Geforce GTX 1060 Thread: faster than RX 480, 120W, $249

[esquared]

There are several of you that are continuing to attack each other and at the same time fill up the reported post box with your complaints. I am still going through them now. It's time consuming and most of these complaints have no merit.

21/29 reported posts on my first page are coming from VC&G. That's 72.4% OF ALL reported posts coming from one of the 36 subforums. You're acting like pre-teen boys with this green team vs red team mentality and name calling. Discuss to the topic like adults.

Be fore-warned that more of you will be sent on 30 day removals from VC&G if you continue to post as you have in this thread.. You all know who you are so I suggest you all dial it back a bit.


esquared
Anandtech Forum Director

 

[Cookie Monster]

Not really true.

In summary, the RX 480 is doing very well in this game, it even beats the 1060 and gets very close to the 980 Ti, even though it has no extra DX12 features. Some other cards fare not so well, but still very close to the average (meaning the performance summary at 1080p). Fury X is basically where it should be. It's actually slightly better as the 980 Ti is less than 10% faster. Fury is 0.4% slower than 980 when on average it is merely 1.8% faster, so a 2% or so delta. 390 is normally 1% faster than 970, here it is 0.8% slower, so basically the same delta as the Fury. If what you say is true, it looks like the advantage Maxwell gains is about 2%.

All in all, it looks like the performance is about equal to the average at 1080p, with a special advantage to the 480 (relative to all other cards), and a special disadvantage to 1060 (relative to 980).

It depends on what benchmarks you look at. TPU in this case shows more even numbers between the 2 IHVs with ever so slight advantage to nVIDIA. But as a whole, all reviews will show that nVIDIA is better in this title (maxwell and pascal) and it could be attributed to those features.

e.g. https://www.computerbase.de/2015-12...abschnitt_benchmarks_von_full_hd_bis_ultra_hd

http://www.pcgameshardware.de/Just-Cause-3-Spiel-9784/Specials/Test-Benchmarks-1179397/

You can see here that nVIDIA cards (maxwell) do well overall compared to TPU numbers.

 

[ELopes580]

249 dollars it is mentioned in the OP. At my usual supplier: alternate.nl , prices range here in the Netherlands from 309 Euro to 389 Euro for the GTX 1060 custom boards. That is ~345 USD to ~434 USD.

The 309 Euro model has a 6 pin connector and is from Asus :
ASUS TURBO-GTX1060-6G
It has DVI, 2x HDMI, 2x Displayport.

But almost all GTX1060 cards are out of stock except one. And no delivery date available.

I order this exact card last week and it arrived on Thursday. Great card, and runs the few games I've tried on full max settings. Temps so far are in the low-40C on idle; and IIRC, mid-60C on load. Haven't bothered to overclock it just yet. I haven't had a desktop/gaming PC since 2008. Wow have I ever missed out on a lot!

 

[IllogicalGlory]

It depends on what benchmarks you look at. TPU in this case shows more even numbers between the 2 IHVs with ever so slight advantage to nVIDIA. But as a whole, all reviews will show that nVIDIA is better in this title (maxwell and pascal) and it could be attributed to those features.

e.g. https://www.computerbase.de/2015-12...abschnitt_benchmarks_von_full_hd_bis_ultra_hd

http://www.pcgameshardware.de/Just-Cause-3-Spiel-9784/Specials/Test-Benchmarks-1179397/

You can see here that nVIDIA cards (maxwell) do well overall compared to TPU numbers.

True enough. But I wonder how much DX12 features have to do with it.

 

[Cookie Monster]

True enough. But I wonder how much DX12 features have to do with it.

Exactly. Hence why I think its still abit premature to really say which GPU will age better. And most often then not, we end up with faster cheaper GPUs that do it better by the time the age thing comes into play.

I don't think we've also had a PC oriented native DX12 title yet iirc.

 

[Silverforce11]

Hmm pretty interesting.

One thing I think is amazing about Pascal (and why it's not just shrunk Maxwell) is how much it caught up to AMD in so many pro AMD Directx 12 titles. Games like Mordor or SW Battlefront put AMD cards ahead in the Maxwell generation, but Pascal cards do better in all those games. It seems like in particular the 1060 seemed to cut in half the so-called "console effect," which is great for gamers everywhere.

