Why does the GTX 1060 perform better than the rx 480 if the 480 has more tflops

[lightning247S]

Is it just Optimization or something else.

 

[RuLeZ]

Tflops doesnt necessarily means performance, it just means the raw power of calculations of the GPU, you can even see between nvidia cards that Tflops doesnt equal performance, like GTX 780 Ti ( 5,040 GFLOPS ) and GTX 970 ( 3,500Gflops) and they have very similar performance in real game scenarios.

[don_svetlio]

Cause DX11? Look at DX12/Vulkan and the 480 is 10% faster.

[arnavvr]

8 minutes ago, don_svetlio said:

Cause DX11? Look at DX12/Vulkan and the 480 is 10% faster.

This.

[Prysin]

14 minutes ago, lightning247S said:

Is it just Optimization or something else.

 

It is because AMD has very very poorly optimized DX11 and OpenGL drivers.

UNder DX12 and Vulcan, the 480 catches up to and or beats the 1060.

[Jon Jon]

It's why I am hyped up for DX12.

[porina]

Flops is more for compute, and does not equal graphical performance.

[handymanshandle]

Smoke and Mirrors and Nvidia potentially buying reviews or paying reviewers to not include DX12/Vulkan. Though Hardware Unboxed did show Vulkan performance of the 480, which handily beat the 1060 in Doom. Though some might say that id might've optimized Vulkan for the 480.

[App4that]

12 minutes ago, wcreek said:

Smoke and Mirrors and Nvidia potentially buying reviews or paying reviewers to not include DX12/Vulkan. Though Hardware Unboxed did show Vulkan performance of the 480, which handily beat the 1060 in Doom. Though some might say that id might've optimized Vulkan for the 480.

Damn man, don't be afraid to show a little bias once in awhile. It's OK. /s

 

Async is still locked on Nvidia cards with Doom, so once that is fixed the 1060 will show a gain.

[othertomperson]

22 minutes ago, Prysin said:

It is because AMD has very very poorly optimized DX11 and OpenGL drivers.

UNder DX12 and Vulcan, the 480 catches up to and or beats the 1060.

Most Dx12 benchmarks I've seen have been within two or three fps of eachother. I wouldn't say that one necessarily "beats" the other.

 

What it does mean is that TFLOPS is a worthless measure of performance. It is a fair comparison between two cards of the same or similar architecture, but otherwise it's fairly meaningless.

[Prysin]

2 minutes ago, othertomperson said:

Most Dx12 benchmarks I've seen have been within two or three fps of eachother. I wouldn't say that one necessarily "beats" the other.

 

What it does mean is that TFLOPS is a worthless measure of performance. It is a fair comparison between two cards of the same or similar architecture, but otherwise it's fairly meaningless.

It IS an indicator of performance. However AMDs cards are so massively underutilized that THEIR TFLOPS reading isnt a metric for performance.

In terms of Nvidia cards, they are pretty much spot on, give or take 10%. AMD can miss with as much as 30%.

 

evidence:

980Ti vs R9 390X in Ashes DX12.

Theoretically, both have the same TFLOPs rating, more or less. Under DX11, it is clear that the R9 390X is underutilized. Under DX12 it rapidly catches up to the 980Ti.

 

This is a known and proven fact.

[othertomperson]

Just now, Prysin said:

It IS an indicator of performance. However AMDs cards are so massively underutilized that THEIR TFLOPS reading isnt a metric for performance.

In terms of Nvidia cards, they are pretty much spot on, give or take 10%. AMD can miss with as much as 30%.

 

evidence:

980Ti vs R9 390X in Ashes DX12.

Theoretically, both have the same TFLOPs rating, more or less. Under DX11, it is clear that the R9 390X is underutilized. Under DX12 it rapidly catches up to the 980Ti.

 

This is a known and proven fact.

In which case in a Dx12 game the one with the most TFLOPs should automatically win. Except it doesn't. It's only Shaders * 2 * Clock Speed. It's a very blunt instrument.

[BuckGup]

13 hours ago, don_svetlio said:

Cause DX11? Look at DX12/Vulkan and the 480 is 10% faster.

Vulkan it is far more then 10%. 30 FPS increases is a large number

[Prysin]

Just now, othertomperson said:

In which case in a Dx12 game the one with the most TFLOPs should automatically win. Except it doesn't. It's only Shaders * 2 * Clock Speed. It's a very blunt instrument.

well, its sort of like comparing cars 0-62mph by amount of horsepower. Sure the "more" horsepower should win. But then there is grip, weight, gear ratio, gear change, torque...

