Overclock.net banner

480 vs 1060 - ashes of the singularity, why is the 1060 and 480 tying each other?

10K views 229 replies 27 participants last post by  rdr09 
#1 ·
all the benchmarks i've seen, like this guru3d one, show the 1060 and 480 neck to neck in ashes. i thought ashes heavily used async? shouldn't the 480 be smoking it?

all the benchmarks i've seen clearly show every single amd card receiving a higher boost with async than any pascal card. in time spy the 480 still receives higher gains with async than the 1060 and there is a lot of controversy over its handling of async in that benchmark to paint nvidia in a better light.

even tomb raider, one all amd cards fail in, even in dx12, still recieve higher gains than pascal's cards, but of course nvidia still wins since its a major gameworks title with heavy use of tessellation.

whats going on with ashes?
 
#2 ·
1. NV did improve performance in Ashes

2. 1060 is 10 or a bit more % faster than 480 overall

3. reference 480 in tests is throttling, probably below 1200MHz in Ashes as GPU is fully utilized (not bottlenecked by CPU)

Beside all that, reference 480 still wins in Ashes, by a small margin tho
 
#3 ·
Quote:
Originally Posted by Ha-Nocri View Post

1. NV did improve performance in Ashes

2. 1060 is 10 or a bit more % faster than 480 overall

3. reference 480 in tests is throttling, probably below 1200MHz in Ashes as GPU is fully utilized (not bottlenecked by CPU)

Beside all that, reference 480 still wins in Ashes, by a small margin tho
yeah but their async performance improvements are so far showing very small gains. ~4% average it seems. while amd.... you get results like doom where they use async pretty heavily and it takes a 480 from 60fps all the way up to 120fps. but then you get ashes, which from what i read uses async even more than doom did and it barely leads the 1060... by what? 1 - 4fps?

i know amd's opengl performance isn't all that great but those kinds of gains in doom are not just overhead issues being cleared up. its async.
 
#4 ·
Actually, in Doom only a potrion of extra performance is from Async, most is from Vulkan itself.
 
#6 ·
Quote:
Originally Posted by muffins View Post

according to id, async is the biggest player in the performance increases.
Rly? Cause I saw some tests with async off (using MSAA or FXAA) and AMD cards performed much better than under OpenGL
 
#7 ·
Quote:
Originally Posted by Ha-Nocri View Post

Rly? Cause I saw some tests with async off (using MSAA or FXAA) and AMD cards performed much better than under OpenGL
after watching that video it seems like turning on async adds a similar percentage boost as just using vulkan over opengl. i wouldn't say just a portion, i would say half the performance increase comes from async and the other half from vulkan.

and anways this is taking this thread off topic. what's going on with ashes?

i was reading this article from extremtech about doom and vulkan release and they make an interesting note:
Quote:
The RX 480 is just one GPU, and we've already discussed how different cards can see very different levels of performance improvement depending on the game in question - the R9 Nano picks up 12% additional performance from enabling versus disabling async compute in Ashes of the Singularity, whereas the RX 480 only sees a 3% performance uplift from the same feature.
 
#8 ·
Vulkan did definitely add more than Async.

But anyway, 480 is throttling more under DX12/Vulkan as it is being utilized more (less CPU bound). So gains aren't as high as with other AMD cards. Also, 480 has somewhat better minimum frame-rate in DX11/OpenGL as Polaris is improved in regards to CPU overhead.
 
#9 ·
Quote:
Originally Posted by Ha-Nocri View Post

Vulkan did definitely add more than Async.

But anyway, 480 is throttling more under DX12/Vulkan as it is being utilized more (less CPU bound). So gains aren't as high as with other AMD cards. Also, 480 has somewhat better minimum frame-rate in DX11/OpenGL as Polaris is improved in regards to CPU overhead.
not really. going from, like a clip i saw in that video, 98 to 125 to 137 are very similar gains. almost half of each other.

i can understand throttling but like that extremetech stated, 3%? even in time spy the gains are larger, 8.5%. doom vulkan is about 10 - 13%.
 
#10 ·
Do I rly have to do this?!





50-50 is like the best case scenario for Async.

Anyway, I explained everything in this thread, you can accept it or keep arguing. I'm out.
 