It's the preemption ability (which Maxwell claimed to have, but did not), Pascal can actually do fast context switching so games that were GCN console optimized, with more compute effects, do not suffer slow context switching like Kepler/Maxwell did. This means no performance penalty for games to go compute heavy.

The game that you notice this effect the strongest is actually Quantum Break where they massively use Copy (subset of Compute) queues interweaving with Graphics workloads.

GTX 1060 OC is 40-50% faster than GTX 970 @ 1550mhz in that game:

https://www.youtube.com/watch?v=rRkhA5V1qXU

This game was actually my "go-to" prediction that Pascal would do well on, prior to the launch. Because people were accusing of it being a poor port (it was, but fixed via patches) didn't realize why it ran so badly on Maxwell. It's just the technology used, lots of Graphics + Compute + Copy queues.

 

[Shivansps]

It's the preemption ability (which Maxwell claimed to have, but did not), Pascal can actually do fast context switching so games that were GCN console optimized, with more compute effects, do not suffer slow context switching like Kepler/Maxwell did. This means no performance penalty for games to go compute heavy.

The game that you notice this effect the strongest is actually Quantum Break where they massively use Copy (subset of Compute) queues interweaving with Graphics workloads.

GTX 1060 OC is 40-50% faster than GTX 970 @ 1550mhz in that game:

https://www.youtube.com/watch?v=rRkhA5V1qXU

This game was actually my "go-to" prediction that Pascal would do well on, prior to the launch. Because people were accusing of it being a poor port (it was, but fixed via patches) didn't realize why it ran so badly on Maxwell. It's just the technology used, lots of Graphics + Compute + Copy queues.

Its funny because Nvidia DX12 guide tells developers to avoid using context switching.
https://developer.nvidia.com/dx12-dos-and-donts

-Even for compute tasks that can in theory run in parallel with graphics tasks, the actual scheduling details of the parallel work on the GPU may not generate the results you hope for
-Be conscious of which asynchronous compute and graphics workloads can be scheduled together

-Don’t toggle between compute and graphics on the same command queue more than absolutely necessary
This is still a heavyweight switch to make

 

[Silverforce11]

Its funny because Nvidia DX12 guide tells developers to avoid using context switching.
https://developer.nvidia.com/dx12-dos-and-donts

It's fine for Pascal.

 

[Shivansps]

It's fine for Pascal.

Im just telling what the guide says, most, if not all, the limitations seems to be related to avoid overwork of the CPU OS scheduler.
It also says that not all compute and graphics can run in async, but im not sure that that means.

Anyway the point is, using lots of context switchs is against what nvidia advises, even if is faster on pascal.

This also looks interesting.

-Don’t rely on the driver to parallelize any Direct3D12 works in driver threads
-On DX11 the driver does farm off asynchronous tasks to driver worker threads where possible – this doesn’t happen anymore under DX12
-While the total cost of work submission in DX12 has been reduced, the amount of work measured on the application’s thread may be larger due to the loss of driver threading. The more efficiently one can use parallel hardware cores of the CPU to submit work in parallel, the more benefit in terms of draw call submission performance can be expected.

 

[Bacon1]

New comparison from BabelTechReview, probably deserves a new thread:

So regardless if this guy hates AMD or not, it is funny that the 980 OC seems to be the better buy for DX11, and DX12 would be 480. Where does the 1060 fit in?

Or are we assuming that Maxwell won't get newer optimizations in games so 1060 will surpass it in DX11?

All of the 480 DX12 #s look completely wrong which questions the validity of the rest of the #s. None seem close to any other reviews so maybe he forgot to turn off background recording or something else.

 

[Unreal123]

So regardless if this guy hates AMD or not, it is funny that the 980 OC seems to be the better buy for DX11, and DX12 would be 480. Where does the 1060 fit in?

Or are we assuming that Maxwell won't get newer optimizations in games so 1060 will surpass it in DX11?

All of the 480 DX12 #s look completely wrong which questions the validity of the rest of the #s. None seem close to any other reviews so maybe he forgot to turn off background recording or something else.

What I have learned so far in anandtech forum and from AMD users is that if a 3rd party review is not recommending AMD card then he considered as a shill. If a game is running better on nvidia than a developer is a shill or the game is broken, if a benchmark software is running better on Nvidia then the software is broken. However, if a 3rd party site recommends AMD card than that guy will be considered as the most honest guy in the world, if AMD is winning in a benchmark is simply due to Nvidia is bad and AMD is superior.