[Brian McKee]

15 minutes ago, App4that said:

Damn man, don't be afraid to show a little bias once in awhile. It's OK. /s

 

Async is still locked on Nvidia cards with Doom, so once that is fixed the 1060 will show a gain.

Nvidia already shows gains with Vulkan on all of their cards. There is zero reason for a user to use open GL with doom unless you're using an old card which is a moot point because we're talking about the 1060...

[App4that]

5 minutes ago, Prysin said:

It IS an indicator of performance. However AMDs cards are so massively underutilized that THEIR TFLOPS reading isnt a metric for performance.

In terms of Nvidia cards, they are pretty much spot on, give or take 10%. AMD can miss with as much as 30%.

 

evidence:

980Ti vs R9 390X in Ashes DX12.

Theoretically, both have the same TFLOPs rating, more or less. Under DX11, it is clear that the R9 390X is underutilized. Under DX12 it rapidly catches up to the 980Ti.

 

This is a known and proven fact.

Only that's old news. Things change

 

 

1 minute ago, SamStrecker said:

Vulkan it is far more then 10%. 30 FPS increases is a large number

More bias I see. Can I make up numbers too?

 

 

[othertomperson]

1 minute ago, SamStrecker said:

Vulkan it is far more then 10%. 30 FPS increases is a large number

Vulkan has a sample size of 1. You can't judge an API from a single game. Would you assert that the RX 480 was crap in Dx12 because of Rise of the Tomb Raider?

 

2 minutes ago, Prysin said:

well, its sort of like comparing cars 0-62mph by amount of horsepower. Sure the "more" horsepower should win. But then there is grip, weight, gear ratio, gear change, torque...

Like I said, it's a crap measure of performance across different architectures and different brands.

[App4that]

Just now, Brian McKee said:

Nvidia already shows gains with Vulkan on all of their cards. There is zero reason for a user to use open GL with doom unless you're using an old card...

What? Please read my post next time. I'm talking about Vulkan.

[BuckGup]

12 hours ago, othertomperson said:

Vulkan has a sample size of 1. You can't judge an API from a single game. Would you assert that the RX 480 was crap in Dx12 because of Rise of the Tomb Raider?

 

Like I said, it's a crap measure of performance across different architectures and different brands.

Well considering see it's performance on Linux in other applications doing better then DX12 on windows. I can safely say it Vulkan becomes mainstream in every game there is no need to use windows anymore

[Brian McKee]

Just now, App4that said:

What? Please read my post next time. I'm talking about Vulkan.

So am I?

 

 

[App4that]

2 minutes ago, othertomperson said:

Vulkan has a sample size of 1. You can't judge an API from a single game. Would you assert that the RX 480 was crap in Dx12 because of Rise of the Tomb Raider?

 

Like I said, it's a crap measure of performance across different architectures and different brands.

But it tells them what they want to hear and they don't even have to listen.

 

Doom shows a larger gain for AMD because AMD wasn't using the same version of OpenGL and was no where near the performance of Nvidia's cards.

[BuckGup]

13 hours ago, App4that said:

Only that's old news. Things change

 

 

More bias I see. Can I make up numbers too?

 

 

How is a fact bias? In Doom the RX480 preformed on average 30 FPS better. There isn't any loopholes or lying. Vulkan is much better then DX12. It is just more work for devs

[App4that]

1 minute ago, Brian McKee said:

So am I?

 

 

Then why did you bring up OpenGL?

 

Just now, SamStrecker said:

How is a fact bias? In Doom the RX480 preformed on average 30 FPS better. There isn't any loopholes or lying. Vulkan is much better then DX12. It is just more work for devs

Where? Where is your evidence that the 480 beats the 1060 by 30fps? Please don't bring biased evidence ether.

[BuckGup]

12 hours ago, App4that said:

Then why did you bring up OpenGL?

 

Where? Where is your evidence that the 480 beats the 1060 by 30fps? Please don't bring biased evidence ether.

I did not say the 480 beats the 1060 by 30 FPS. I said the 480 gets 30 more FPS when using Vulkan rather then OpenGL. Don't be so venomous. 

[App4that]

Just now, SamStrecker said:

I did not say the 480 beats the 1060 by 30 FPS. I said the 480 gets 30 more FPS when using Vulkan rather then OpenGL. Don't be so venomous. 

My bad, my bad. Misunderstood.