again, AMD have 63% for bankruptcy, nVidia only 1%! So your argument is invalid! DX12 is the future, yes, but not the present! AMD make bulldozer architecture for future (after good Phenom processors), and what happend to them? They came close to bankruptcy… Nvidia plays a much safer game, and I can not blame them for that. That story about how RX480 beat the 1080 talk somewhere else… It’s not even funny!
You can see the results on the 3Dmark hall of fame. The FX range are being replaced by Zen which will put them back on a level playing field with Intel as the issue with X86 advancement is X86 itself.
https://uploads.disquscdn.com/images/df41781174d04eb9919ee7a8bd97177cb89e17a52a7df5087ae89fa93538d495.jpg
I said mGPU not single https://uploads.disquscdn.com/images/df41781174d04eb9919ee7a8bd97177cb89e17a52a7df5087ae89fa93538d495.jpg
LOL, look better on first and last place?!?!??!?!? 😀
I am a graphic designer so, pls, spare me of these tables and charts…
Right just like ashes of the singularity is still working with Nvidia. Nvidia sadly has poor DX12 performance and the problem persists in the 1000 series. All three of you are fanboying! Now stop arguing over nonsense! There are many much more important things happening like the possibility of losing your free speech and other freedoms! Stop crying about graphics cards and wake up!
I thought the review was OK I just disagree with the methodology and conclusion, that’s why you have comments and debate right?, better than deafening silence.
When’s the next Leo video, I swear he’s my long lost brother. Please tell me he’s going to do a video review of Robot Wars.
Last gen I recommended the 970 every single time to friends because it was without comparison within its bracket making it a true MUST HAVE (despite having a 290X vapor-x myself which was a better card but some £150 more expensive therefore making it not a card I could recommend).
The 1060 has strengths and weakness’s when compared to the RX480 and the limited range of games here does not give enough evidence to make a decision, joking aside for one moment where are the DX12 games? Hitman, Tombraider, Warhammer? What about OpenGL Doom?
If you have no doubts at all then its a MUST HAVE, if you are still making your mind up between this and the RX480 then surely its just WORTH CONSIDERING. I’m not going to buy a card based on a chart showing the few watts I’d save which amounts to a couple of cans of Pepsi a month.
This product is a serious compromise with some petty omissions aimed squarely at us the informed gamer, Nvidia know that if they gave us the 1060 we want with SLI capability we’d have bought them by the bucket load, £200 now and £200 6 months or 12 months down the line is a hell of a lot more realistic than £600 now.
Please dont tell Allan he’ll murder me? but the most critically acclaimed gamed ever The Witcher 3 (nothing gets my card screaming for air like that game but the lack of a built in benchmark is a bit of a hindrance, Hitman and definitely Warhammer would flesh out the DX12 tables a little. Doom for OpenGL. I’d leave it at that for now with so many big games on the horizon. Hope that helps a little.
Am I the only one that noticed the VRAM aberration in the GTA 5 Pics? 2064 MB RAM?
I can link 3 or 4 that show on average nope, they are truly comparable, and H is absolutely NOT and AMD fan, they have NOT been for a very long time, so their testing to be wrong, even though they show exact how was, frame time etc, sure it must be wrong, vs most of the others not going anywhere close to as in depth as they did, hmmmmmmmmmmmmmm
Hi Neil,
The screenshots for the game settings used came from the first GPU put on the test bench – the GTX 750 Ti I believe. However, since the settings are consistent irrespective of the GPU used, there’s not much point in taking new screenshots every time since it would only be the “available VRAM” that changes.
Ryan
Debate is always welcome, thanks for taking the time to participate, I appreciate the fact you’re disagreeing with the methodology and resultant conclusion, rather than resorting to insults, conspiracy theories, accusations and the like.
To be honest I have DX12 results coming in courtesy of Rise of the Tomb Raider and the difference doesn’t seem to have changed between either card, it’s similar to the margins shown in other games, the 1060 remains ahead.
I still stand fully behind the conclusion, if you’ve got the money to spend (e.g. a £250-300~ budget) and based on the results of my testing it is the better buy and a must have for a 1080/1440 gamer. I had a look at other reviews like TechPowerUp and they also showed advantages for the 1060 (10%) at the same resolutions we tested (1080+1440). But at £200 I’d go 4GB RX 480 just because the gap in performance is slightly less than the gap in price, though regional pricing is quite variable.
Lack of SLI is a compromise but multi-GPU isn’t that popular, though until steam’s hardware survey shares multi-GPU stats it is difficult to say.
Maurice, I think you have a fundamental misunderstanding of what is “right” and “wrong” when it comes to GPU testing.
Wrong implies something has not been tested properly and incorrect results are being presented, and that if this “wrong thing” were changed the results would magically be different, or correct. I can assure you there’s nothing you could “fix” to get different results based on the benchmarks I’ve run.
You could benchmark in a different way using different methods (e.g. use DX12 not DX11, use a different test platform/CPU, benchmark different resolutions) but that doesn’t make the results presented here wrong. They are right for the chosen methodology.
You are pretty much starting from the statement “the AMD RX 480 should be the same as the GTX 1060 if not faster” and therefore I should continually adjust my benchmarks to achieve this outcome. Which is effectively projecting the result you want to see, onto your methodology to deliver those results.
I think it’s a better approach to allow the data to speak for itself and remove bias. That’s not to say this methodology is perfect, it could be improved and I’m all for continuous improvement and reader feedback. But methodological decisions should be guided by the context of the testing population not the results you expect to see.
Ryan
Fair enough, I kinda figured it was a stock picture.
Thanks, I’ll definitely try and get Hitman into my results going forward. For this review I have now added in Rise of the Tomb Raider results if you’re interested. The results, for me at least, aren’t very surprising. I guess not all DX12 titles respond the same so I will try and add others.
Horrible card. Performance is all over the place. Stuttering galore. Ugh. And the drivers are pure trash.
Can’t hear this guy’s crying over my 1080 Ti sorry