Gtx 1060

[Deleted]

So I'm buying a new gaming PC and my GTX 1060 will be my GPU, could someone tell me if 1060 is a good GPU to buy atm?

Hard to say, depends on your budget, how cheap you can get one and what you are comparing it to. I myself am trying to decide between GTX 1060 and RX 480. I decided to wait for RX 480 aftermarket cards, however.

[Remilia]

Am I the only one, that has a closed case, placed under the desktop?
I see my interior maybe twice a year when I clean it, lol.

Closed case! Yes.
I don't even clean it (yay dust filter) unless I actually have to open the case for whatever reason. And then I still don't have much of any dust.

Quantum Break and Rise of the Tomb Raider are similar in that they heavily favor AMD or Nvidia. But they're both riddled with problems, and yet people use them in benchmarks. Time Spy is a strange synthetic benchmark, cause it doesn't seem to push Async Compute as hard as Doom does. Doom is very well optimized for Vulkan. Hitman we know has been optimized by AMD as well as Ashes of the Singularity. But of all these games, Doom shows a lot of promise with Vulkan. The next big title to look out for is Deus Ex: Mankind Divided, which will be out August 23rd.

Think Quantum Break got a patch to fix the gigantic launch clusterfuck now.

[coprax]

Closed case! Yes.
I don't even clean it (yay dust filter) unless I actually have to open the case for whatever reason. And then I still don't have much of any dust.

I have dust filters, too (fractal design R5, fyi) but there is still dust after some time, it's not 100% dustproof^^

[Ahovv]

Time Spy gains roughly 10% on AMD cards with Async enabled which I believe is roughly the same on Doom. Not talking about the OpenGL->Vulkan gains here.

https://www.reddit.com/r/Amd/comment...rding_timespy/

[Vash The Stampede]

Time Spy gains roughly 10% on AMD cards with Async enabled which I believe is roughly the same on Doom. Not talking about the OpenGL->Vulkan gains here.

Vulkan gives AMD cards nearly 30% increase, while Nvidia gets nearly 10%. Hitman and Ashes gives AMD roughly 10% while for Nvidia is gives nothing or less. Hard to tell what Async Compute does alone when only TIme Spy is the only one with the ability to enable or disable Async Compute. But then again, Time Spy doesn't have a DX11 switch either.

- - - Updated - - -

So I'm buying a new gaming PC and my GTX 1060 will be my GPU, could someone tell me if 1060 is a good GPU to buy atm?

No such thing as a bad GPU, just a bad price. At 250 USD the 1060 is a steal, but the problem begins with the 300 USD Founders Edition price, which can influence the value of the 1060's. I would not spend more than 260 USD for a 1060, cause any more and you'd be looking at 1070 territory. But the 1070's aren't at their 380 USD price either, thanks to FE's.

I would keep an eye out for 1060's to see if they've come down in price, which can take months. But the RX 480 will soon have partner cards out, which hopefully bridge the performance gap a bit and push the 1060's for their MSRP. Also keep in mind that these new cards do depend on what game you play. A YouTuber called MXR who reviews Skyrim and Fallout mods has upgraded from a GTX 970 to a GTX 1080 and didn't see any fps increase. Of course he plays with modded games. Depending on the games you play, alternatives like the 970 and 980 might not be a bad idea. Especially because the 1060 has nothing architecturally over them that would warrant you to go buy it. The 1060 sits in between them in performance, and those cards aren't having supply issues, or Founders Edition price gouging issues. The GTX 980 is probably too much right now even compared to a FE 1060, but prices should plummet soon.

Don't forget the RX 480 cause it isn't a bad card either, and cheaper. Plus we know it can match or beat a 1060 in DX12/Vulkan games. But if you plan to play Battlefield 3 for 5 years then you'd probably want 1060. If you want to play Deus X Mankind at War for 3 years, the RX 480 makes sense.

- - - Updated - - -

https://www.reddit.com/r/Amd/comment...rding_timespy/

I'm going to leave that alone cause nothing from Futuremark states that they used a path that fits Pascal architecture capabilities. I never took 3Dmark scores seriously in the past, and not about to start now. In this day in age what does 3Dmark scores prove? Absolutely nothing when Nvidia or AMD can get their hands on a game and completely change the way it runs on your graphics card. Rise of the Tomb Raider? More like Rise of the Nvidia Raider. Hitman? More like AMDman. Need to stop favoring one side. Either optimize for both or don't optimize for either.

