• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

RX480 vs GTX1060

MyBrains said:
Idle and load temps for comparison.. thought i'd match your 40% fan speed for load temps.

Load temps is running DayZ last night for over half hour before logging temps, i use Vsync in the few games i play and have noticed if i disable Vsync the temps can rise a bit but without Vsync i seem to get some stutters like i had on my 7950 but Vsync feels so smooth locked to 60fps :)

Idle.. (28c @ 14% Fan Speed = around 610rpm)

https://imgur.com/a/zs8BV

Load.. (58c @ 40% Fan Speed = around 1650rpm) < also overclocked here @ 1366/1800.

https://imgur.com/a/bxb2O

Didnt embed images as i cant find the option to do so anymore with imgur, maybe i have to make an account there now, but something has changed :(

Heres a pic of my Custom Fan Profile..

https://imgur.com/a/wXFVw (Idling @ 27c atm :p )

Note: DayZ Standalone isnt the CPU dependant game it used to be since the new renderer (Enfusion Engine) brought DX11 and uses the GPU a hell of a lot more now.
Click to expand...

That's good to see. I was a bit puzzled when some reviewers were saying that the card almost got to 80c, which seems ridiculous when the reference model got the about there, seemed like a custom cooler did nothing.

I could bust open Dayz for a more direct comparison as well at 40% speed.
 
So I was bored, stock speed 1060s cost the same as a devil
4Bh6J9u.png
 
The 1060 6GB is clearly the faster card right now and is a good buy for people that upgrade often.

On the other hand, the 480 is probably going to shine within 2017 if more games like Doom come along and DX12/Vulkan along with shader intrinsics and all that become more commonly used. There's claims that the 480 is 30% faster than the 1060 in Doom when the 480 is running on Vulkan.

I can't help but feel certain that people who hold onto a 6GB 1060 going into 2018 will regret it.

The same can be said for that 3GB card Nvidia calls a 1060, vs the 470 I guess...
 
akarypid said:
The 1060 6GB is clearly the faster card right now and is a good buy for people that upgrade often.

On the other hand, the 480 is probably going to shine within 2017 if more games like Doom come along and DX12/Vulkan along with shader intrinsics and all that become more commonly used. There's claims that the 480 is 30% faster than the 1060 in Doom when the 480 is running on Vulkan.

I can't help but feel certain that people who hold onto a 6GB 1060 going into 2018 will regret it.

The same can be said for that 3GB card Nvidia calls a 1060, vs the 470 I guess...
Click to expand...

I guess it depends on just how many game companies will be using dx12 and vulcan, and if a person with a 1060 actually intends to play these games. For example, I have a 1060 but have no interest in doom. Therefore, I've not lost out in my mind. But, a game I am really looking forward to is Halo 5: Forge. If that used DX12/Vulcan and I had terrible performance I would be a bit miffed. Unless there is an overwhelmingly big list of companies committing to DX12/Vulcan then I'm not sure what the true outcomes will be

Another issue is why only AMD has, at the moment, been able to take proper advantage of DX12 (most of the time) and Vulcan (all of the time) over Nvidia. What specifically makes it so that Nvidia hasn't gained any performance for these API's yet?
 
Last edited:
southernorth said:
I guess it depends on just how many game companies will be using dx12 and vulcan, and if a person with a 1060 actually intends to play these games. For example, I have a 1060 but have no interest in doom. Therefore, I've not lost out in my mind. But, a game I am really looking forward to is Halo 5: Forge. If that used DX12/Vulcan and I had terrible performance I would be a bit miffed. Unless there is an overwhelmingly big list of companies committing to DX12/Vulcan then I'm not sure what the true outcomes will be

Another issue is why only AMD has, at the moment, been able to take proper advantage of DX12 (most of the time) and Vulcan (all of the time) over Nvidia. What specifically makes it so that Nvidia hasn't gained any performance for these API's yet?
Click to expand...

This was explained to me in another thread. Apparently Maxwell lacks the hardware support to be able to efficiently implement DX12. Nvidia claimed they would do it in software, but that turned out to be a disaster (games were actually slower in DX12).

Pascal has two important features that allow for a meaningful DX12 implementation: reasonably fast context-switching (they advertise <100microseconds) and something branded "dynamic load balancing" which allows the card to switch from doing graphics work to compute work.