This is an inflammayory post and is not allowed in VC&G.

Post to the topic or not at all.


esquared
Anandtech Forum Director

 

[coercitiv]

What I have learned so far in anandtech forum

Well, just like in every community, you first need to learn how to listen. Simply lending your ear to the most vocal members does not necessarily mean you are about to learn anything of value.

So regardless if this guy hates AMD or not, it is funny that the 980 OC seems to be the better buy for DX11

As I previously stated a while back, I was looking forward to people recommending discounted Maxwell against Pascal. (first 970 then 980 once price dropped low enough)

It's not the value that intrigued me, but rather the arguments that would need to be brought into light in order to recommend Pascal.

 

[rh71]

So I plug in the 1060 replacing my 6850 and I can now play Garden Warfare 2 on max res/settings, smooth as glass. Don't even need to upgrade my Q6600, which would have cost me another $500 all told. So sweet at this price. One issue though, the game is literally taking 4+ minutes to load past the splash screen, whereas it was about a minute before. Reinstalled the game too. AMD vs. Nvidia driver issue?

 

[Silverforce11]

What I have learned so far in anandtech forum and from AMD users is that if a 3rd party review is not recommending AMD card then he considered as a shill. If a game is running better on nvidia than a developer is a shill or the game is broken, if a benchmark software is running better on Nvidia then the software is broken. However, if a 3rd party site recommends AMD card than that guy will be considered as the most honest guy in the world, if AMD is winning in a benchmark is simply due to Nvidia is bad and AMD is superior.

Then you obviously have a selective bias at what you pick up. A few people tried to say the same thing since [H] recommended the 480 but not the 1060, no awards for it. I said it didn't change anything since they only use 4-5 games. They happen to pick a few that favors the 480, that is all.

Some sites are not valid, because of their small sample size or biased game selection/settings. Some reviewers are simply not valid based on their actions in the past & current. That's all it is.

Oh, welcome to ATF. You are off the a grand start.

 

[3DVagabond]

Hmm pretty interesting.

One thing I think is amazing about Pascal (and why it's not just shrunk Maxwell) is how much it caught up to AMD in so many pro AMD Directx 12 titles. Games like Mordor or SW Battlefront put AMD cards ahead in the Maxwell generation, but Pascal cards do better in all those games. It seems like in particular the 1060 seemed to cut in half the so-called "console effect," which is great for gamers everywhere.

I needed a beast card for MadVR so I don't regret my 480 purchase, but my friend who I am helping build a mini ITX VR machine should probably get a 1060 instead. I personally think the 480 will age a little better than the 1060 but it won't be as dramatic as last generation and AMD has a lot of ground to catch up after OCing. The benefits of not going green team are quickly evaporating, I hope AMD treads carefully the next 12 months.

Don't trust Babeltech. Look at more reviews before you make any decisions. apoppin hates AMD about as much as anyone could.

 

[3DVagabond]

Im just telling what the guide says, most, if not all, the limitations seems to be related to avoid overwork of the CPU OS scheduler.
It also says that not all compute and graphics can run in async, but im not sure that that means.

Anyway the point is, using lots of context switchs is against what nvidia advises, even if is faster on pascal.

This also looks interesting.

That guide was made before Pascal and was directed at Maxwell.

 

[3DVagabond]

What I have learned so far in anandtech forum and from AMD users is that if a 3rd party review is not recommending AMD card then he considered as a shill. If a game is running better on nvidia than a developer is a shill or the game is broken, if a benchmark software is running better on Nvidia then the software is broken. However, if a 3rd party site recommends AMD card than that guy will be considered as the most honest guy in the world, if AMD is winning in a benchmark is simply due to Nvidia is bad and AMD is superior.

Just go to their forum and read it. You can draw your own conclusions.

 

[vissarix]

So regardless if this guy hates AMD or not, it is funny that the 980 OC seems to be the better buy for DX11, and DX12 would be 480. Where does the 1060 fit in?

Or are we assuming that Maxwell won't get newer optimizations in games so 1060 will surpass it in DX11?

All of the 480 DX12 #s look completely wrong which questions the validity of the rest of the #s. None seem close to any other reviews so maybe he forgot to turn off background recording or something else.

