Total War: Warhammer DirectX 12 benchmarks

[sontin]

Wow what is with the stuttering while panning on the 1070, that looks awful. Both Fury X (even while running closer to 34 fps vs 40s) and 1080 are smooth, but the 1070 stutters a lot.

Broken DX12 path and video recording. It is just another premium job by a AMD's Gaming Evolved partner. :thumbsup:

A few bits about the DX12 performance on Maxwell: http://www.dsogaming.com/news/report-total-war-warhammer-runs-27-slower-dx12-nvidias-hardware/

 

[24601]

Broken DX12 path and video recording. It is just another premium job by a AMD's Gaming Evolved partner. :thumbsup:

A few bits about the DX12 performance on Maxwell: http://www.dsogaming.com/news/report-total-war-warhammer-runs-27-slower-dx12-nvidias-hardware/

Expecting an AMD Gaming Evolved partner to do the low level optimization required to beat Nvidia's driver team at their job is utter lunacy.

 

[Bacon1]

Broken DX12 path and video recording. It is just another premium job by a AMD's Gaming Evolved partner. :thumbsup:

A few bits about the DX12 performance on Maxwell: http://www.dsogaming.com/news/report-total-war-warhammer-runs-27-slower-dx12-nvidias-hardware/

Really it just happens to be broken for a single card and not the 1080?

 

[Yakk]

The blood and gore download is cool!

BTW, the Blood & Gore DLC may be causing performance issues, so expect contradicting benchmarks about DX11 & DX12 from people to do not realize they have it on.

 

[sontin]

Really it just happens to be broken for a single card and not the 1080?

It is not as broken as it on Maxwell: https://www.youtube.com/watch?v=s9dIGqnYK_A

Still, 20% performance decrease is impressive.

 

[Bacon1]

http://www.computerbase.de/2016-03/rise-of-the-tomb-raider-directx-12-benchmark/2/

45FPS with a Fury X in the second test. This is a CPU hearvy scene. The GTX980TI is 63% faster and loses no performance with DX12.

But i guess this is AMD's experience with a low level API. :\

http://forums.anandtech.com/showpost.php?p=38093051&postcount=85

It is not as broken as it on Maxwell: https://www.youtube.com/watch?v=s9dIGqnYK_A

Still, 20% performance decrease is impressive.

Guess that's Nvidia's experience with a low level API. :\

 

[DarkKnightDude]

Its definitely boosted performance for my Fury X. Battles are now even smoother, even on Ultra size. I could swear the game looks better, or maybe its placebo.

 

[sontin]

I see no problem with it. Shows that low level APIs are not designed for multiple architectures. Something people have tried to communicate here.

/edit: I think it is an archivement that DX12 doesnt even provide any CPU benefits in this game. There doesnt exist any advantage of DX12 besides AMD is able to get rid of their inferior dx11 driver.

 

[itsmydamnation]

I see no problem with it. Shows that low level APIs are not designed for multiple architectures. Something people have tried to communicate here.

Maybe NV DX12 drivers just suck, no that couldn't possible be it . Everyone know's DX12 is ultra low level and doesn't even have drivers .

AMD game issues == amd driver team sucks
NV games issues != NV driver team suck under any circumstances

 

[Zstream]

I think its kinda pointless to argue anything further, its pretty much over at this point. Have only 200 bucks to spend get a 480, if you have more then buy a 1070 or 1080, there's literally nothing left to discuss on these forums.

Pretty much this - the only caveat to that is that I have plenty of money but refuse to spend more

 

[3DVagabond]

That's very common with CPU bottleneck at low resolution.

Crossfire has additional CPU use. If the game is already bottlenecked it would get worse with the additional Crossfire overhead.

 

[3DVagabond]

Why do reviewers still test the 290X instead of the 390X?

So they can use cards that throttle and under perform. Using the 390/X would just mean more red scores near the top of the charts.

 

[3DVagabond]

Only in the internal benchmark. Ingame Fury X has no chance against a broken DX12 path on Pascal:
https://www.youtube.com/watch?v=uxnFv8cnjZA

So the vid you referenced shows all of the cards being markedly slower. Except for the 970 which is comparable to Guru3D's (running mid 40's). (Never mind. Misread and was comparing 970 instead of 1070.) The 1080 loses ~30%. The Fury X drops from 68fps to ~31fps though. lol >50% loss in performance. I don't think these two tests can really be compared. I guess it comes down to who do you believe, Guru3D or DudeRandom84.

Interesting, so Geforce GTX 1070 does much better than Fury X in-game.

If you believe that I've got some land in Florida.

 

[Flapdrol1337]

Strange to see the 8350 so close to the 2500K, but both being beat by a haswell i3.