#11 ·
Quote:
Originally Posted by Ha-Nocri View Post

Do I rly have to do this?!





50-50 is like the best case scenario for Async.

Anyway, I explained everything in this thread, you can accept it or keep arguing. I'm out.
Well I can pick screenshots too

Either way async gives it a big boost as well and it's not just a small portion. Clearly depends on how the scene. And more importantly, a bigger boost than what it appears in ashes.

If extremetech is right, a 3% something has to be going on, which is the whole point of this thread. Not to steer off course into another game.

Would be nice if there were benchmarks that benched the 480 in ashes with async on and off.

Maybe be possible someone with a 480 could do some testing?
 
#12 ·
The bigger problem is no availability.. All the benches in the world wont matter if you cant buy either of them. well unless you want to pay $300+ for a 480. The limited availability of the 480 shot amd in its own foot.

I think you'll see large gains out of nvidia with maturing drivers, especially since they are more software based, the 1060 is less than a week old, so we'll see.. I think nvidias driver development is miles ahead of amd. The 1060 also overclocks better, runs cooler and uses less power. What I find interesting the the power usage, amd seems to be behind the curve by a wide margin.

Any benchmarks I've seen of the 480 and 1060, the 1060 is just faster on a wider range of titles.. The only games I really see the 480 edge out the 1060 were in 2 titles, and both were amd backed titles and it was a slim margin, like a couple fps. I also dont completely believe amd and its async hype, it may be better and thats fine if it is, but I dont think its that much better, and not so sure hardware based is any better than software based, again I think we'll have to wait and see, I think software based is more flexible... I find it interesting each company went a different route.

I think nvidia knows every move amd is making and they arent really worried about amd. I think thats exactly why we seen nvidias 1070's and 1080's first, nvidia knew amd was going with a lower tier card and nvidia knew they already had a card designed to compete with it.. I hope amd produces the 490, if thats what they are going to call it, and it falls right between the 1070 and 1080 and costs the same or less than the 1070.. That would be something for AMD. say 10% faster than the 1070 and and around $350-$375..

The aib cards might help amd out, I hope they do, we need the competition..

I own a 480 and a 1070, bought them a day apart.. I can tell you from experience amd needs a to work on their drivers and software package. I literally think amd let the consumer be their beta testers.
 
#13 ·
Async does not add much. At most we will see over 10% but NEVER the 47% increases (calculated by AMD as best cases). There is more to AMD's gains than just Async...

For myself, neither the 1060 nor 480 are really what I am looking for. To be fair, even the 1080s arent a big enough increase in performance for me to be happy. So it is waiting.

However for my friends, it is the RX 480 that they want. Because it will outlive the 1060 and we here are used to upgrading GPUs ever 4-5-6 years. The 1060 will not survive that well.
 
#14 ·
I dont think the 480 will outlive the 1060, one reason is the heat it generates on the reference models...The 1060 also runs much cooler so its going to suffer from less thermal degradation.

I have yet to see how the 1060 overclocks compared to the 480. I know the 1060 is reportedly going over 2000mhz, my 480 with a heavily modified cooler will hit 1400 and thats getting close to its limit.

Also, as resolutions increase and become more affordable it will render these lower tier cards useless, and thats likely within a years time.. Thanksgiving/Christmas time 1440p monitors will be cheap..

I upgraded to a 32" 1440p monitor on my main computer, so I had no need to purchase neither the 480 or 1060, and I went with the 1070 which is perfect for 1440p, actually its a little overkill.. From what I've seen so far the 1060 does better than the 480 at 1440p.
 
#15 ·
Its been shown that Nvidia cheats benchmarks by disabling or reducing visual effects on a driver level for DX12, Vulkan, and even DX11. Enough of an answer?
 
#17 ·
Quote:
Originally Posted by danjal View Post

I think amd is doing the same in its titles, thats why you dont get consistent results in Ashes..
Certainly possible, although no evidence has ever been shown, unlike Nvidia where very visible results can be seen in side by side screenshots of certain titles.
 