Intel graphic users are probably looking at this and going, "I don't even?".

[Zenny]

https://www.reddit.com/r/Amd/comment...rding_timespy/

"but DX12 is all about low level optimization utilizing hardware specific code path"

No that's bullshit, both AMD and Nvidia have recommended to Futuremark that they not do so. Time Spy doesn't have any vendor specific optimizations. Both AMD and Nvidia signed off on it: https://www.futuremark.com/business/...opment-program. Hell: "The work that Time Spy places into the COMPUTE queue and the specific implementation of that work is the result of deep co-operation with all BDP members including AMD, Intel, Microsoft, and NVIDIA among others." Lets be real here, the only reason people are complaining is because AMD is not crushing Nvidia, if the results were different certain people wouldn't have said a thing.

The funny thing is if people wanted this DX12 benchmark to be "pushed to the limit" most AMD cards including the Fury X wouldn't be able to run it. Simply because GCN 1.0, 1.1 and 1.2 do not support the highest feature level of DX12 while Maxwell and Pascal does.

As Futuremark themselves have noted, Ashes of the Singularity has zero vendor specific optimizations either, while Doom (a Vulkan title) has optimizations for AMD but none for Nvidia. Should we toss Doom results out now for not being fair?

- - - Updated - - -

Vulkan gives AMD cards nearly 30% increase, while Nvidia gets nearly 10%. Hitman and Ashes gives AMD roughly 10% while for Nvidia is gives nothing or less. Hard to tell what Async Compute does alone when only TIme Spy is the only one with the ability to enable or disable Async Compute. But then again, Time Spy doesn't have a DX11 switch either.

I'm not referring to the boost AMD gets from Vulkan, that is indeed in excess of 30%, I'm referring to Async on/off in vulkan itself. The boost for that is roughly 10%:

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

- - - Updated - - -

Also a very interesting read on async compute on pascal vs preemption:

Async Compute:

http://www.anandtech.com/show/10325/...ition-review/9

Preemption:

http://www.anandtech.com/show/10325/...tion-review/10

[Vash The Stampede]

Had another look at HardOCP's benches and notice their Rise of the Tomb Raider benchmark was different. Starting to wonder if the other review sites had applied the Async Compute patch for the game. Some websites don't even know that Doom's Vulkan only works properly with TSAA or NO AA. I tried to see if there was any mention of review sites using the latest RotTR Async patch and I can't find any. Why does HardOCP's RotTR score higher for the RX 480 compared to the 1060, when everyone else's scores show the 1060 takes a dump on the RX 480 in this game?

Something's wrong here.

[Zenny]

Had another look at HardOCP's benches and notice their Rise of the Tomb Raider benchmark was different. Starting to wonder if the other review sites had applied the Async Compute patch for the game. Some websites don't even know that Doom's Vulkan only works properly with TSAA or NO AA. I tried to see if there was any mention of review sites using the latest RotTR Async patch and I can't find any. Why does HardOCP's RotTR score higher for the RX 480 compared to the 1060, when everyone else's scores show the 1060 takes a dump on the RX 480 in this game?

Something's wrong here.

That maximum for the 1060 seems way out of line compared to the RX 480 and 980. Hell those maximums for the 980 and RX480 seem way out of line for any benchmark I've seen for the game. Don't think they are using the inbuilt benchmark for Rise of the Tomb Raider.

[Vash The Stampede]

That maximum for the 1060 seems way out of line compared to the RX 480 and 980. Hell those maximums for the 980 and RX480 seem way out of line for any benchmark I've seen for the game. Don't think they are using the inbuilt benchmark for Rise of the Tomb Raider.

When I first heard about the Async Compute patch for RotTR, the results between the 980 and RX 480 looked much like that. Older tests had a massive 20 fps difference, much like the ones I see on other websites. Here's TechPowerUps RotTR at 1440p DX12. I'm thinking some websites aren't updating their games.

https://www.techpowerup.com/reviews/...ming_X/19.html



- - - Updated - - -

AdoredTV has a take on the 1060 benchmarks and he agrees that the benchmarks that websites are performing are more of a topic than the 1060 itself. Why no Vulkan Doom? Why lots of old DX11 games? Why no DX12 for Hitman?