Hahahaha dude are you sure you dont need glasses?🙄 the gtx 1060 is anywhere from 10 to 20% faster on dx11 games and its faster on dx12 aswell compared to the rx480 on that review...look closely, zoom in or just call this fake, biased, shill review 😆😆

 

[Bacon1]

Hahahaha dude are you sure you dont need glasses?🙄 the gtx 1060 is anywhere from 10 to 20% faster on dx11 games and its faster on dx12 aswell compared to the rx480 on that review...look closely, zoom in or just call this fake, biased, shill review 😆😆

Maybe you need glasses?

980 OC is faster than 1060 OC in:

FireStrike
Timespy
Metro Last Light Redux
Shadows of Mordor
Alien Isolation
DA:I
Dying Light
Total War: Attila
GTA V (>1080p)
Project Cars
Witcher 3
Mad Max
Batman: Arkham Knight
FO 4
SW: Battlefront
Assassin's Creed Syndicate
Just Cause 3
R6S
Dirt Rally
FC Primal
The Division
DOOM
Mirrors Edge Catalyst (>1080p)
Hitman (1440p)
Ashes of the Singularity
Total War Warhammer

1060 faster in:

Crysis 3
GTA V (1080p)
ME: Catalyst (1080p)
Hitman (1080p)

Tied:
ROTTR

 

[Keysplayr]

Maybe you need glasses?

980 OC is faster than 1060 OC in:

Schnip*

To actually make your statement real, we would need to swallow the red pill.

Anyone that has eyes, and a brain, can see you are playing a game.

1060 is a lot faster than 480 in all DX11 games. And is for the most part equal in DX12 games. 1060 is usually tied or within 5 frames to the 980.
Most of the time very close. .1 to 3 fps.

Just keep it real man. It doesn't hurt.

 

[frozentundra123456]

Not to mention new vs new, the 980 is a hundred dollars more expensive and will use more power.

 

[Keysplayr]

Not to mention new vs new, the 980 is a hundred dollars more expensive and will use more power.

And while 4GB is enough for today's gaming (for the most part barring 4K ultra texture fubars) the extra 2GB or 4GB on the 1060 and 480 respectfully, is nice too.

 

[raghu78]

To actually make your statement real, we would need to swallow the red pill.

Anyone that has eyes, and a brain, can see you are playing a game.

1060 is a lot faster than 480 in all DX11 games. And is for the most part equal in DX12 games. 1060 is usually tied or within 5 frames to the 980.
Most of the time very close. .1 to 3 fps.

Just keep it real man. It doesn't hurt.

Depending on the game the Rx 480 can be significantly faster even in DX11 game. I agree on average the GTX 1060 is faster than Rx 480 in DX11 games. btw I can bet there are more CoD players than any other game.

https://www.computerbase.de/2016-07...diagramm-call-of-duty-black-ops-iii-1920-1080

But even that average is 5-10% depending on reviewer test suite. The same goes for DX12 / Vulkan performance where AMD is roughly 5-10% faster. btw the latest 16.7.3 driver brought a nice perf jump in RoTR which was one of the DX12 titles where AMD lagged Nvidia badly. So we will see reviews updated with this perf gain by the time Rx 470 launches.

http://hexus.net/tech/reviews/graphics/94969-sapphire-radeon-rx-480-nitro-4gb-8gb-oc/?page=10

In the latest hexus review Sapphire Rx 480 Nitro 8G won more games against MSI GTX 1060 Gaming. Sapphire Rx 480 8G won big in Doom Vulkan (15% at 1080p and 20% at 1440p) , Hitman (14% at 1080p, 16% at 1440p) while winning Total War Warhammer by (4% at 1080p and 5% at 1440p) and Division (by 3% at 1080p and 4% at 1440p). GTX 1060 won Dirt Rally (by 15% at 1080p, 11% at 1440p) , RoTR (9% at 1080p, 3% at 1440p), Fallout (8% at 1080p, 7% at 1440p). Sapphire Rx 480 was the faster card on avg.

You can say that hexus test suite is not large which I can agree is a fair point. But thats the case. Depending on choice of games any test suite can favour one vendor or the other. So it comes down to what games you intend to play. Do your research and pick your GPU. As of now AMD Rx 480 wins DX12 games > 5% avg with Doom Vulkan being a specific outlier because of GCN shader intrinsics and Vulkan optimizations.