#18 ·
Quote:
Originally Posted by Zero4549 View Post

Certainly possible, although no evidence has ever been shown, unlike Nvidia where very visible results can be seen in side by side screenshots of certain titles.
When amd released the 480, the video of it running ashes was suspicious in itself, and amd jumped up with a blatant lie about that within 24hrs, also adding, I find it odd that the ashes benchmark never renders the same.. Why is that? Is it adjusting for select gpus? It should never be used as a baseline benchmark ever, since it never renders the same textures and such in a row..

As devious as amd has been lately, with powergate and them trying to kill older cards, I would bet they are doing the same.
 
#19 ·
Quote:
Originally Posted by danjal View Post

When amd released the 480, the video of it running ashes was suspicious in itself, and amd jumped up with a blatant lie about that within 24hrs, also adding, I find it odd that the ashes benchmark never renders the same.. Why is that? Is it adjusting for select gpus? It should never be used as a baseline benchmark ever, since it never renders the same textures and such in a row..

As devious as amd has been lately, with powergate and them trying to kill older cards, I would bet they are doing the same.
What devious things have AMD done recently? The PCIE issue was fixed VIA driver and can be fixed through BIOS mod found on this forum.
 
#20 ·
Quote:
Originally Posted by Zero4549 View Post

Certainly possible, although no evidence has ever been shown, unlike Nvidia where very visible results can be seen in side by side screenshots of certain titles.
Do you really think AMD is going to let Nvidia get away with cheating on image quality?

If there was something that could be proven, I guarantee AMD would be all over it just like they have in the past.
 
#21 ·
Quote:
Originally Posted by mcg75 View Post

Do you really think AMD is going to let Nvidia get away with cheating on image quality?

If there was something that could be proven, I guarantee AMD would be all over it just like they have in the past.
I think thats exactly why amd jumped so quick with the ashes controversy, and also powergate
 
#22 ·
Quote:
Originally Posted by Worldwin View Post

What devious things have AMD done recently? The PCIE issue was fixed VIA driver and can be fixed through BIOS mod found on this forum.
I stated the devious things in the post you responded too.

I owned a 290 and amd tried to kill it with a driver update, they set default fan speed at minimum and target temp at maximum, not once, but with two driver updates... The power issue of the 480, amd wouldnt have done anything except let people burn up their motherboards if nothing was found by independent testing.. Why didnt amd know of the power issue prior to release? Are they not testing their own gpu's? Are they just letting the consumer do the testing? Thats exactly what I think amd is doing to cut costs.
 
#23 ·
Oh and before anyone mentions the GeForce 196.75 driver scandal with the fan control issue (locking fan to minimum speed), it already happened... AMD should have learnt from nVidia's mistake and screen their official (not Beta) drivers properly.
 
#24 ·
OP, 200$ vs 300$, equal performance, 480 IS smoking it
smile.gif
 
#25 ·
Quote:
Originally Posted by Desolutional View Post

Oh and before anyone mentions the GeForce 196.75 driver scandal with the fan control issue (locking fan to minimum speed), it already happened... AMD should have learnt from nVidia's mistake and screen their official (not Beta) drivers properly.
Don't forget 320.18. I had a busy week back then.
 
#26 ·
Quote:
Originally Posted by danjal View Post

I stated the devious things in the post you responded too.

I owned a 290 and amd tried to kill it with a driver update, they set default fan speed at minimum and target temp at maximum, not once, but with two driver updates... The power issue of the 480, amd wouldnt have done anything except let people burn up their motherboards if nothing was found by independent testing.. Why didnt amd know of the power issue prior to release? Are they not testing their own gpu's? Are they just letting the consumer do the testing? Thats exactly what I think amd is doing to cut costs.
The power issue claim as far as we know didn't consume any motherboards and it was sorted as soon as it was brought to their attention regardless of it was actually an issue or not since the guys that drafted the spec themselves said it was alright.

As for the performance of the 1060, oxide creates a specific code path for each IHV which is arguably the greatest thing about DX12 and pascal has premption which can help with async loads. As far as my eyes can tell, there was no difference in the quality of the textures on both cards but just Nvidia cards rendering less snow. Whether this offloads work off Nvidia cards, nobody actually knows and none of the review sites that reported on this at first have posted a follow up after bringing it up with Nvidia. If they ever did.
 
  • Rep+
Reactions: LAKEINTEL
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top