In both cases GCN has had superior capabilities for a while now, as it has hardware to address both of these things. Context switching is a matter of clock cycles (near instant) and the switching from graphics to compute therefore can occur with no constraints (and in fact there's schedulers implemented in hardware to do this).

You can look it up, there's a lot of misinformation on the subject from both sides, but the bottom line from when I looked into it was:

- Maxwell can't do DX12 and never will no matter what Nvidia say
- Pascal can do DX12 (lots of people claim it lacks async features but that's not true) but not as good as GCN

EDIT: regarding Maxwell, I mean that it will never be able to run faster in DX12 than DX11 and any software emulation of DX12 will always be considerably slower

EDIT 2: as noted by Rroff below, I'm talking about async, but that's the gist of the DX12 API
 
Last edited:
^^ Its Async you are talking about - both Maxwell and Pascal have full hardware for most DX12 features. There is more to DX12 than just Async.

Gen 2 Maxwell can do graphics + compute but there are some severe issues if you just throw work at it indiscriminately which can even crash the OS hence why nVidia hasn't hurried to driver support (and as mentioned with context switching you can get some big performance penalties if you just throw work at it as if it is a GCN architecture). Pascal has some significant advantages one of which is being able to fold work in behind the graphics job if it completes early enough which Maxwell can't do and will never be able to do - which will give the 1000 series cards a decent boost in anything that can take advantage of that over the 980ti and Maxwell TX.
 
Last edited:
southernorth said:
I guess it depends on just how many game companies will be using dx12 and vulcan, and if a person with a 1060 actually intends to play these games. For example, I have a 1060 but have no interest in doom. Therefore, I've not lost out in my mind. But, a game I am really looking forward to is Halo 5: Forge. If that used DX12/Vulcan and I had terrible performance I would be a bit miffed. Unless there is an overwhelmingly big list of companies committing to DX12/Vulcan then I'm not sure what the true outcomes will be

Another issue is why only AMD has, at the moment, been able to take proper advantage of DX12 (most of the time) and Vulcan (all of the time) over Nvidia. What specifically makes it so that Nvidia hasn't gained any performance for these API's yet?
Click to expand...

Halo 5 is dx12 only like pretty much all other Microsoft games. These tend to favor AMD for obvious reason's being xbox orientated.
 
Rroff said:
^^ Its Async you are talking about - both Maxwell and Pascal have full hardware for most DX12 features. There is more to DX12 than just Async.

Gen 2 Maxwell can do graphics + compute but there are some severe issues if you just throw work at it indiscriminately which can even crash the OS hence why nVidia hasn't hurried to driver support (and as mentioned with context switching you can get some big performance penalties if you just throw work at it as if it is a GCN architecture). Pascal has some significant advantages one of which is being able to fold work in behind the graphics job if it completes early enough which Maxwell can't do and will never be able to do - which will give the 1000 series cards a decent boost in anything that can take advantage of that over the 980ti and Maxwell TX.
Click to expand...

Maxwell does Async too according to Nvidia:

http://international.download.nvidi...tx-980-ti-directx-12-advanced-api-support.png

:p
 
Rroff said:
^^ Its Async you are talking about - both Maxwell and Pascal have full hardware for most DX12 features. There is more to DX12 than just Async.

Gen 2 Maxwell can do graphics + compute but there are some severe issues if you just throw work at it indiscriminately which can even crash the OS hence why nVidia hasn't hurried to driver support (and as mentioned with context switching you can get some big performance penalties if you just throw work at it as if it is a GCN architecture). Pascal has some significant advantages one of which is being able to fold work in behind the graphics job if it completes early enough which Maxwell can't do and will never be able to do - which will give the 1000 series cards a decent boost in anything that can take advantage of that over the 980ti and Maxwell TX.
Click to expand...

I really hate how "Direct3D feature levels" are synonymous to the API version. Feature levels can implemented across both DX11 and DX12 (in fact I understand some were even moved from 12 back to 11.3 or something like that?)

Anyway, from an API (not feature level) point of view the important change from DX11 to DX12 is that there's asynchronous calls now and there's separate queues (graphics/compute/copy) for which threads can enqueue tasks in a non-blocking manner. Note that this asynchronous API is not to be confused with the hardware async features I was talking about above.

The non-blocking nature of DX12 make the hardware-side async capability super-important because as mixed work arrives from the CPU it becomes important that the hardware be able to prioritise work (context switching) and exploit idle compute capacity (dynamic load balancing / hardware schedulers) which was not as important in DX11 where work was arriving at the rate which the DRIVER (not game) decides.