We are going to see a slew of high profile DX12 games in the next 6 months and majority of them are AMD Gaming Evolved titles. Deus Ex Mankind Divided, Battlefield 1, Watch Dogs 2, Civilization VI, Gears of War 4, Forza Horizon 3. AMD is pushing GPUOpen very aggressively to combat Gameworks and we can expect more games to run much better on one vendor or the other based on who the developer aligns with. So it makes sense for the gamer to buy based on perf in games he/she is going to be playing.

 

[Keysplayr]

Depending on the game the Rx 480 can be significantly faster even in DX11 game. I agree on average the GTX 1060 is faster than Rx 480 in DX11 games. btw I can bet there are more CoD players than any other game.

https://www.computerbase.de/2016-07...diagramm-call-of-duty-black-ops-iii-1920-1080

But even that average is 5-10% depending on reviewer test suite. The same goes for DX12 / Vulkan performance where AMD is roughly 5-10% faster. btw the latest 16.7.3 driver brought a nice perf jump in RoTR which was one of the DX12 titles where AMD lagged Nvidia badly. So we will see reviews updated with this perf gain by the time Rx 470 launches.

http://hexus.net/tech/reviews/graphics/94969-sapphire-radeon-rx-480-nitro-4gb-8gb-oc/?page=10

In the latest hexus review Sapphire Rx 480 Nitro 8G won more games against MSI GTX 1060 Gaming. Sapphire Rx 480 8G won big in Doom Vulkan (15% at 1080p and 20% at 1440p) , Hitman (14% at 1080p, 16% at 1440p) while winning Total War Warhammer by (4% at 1080p and 5% at 1440p) and Division (by 3% at 1080p and 4% at 1440p). GTX 1060 won Dirt Rally (by 15% at 1080p, 11% at 1440p) , RoTR (9% at 1080p, 3% at 1440p), Fallout (8% at 1080p, 7% at 1440p). Sapphire Rx 480 was the faster card on avg.

You can say that hexus test suite is not large which I can agree is a fair point. But thats the case. Depending on choice of games any test suite can favour one vendor or the other. So it comes down to what games you intend to play. Do your research and pick your GPU. As of now AMD Rx 480 wins DX12 games > 5% avg with Doom Vulkan being a specific outlier because of GCN shader intrinsics and Vulkan optimizations.

We are going to see a slew of high profile DX12 games in the next 6 months and majority of them are AMD Gaming Evolved titles. Deus Ex Mankind Divided, Battlefield 1, Watch Dogs 2, Civilization VI, Gears of War 4, Forza Horizon 3. AMD is pushing GPUOpen very aggressively to combat Gameworks and we can expect more games to run much better on one vendor or the other based on who the developer aligns with. So it makes sense for the gamer to buy based on perf in games he/she is going to be playing.

And absolutely no new DX12 titles will be TWIMTBP? What's good for one is good for the other.

And why only link CoDIII? That's the only one that shows 480 ahead of 1060 and that's the one you linked.. See this is what I'm talking about on these forums. Nobody can keep things real.

 

[Sweepr]

New comparison from Hardware.fr:

Gainward GTX 1060 vs RX 480 Sapphire Nitro+: €280 match

- Power: Gainward GTX 1060 draws 83W less under load (TW3 Hairworks Off)
- Noise: 10.7 dBa less under load (!), 6.8 dBa less than Nitro+ at Silence Mode
- Overclock:
208 MHz GPU (+ 11%) and 2300 MHz memory (+ 15%) vs +1380 MHz (+ 3%) and 2250 MHz memory (+ 12.5%)
- Actual Gain from Overclocking (Overall): 12.3% vs 6.1%

A € 280, so without much hesitation we give our preference to these Gainward 1060 GTX or GeForce Palit offers a very good compromise for playing 1080p.

www.hardware.fr/focus/118/rx-480-sapphire-nitro-vs-gtx-1060-gainward-match-280.html

What I have learned so far in anandtech forum and from AMD users is that if a 3rd party review is not recommending AMD card then he considered as a shill.

Quoting an inflammatory post with a thumbsup, is the same as posting it yourself.

Stop it now. I will lock this thread if it continues.


esquared
Anandtech Forum Director