[Artorius]

The 1060 beats the RX480 in Time Spy

Time Spy is a very limited dx12 "approach" and sincerely speaking it doesn't even matter for what we're talking about here. There will be games that try their best to give the most performance possible using those new APIs, Doom is a good example. Time Spy aims to be the lowest common denominator, which means giving a "fair fight" for cards as old as Kepler cards. It isn't by any means showcasing what the APIs are capable of doing.

and Rise of the Tomb Raider,

Nvidia sponsored game that also hardly matters when we're talking about Vulkan/DX12 performance.

it matches it in Ashes of the Singularity

This is actually quite impressive but it still losing.

and Warhammer and looses in Hitman and Quantum Break.

Hitman doesn't count, it's a AMD sponsored game and heavily tuned for AMD cards. Don't really know about the other two but any decent DX12/Vulkan implementation will put the RX480 ahead of the 1060, regardless of how much compute there is.

"Loosing horribly" is utterly wrong.

A 31% lead for the RX480 on Vulkan is "losing horribly" for the 1060.

Before anyone starts claiming that Time Spy is not proper DX12 or doesn't have real async compute (you know despite offering zero proof to support that retarded claim) read this:

It isn't. It's the "least" DX12 as possible.

Let's compare 2 different implementations of those low overhead APIs:





Can you spot the difference?

I won't lose time writing this again so I'll just quote a poster from Anandtech.

The issue here is, they (3Dmark) are purposely neutering what a full DX12 / Vulkan engine can actually do, to appease a vendor who uses drivers to get around their lack of hardware functions.
Since they must do it in drivers, that introduces expensive context switching, so, obviously, they (nvidia) don't want to do that, since it shows their hardware severely under performing the competition. (BTW, hardware context switching is faster than software, but you still get a penalty of doing it for "half support".)

As was previously shown in this thread, the Fury card is faster than Nvidia's newest lineup using a fairly well optimized Vulkan game (Doom), and that is a hard pill to swallow for people who just dropped down $500+ for their newest toy.

So, once again, it proves that 3Dmark is basically worthless as a real world test to show what a full implementation of DX12/Vulkan can do.

Original post.

[Bigvizz]

No such thing as a bad GPU, just a bad price. At 250 USD the 1060 is a steal, but the problem begins with the 300 USD Founders Edition price, which can influence the value of the 1060's. I would not spend more than 260 USD for a 1060, cause any more and you'd be looking at 1070 territory. But the 1070's aren't at their 380 USD price either, thanks to FE's.

I would keep an eye out for 1060's to see if they've come down in price, which can take months. But the RX 480 will soon have partner cards out, which hopefully bridge the performance gap a bit and push the 1060's for their MSRP..

Months? they're already at MSRP for many AIB models. They just have to come back in to stock give it a week or so.

[Vash The Stampede]

Months? they're already at MSRP for many AIB models. They just have to come back in to stock give it a week or so.

Just like the RX 480, and they're sold out or back order. The RX 480 has been out for a good long time and it certainly hasn't normalized in price and availability. Why you think I said whoever can produce more cards for their MSRP will win, cause it doesn't look like the sales for either the 480 or 1060 are slowing down. The demand is just huge.

Interestingly, I've not heard of anyone who's gotten a 4GB RX 480 that wasn't really a 8GB. And I heard that AMD might have stopped making the 4GB cards. Nothing confirmed, just what I hear through the grape vine.

[Zenny]

Time Spy is a very limited dx12 "approach" and sincerely speaking it doesn't even matter for what we're talking about here. There will be games that try their best to give the most performance possible using those new APIs, Doom is a good example. Time Spy aims to be the lowest common denominator, which means giving a "fair fight" for cards as old as Kepler cards. It isn't by any means showcasing what the APIs are capable of doing.

Time Spy is the exact same DX12 feature level as every single DX12 game on the market. DX12 feature levels have zero corresponding impact on what performance the API is capable of and higher feature levels won't give a magical boost in performance. It's hilarious that your bring up Kepler as the reason for this when the highest feature level of DX12 won't even run on any pre-polaris cards. I've pointed this out to you before but I guess you conveniently ignored me.

Nvidia sponsored game that also hardly matters when we're talking about Vulkan/DX12 performance.