As far as Maxwell is concerned, I still believe (EDIT to the limits of my understanding) there will never be a meaningful DX12 driver.
 
Last edited:
TheRealDeal said:
Halo 5 is dx12 only like pretty much all other Microsoft games. These tend to favor AMD for obvious reason's being xbox orientated.
Click to expand...

This is the worst thing about all this. I really wish everyone would pick up Vulkan instead...

:mad:

I'm mad because Nvidia really wanted to get behind Vulkan, and they've been able to get away with just about anything in the gaming graphics world, but in this case they've got to go up against Microsoft...
 
akarypid said:
This is the worst thing about all this. I really wish everyone would pick up Vulkan instead...

:mad:

I'm mad because Nvidia really wanted to get behind Vulkan, and they've been able to get away with just about anything in the gaming graphics world, but in this case they've got to go up against Microsoft...
Click to expand...

It could mean nothing, but:

https://content.halocdn.com/media/D...c-sheet--dd86eb67b05a4fab813259e2df7ce4af.jpg

Shows only Nvidia cards on the poster. Of course it will support AMD cards and you just need to work out the equivalent Nvidia to AMD card, but was putting Nvidia cards on the poster a subtle nod that performance won't be terrible?

Or maybe I'm just mad.
 
akarypid said:
I really hate how "Direct3D feature levels" are synonymous to the API version. Feature levels can implemented across both DX11 and DX12 (in fact I understand some were even moved from 12 back to 11.3 or something like that?)

Anyway, from an API (not feature level) point of view the important change from DX11 to DX12 is that there's asynchronous calls now and there's separate queues (graphics/compute/copy) for which threads can enqueue tasks in a non-blocking manner. Note that this asynchronous API is not to be confused with the hardware async features I was talking about above.

The non-blocking nature of DX12 make the hardware-side async capability super-important because as mixed work arrives from the CPU it becomes important that the hardware be able to prioritise work (context switching) and exploit idle compute capacity (dynamic load balancing / hardware schedulers) which was not as important in DX11 where work was arriving at the rate which the DRIVER (not game) decides.

As far as Maxwell is concerned, I still believe (EDIT to the limits of my understanding) there will never be a meaningful DX12 driver.
Click to expand...

There is a lot of misunderstanding when it comes to Maxwell. At the end of the day though it isn't a great choice for DX12 whatever way you shake it - even with its capabilities fully exploited it will fall short of Pascal and future GCN hardware as even with "working" "async" it will always be at the mercy of the longest thread and fall down hard if developers don't pay attention to that.
 
southernorth said:
It could mean nothing, but:

https://content.halocdn.com/media/D...c-sheet--dd86eb67b05a4fab813259e2df7ce4af.jpg

Shows only Nvidia cards on the poster. Of course it will support AMD cards and you just need to work out the equivalent Nvidia to AMD card, but was putting Nvidia cards on the poster a subtle nod that performance won't be terrible?

Or maybe I'm just mad.
Click to expand...

My Vulkan preference was more from an "I hope I can finally play games in Linux instead of having to boot back into Windows" direction. At least for non-Microsoft games... :p
 
Guys, I would be really interested to see the benchmark performance average of a 480 using the big games from 2016 only versus the same with the 1060.

Unfortunately, a lot of sites, especially TechPowerUp, have a bench suite of games that heavily favour Nvidia GPUs (W1zzard has BF3 and BF4, Crysis 3 but no Doom, Warhammer or other notable big PC games released this year). I reckon if both cards were tested with newer games only, performance average would be very close.

As we know, DOOM Vulkan, Forza Apex, Warhammer: TW and now Deus Ex are faster on the 480. All games released in the last few months or this year at least.
 
Still trying to decide between the 1060 and RX 480. Currently leaning towards either the EVGA 1060 SC or Zotac 1060 Amp as these seem to have a good reputation here.

But I wonder how one of the AIB RX 480s at around the same price point (£260) would compare in terms of running cool and quiet. Which of the AIB RX 480s runs the quietest and coolest?

I know for many people these aren't important considerations, but for me they are. My PC has to contend with a rather high ambient temperature and my build is focused on keeping fan noise to a minimum where possible.

I've settled on £260 or thereabouts as my budget cap because any higher than this and I should be considering the 1070. So this tends to rule out Asus and MSI.
 
You must log in or register to reply here.
Back
Top Bottom