Fable Legends DX12 Benchmarked

[the54thvoid]

Here:

http://www.anandtech.com/show/9659/fable-legends-directx-12-benchmark-analysis/2

Interesting beta preview.

Summary - as predicted - AoS bench was way too early to call any future DX12 outcome. It's a healthy mixed bag. Oh, 3rd graph down 290x beats the Fury X and a 970 ties a 980ti

 

[LightningJR]

3rd graph down 290x beats the Fury X and a 970 ties a 980ti

Low graphics and 720p, not too relevant imo. Nice to see a 750ti getting 85fps though.

 

[rtwjunkie]

It's good to see that where the GPU matters, 1080P and above that AMD and Nvidia are both represented at the top, instead of just one brand making a sweep!

I think if the 380 and the 980 had been represented, they would be 3rd and 4th place.

 

[FordGT90Concept]

It means Fable Legends isn't making heavy use of async compute. If you think about the type of game AoS is, it makes sense it would benefit more from async compute than a game like Fable Legends.

The game appears to be very graphically demanding.

 

[the54thvoid]

It means Fable Legends isn't making heavy use of async compute. If you think about the type of game AoS is, it makes sense it would benefit more from async compute than a game like Fable Legends.

The game appears to be very graphically demanding.

Which is the whole point of the 'nonsense' around the DX12 sucks for Nvidia myth. Each game with each graphical requirement will lean towards various elements of the API.
I'm not a Fable person though, my horizon has Deus Ex on it. Although it's got some AMD only stuff (Tress FX) I think it'll be a good environment for both red and green.

 

[Ahhzz]

This one's only going to be supported on Win10?

 

[FordGT90Concept]

Yes but I'm sure if you sail the high seas, they'll find a way to fix that.

 

[crazyeyesreaper]

Yet OC3D had the exact opposite results so who to believe?
http://www.overclock3d.net/articles/gpu_displays/fable_legends_dx12_benchmark_-_amd_vs_nvidia/1

 

[TheMailMan78]

Why is the 970 almost double the 680 at 1080p? Am I missing something?

As usual Ill wait for W1zz.

 

[crazyeyesreaper]

Pretty much Mailman, looking at other sites I am seeing performance swing from 3-4% in favor of Nvidia to AMD / Nvidia being equal to nearly 20% in favor of Nvidia. So numbers are all over the place right now.

 

[rtwjunkie]

Two Identical systems (other than one has k cpu and the other is non k) in my house, and the 970 one averages nearly 10 fps more than the 780, so yeah, I can believe the benchmark above that has a 970 doing 20 fps more than a 680. And our 970 is the one paired with the non-k 3770.

 

[W1zzard]

As usual Ill wait for W1zz.

I'm not involved in any Fable pre-release benchmarking.

Once the game is released, and it doesn't suck completely, it'll get added to our benchmarks.

 

[jigar2speed]

I am very proud of my HD 7970 GHZ edition, the results are for normal HD 7970 so i am pretty sure my OCed card should hit 50 FPS at Ultra setting while playing FABLE.

 

[TheMailMan78]

Two Identical systems (other than one has k cpu and the other is non k) in my house, and the 970 one averages nearly 10 fps more than the 780, so yeah, I can believe the benchmark above that has a 970 doing 20 fps more than a 680. And our 970 is the one paired with the non-k 3770.

I'm not buying a 970 is almost double the performance of a 680.

 

[rtwjunkie]

I'm not buying a 970 is almost double the performance of a 680.

Well, 20 frames, not 40...40 would be double. But I can only tell you how well it performs in my own experience compared to a 780, which is excellent! I can't account for anyone else's experience.

 

[Moofachuka]

I am very proud of my HD 7970 GHZ edition, the results are for normal HD 7970 so i am pretty sure my OCed card should hit 50 FPS at Ultra setting while playing FABLE.

+1 still rocking my dual 7970. Best purchase so far... was actually debating between 7970 and 680 at that time, and now the 7970's are aging so well.

 

[m&m's]

Why is the 970 almost double the 680 at 1080p? Am I missing something?

I'm not buying a 970 is almost double the performance of a 680.

Adjust your goggles old man cuz your eyes need a fix. 39.68 vs 59.70, it's more like +50% than 2x.
When the GTX 970 was brand new, it already was +30% faster @1080P than a 680.
So who knows if DX12 doesn't bring something that makes the gap between 30% and 50%.

 

[medi01]

Yet OC3D had the exact opposite results so who to believe?
http://www.overclock3d.net/articles/gpu_displays/fable_legends_dx12_benchmark_-_amd_vs_nvidia/1

Anand "for some reason" didn't use the latest AMD drivers.


PS
Isn't Fable on Unreal which is, well, kinda sponsored by nVidia? (not that previous DX12 comparison was legit either though)

 

[rtwjunkie]

Anand "for some reason" didn't use the latest AMD drivers.


PS
Isn't Fable on Unreal which is, well, kinda sponsored by nVidia? (not that previous DX12 comparison was legit either though)

Hmmmmm...so presumably a few frames better would be the result for the AMD cards? Is that what you're saying? If they are better drivers, then it makes a difference.

 

[Xzibit]

The test seem to indicate that the 380, 390 & 390X are a much better buys for DX12.

Both ins AoS and FL

 

[the54thvoid]

PCPER have it too.

And FTR, OC3D didn't run the test - it was sourced from Extremetech who re ran it and got this:

They re ran it because they realised they had a glitch due to some power play from Win 10 affecting the result.

The test seem to indicate that the 380, 390 & 390X are a much better buys for DX12.

Both ins AoS and FL

The power of GCN for you. the 390X is doing especially well.

 

[Xzibit]

The power of GCN for you. the 390X is doing especially well.

And this is based on UE engine. Something Nvidia is suppose to be superior and the latest build said AsynComp was just starting experimental phase. Who know what this FL build is based on.

 

[Lionheart]

Funny how the 390X is a refreshed / rebranded GPU & its not that far behind the Fury X performance wise. I feel like the Fury X is being held back, driver optimization or ROP starved maybe???

 

[the54thvoid]

And this is based on UE engine. Something Nvidia is suppose to be superior and the latest build said AsynComp was just starting experimental phase. Who know what this FL build is based on.

Everybody should remember that Fiji is a credible architecture built on from Hawaii, held back by AMD's lesser software optimising under DX11. It has more transistors, more bandwidth, more TFlops and more cores than GM200 - its only deficit is ROPS. Under DX12 and the bare metal cliche, Both AMD and Nvidia will be trading blows regularly, game dependent with no outright winner. On paper - Fiji should do better.

 

[arbiter]

It means Fable Legends isn't making heavy use of async compute. If you think about the type of game AoS is, it makes sense it would benefit more from async compute than a game like Fable Legends.

The game appears to be very graphically demanding.

No its a game that doesn't use tech from 1 side to tilt the table in 1 sides favor, Async was original AMD made so using Async tilts the field in AMD"s favor its like how people complain physx did it. I know someone will say that async is a standard but that was only made a standard recently and before that it was amd locked tech.

Funny how the 390X is a refreshed / rebranded GPU & its not that far behind the Fury X performance wise. I feel like the Fury X is being held back, driver optimization or ROP starved maybe???

Very possible it is, remember original PR benchmarks AMD released turned anything off that wasn't shader based to show the cards power. Some of those they turned off uses the rops so could be very possible that 64 ROPS which 390x has as well is holding it back some. 980ti has 50% more of them with 96.