Yeah, lets ignore the performance improvements made to the game and the fact async is now supported.

This is actually quite impressive but it still losing.

Ah yes, lets not bring up the other benchmarks sites that show a far closer match:
http://arstechnica.co.uk/gadgets/201...x-1060-review/
http://www.guru3d.com/articles_pages...review,16.html

Faster by a single frame in the one, slower by a single fram in the other. Guess the Ashes team sold out to Nvidia

Hitman doesn't count, it's a AMD sponsored game and heavily tuned for AMD cards. Don't really know about the other two but any decent DX12/Vulkan implementation will put the RX480 ahead of the 1060, regardless of how much compute there is.

These sentences just reflect how much bias you have. "Screw what benchmarks say, I KNOW NVIDIA IS SLOWER". Facts don't matter!

A 31% lead for the RX480 on Vulkan is "losing horribly" for the 1060.

I'm sure Doom is the real performance indicator of the difference between these two cards, despite it running on a older version of Vulkan on Nvidia cards and the official statement from the publisher is that they are still working with Nvidia on it.

It isn't. It's the "least" DX12 as possible.

Because that would make it the most compatible? The higher feature levels don't bring significant performance increases to the API, low level access and async compute are all inherent to DX12 no matter the feature level.
https://en.wikipedia.org/wiki/Featur...ls_in_Direct3D

As I mentioned earlier the most advanced feature levels of DX12 are not even supported by pre-Polaris cards. While Maxwell and Pascal do, heck Pascal supports the optional features as well. It would be a lark for DX12 games to seriously support these features and not have AMD cards being able to run the game.

Let's compare 2 different implementations of those low overhead APIs:





Can you spot the difference?

I sure can, one is a vendor neutral engine built from the ground up and approved fully by AMD, Nvidia and Intel, the other is a game that runs on a older version of a API for a specific vendor, has specific optimizations for the other vendor and is still being worked on by the developer with the 1st vendor. But I'm sure it's a 100% accurate representation of the new API's and every single DX12 implementation (including the only true DX12 engine) is simply wrong.

Psst, other sites found the difference in performance to be a weee bit closer:

http://www.eurogamer.net/articles/di...formance-gains
https://www.youtube.com/watch?v=ZQi6wwSOhZU

I won't lose time writing this again so I'll just quote a poster from Anandtech.

So let me get this straight, I'm supposed to believe a random forum poster (which appears to pull all of this out of his ass) over the review from the site itself?

http://www.anandtech.com/show/10325/...ition-review/9

Wait let me quote someone from that forum as well!

Your "issue" is tin foil hat theorizing and speculation. There's no proof that async compute was neutered to appease Nvidia; in fact there's no hard proof that async compute was neutered at all, just speculation. Would you support it if Time Spy was rendering with a full FL12.1 codepath and thus using features only Nvidia cards support even if it meant Nvidia still had the performance advantage? It's alright to talk about this, but it's kind of presumptuous to act like anything's been proven.

Awww yeah.

[Ahovv]

"but DX12 is all about low level optimization utilizing hardware specific code path"

No that's bullshit, both AMD and Nvidia have recommended to Futuremark that they not do so. Time Spy doesn't have any vendor specific optimizations. Both AMD and Nvidia signed off on it: https://www.futuremark.com/business/...opment-program. Hell: "The work that Time Spy places into the COMPUTE queue and the specific implementation of that work is the result of deep co-operation with all BDP members including AMD, Intel, Microsoft, and NVIDIA among others." Lets be real here, the only reason people are complaining is because AMD is not crushing Nvidia, if the results were different certain people wouldn't have said a thing.

The funny thing is if people wanted this DX12 benchmark to be "pushed to the limit" most AMD cards including the Fury X wouldn't be able to run it. Simply because GCN 1.0, 1.1 and 1.2 do not support the highest feature level of DX12 while Maxwell and Pascal does.

As Futuremark themselves have noted, Ashes of the Singularity has zero vendor specific optimizations either, while Doom (a Vulkan title) has optimizations for AMD but none for Nvidia. Should we toss Doom results out now for not being fair?

- - - Updated - - -



I'm not referring to the boost AMD gets from Vulkan, that is indeed in excess of 30%, I'm referring to Async on/off in vulkan itself. The boost for that is roughly 10%:

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

- - - Updated - - -

Also a very interesting read on async compute on pascal vs preemption:

Async Compute:

http://www.anandtech.com/show/10325/...ition-review/9

Preemption:

http://www.anandtech.com/show/10325/...tion-review/10

The short of it is, Vulkan is a better default platform to develop around, and it is in Nvidia's best interest to optimize around it. Yes, I fully understand that Vulkan was based off Mantle (AMD clearly contributed in a major way).

However, even Nvidia admits/flaunts Vulkan as the next-generation, cutting edge API. What's going to happen in a few years when most modern games support/optimize around Vulkan? Will there be excuses then, or will Nvidia adapt their technology to better support it? The answer is clear, and it's not valid whatsoever to compare GPUs and then fail to acknowledge the current advantage AMD has on futureproofing.

Anyway, even if we toss out every other bit of data regarding these GPUs, the rx 480 4gb version is by far the cheapest 60fps 1080p ultra GPU on the market.

1440p makes for an interesting comparison between the 8gb 480 and the 1060 since they are similarly priced. But again, there is futureproofing to consider, and by that alone I would opt for the 480.

I'm not an AMD fanboy; years ago I had the 8800gtx and was happy with it, although it was pricey. I just think there is a more compelling argument to be made for the 480.

[Deleted]

Had another look at HardOCP's benches and notice their Rise of the Tomb Raider benchmark was different. Starting to wonder if the other review sites had applied the Async Compute patch for the game. Some websites don't even know that Doom's Vulkan only works properly with TSAA or NO AA. I tried to see if there was any mention of review sites using the latest RotTR Async patch and I can't find any. Why does HardOCP's RotTR score higher for the RX 480 compared to the 1060, when everyone else's scores show the 1060 takes a dump on the RX 480 in this game?

Something's wrong here.

Didn't they release a new patch, sorta fixing the whole DX12 thing. If so, could be [H] is using the new patch whereas other sites can't be bothered to test it using the new patch.

[Zenny]

The short of it is, Vulkan is a better default platform to develop around, and it is in Nvidia's best interest to optimize around it. Yes, I fully understand that Vulkan was based off Mantle (AMD clearly contributed in a major way).

However, even Nvidia admits/flaunts Vulkan as the next-generation, cutting edge API. What's going to happen in a few years when most modern games support/optimize around Vulkan? Will there be excuses then, or will Nvidia adapt their technology to better support it? The answer is clear, and it's not valid whatsoever to compare GPUs and then fail to acknowledge the current advantage AMD has on futureproofing.

Anyway, even if we toss out every other bit of data regarding these GPUs, the rx 480 4gb version is by far the cheapest 60fps 1080p ultra GPU on the market.

1440p makes for an interesting comparison between the 8gb 480 and the 1060 since they are similarly priced. But again, there is futureproofing to consider, and by that alone I would opt for the 480.

I'm not an AMD fanboy; years ago I had the 8800gtx and was happy with it, although it was pricey. I just think there is a more compelling argument to be made for the 480.

DX12 and Vulkan are both very very similar to each other, Doom is the only Vulkan benchmark we have at the moment and using it as a single data point seems a bit premature. Especially since we know the Nvidia cards have not been fully optimized for it yet.

[Drunkenvalley]

Especially since we know the Nvidia cards have not been fully optimized for it yet.

I don't think that's a meaningful option for nVidia with DX12/Vulkan.

[Deleted]

There is actually another DX12 game out right now, which seemed to slipped past by just about everyone... Shame on us.. Forza 6, should be free in the windows store.

A german site tested different DX12 games for the 480 and 1060.

[Zenny]

I don't think that's a meaningful option for nVidia with DX12/Vulkan.

Why not? We know Async works on Pascal, we have 2 examples of it so far.

- - - Updated - - -

There is actually another DX12 game out right now, which seemed to slipped past by just about everyone... Shame on us.. Forza 6, should be free in the windows store.

A german site tested different DX12 games for the 480 and 1060.

Bizzare, that site has the RX480 beat the 1060 while this benchmark has the opposite:

[Deleted]

Bizzare, that site has the RX480 beat the 1060 while this benchmark has the opposite:

I don't know if the game as a benchmark build in, so could be they used different settings.

The german site specifically states they have it raining during their benchmark run. Could be gameaxis does not have it on.