Warhammer Total War Bench Thread

Radox-0 · 29 Sep 2016 at 16:29

Wow, did not expect the Nano under water, so running higher then Fury X clocks to be quiet so behind the GTX 1060 in the title. Did not see the CPU being a boundary either,, but possibly a reason.

Kaapstad · 29 Sep 2016 at 16:44

Radox-0 said:
Wow, did not expect the Nano under water, so running higher then Fury X clocks to be quiet so behind the GTX 1060 in the title. Did not see the CPU being a boundary either,, but possibly a reason.

The RX480 is also quite a bit behind, I think it could be that AMD need to work on the drivers for this game.

TheRealDeal · 29 Sep 2016 at 17:02

Radox-0 said:
Wow, did not expect the Nano under water, so running higher then Fury X clocks to be quiet so behind the GTX 1060 in the title. Did not see the CPU being a boundary either,, but possibly a reason.

It shouldn't be as all other sites show the AMD cards doing well with Nano way faster than the gtx1060.

Has there been an update as AMD usually see a nice boost in this game with dx12 but not in this thread. Link below shows what Dx12 should be doing for both sides. Without that boost AMD cards are way below NV cards like this thread is showing. Nvidia are seeing the drop off in dx12 as has always been present but AMD are seeing nothing almost like dx12 is not running.

http://www.techspot.com/review/1221-amd-radeon-rx-460/page2.html

Kaapstad · 30 Sep 2016 at 08:06

TheRealDeal said:
It shouldn't be as all other sites show the AMD cards doing well with Nano way faster than the gtx1060.

Has there been an update as AMD usually see a nice boost in this game with dx12 but not in this thread. Link below shows what Dx12 should be doing for both sides. Without that boost AMD cards are way below NV cards like this thread is showing. Nvidia are seeing the drop off in dx12 as has always been present but AMD are seeing nothing almost like dx12 is not running.

http://www.techspot.com/review/1221-amd-radeon-rx-460/page2.html

Your link highlights exactly what review sites do wrong. They are not using maximum settings which means the CPU becomes a bottleneck which in turn means the cards are a lot closer together than they should be.

The difference in performance between DX11 and DX12 on the 1060 in your link is quite small, in this bench thread it is much bigger as there is no CPU bottleneck to skew the figures.

TheRealDeal · 30 Sep 2016 at 09:22

Kaapstad said:
Your link highlights exactly what review sites do wrong. They are not using maximum settings which means the CPU becomes a bottleneck which in turn means the cards are a lot closer together than they should be.

The difference in performance between DX11 and DX12 on the 1060 in your link is quite small, in this bench thread it is much bigger as there is no CPU bottleneck to skew the figures.

It's not that either as Guru are running max detail at 4k and the Amd cards still show up pretty well. There is no cpu bottleneck here or at least it's comparable to 1080p max with 8xaa.

http://www.guru3d.com/articles_pages/zotac_geforce_gtx_1080_arcticstorm_review,17.html

subbytna · 30 Sep 2016 at 09:39

I take ti you need the game to run the bench yes?

Radox-0 · 30 Sep 2016 at 10:05

subbytna said:
I take ti you need the game to run the bench yes?

Yup, but it is an awesome game in its own right if Total War is your thing.

TheRealDeal said:
It's not that either as Guru are running max detail at 4k and the Amd cards still show up pretty well. There is no cpu bottleneck here or at least it's comparable to 1080p max with 8xaa.

http://www.guru3d.com/articles_pages/zotac_geforce_gtx_1080_arcticstorm_review,17.html

Hmmm, will take another look at running the results. Could just be the CPU itself.

TheRealDeal · 30 Sep 2016 at 10:17

Radox-0 said:
Yup, but it is an awesome game in its own right if Total War is your thing.

Hmmm, will take another look at running the results. Could just be the CPU itself.

Do you have the Dx12 patch installed? At one point it was not an auto update but i think it is now.

Radox-0 · 30 Sep 2016 at 10:21

TheRealDeal said:
Do you have the Dx12 patch installed? At one point it was not an auto update but i think it is now.

Yep, DX12 patch installed. Results I provided were both in DX11 and DX12 with DX12 run being a few fps behind.

TheRealDeal · 30 Sep 2016 at 10:24

Radox-0 said:
Yep, DX12 patch installed. Results I provided were both in DX11 and DX12 with DX12 run being a few fps behind.

It's a mystery to me. Everywhere shows a huge boost in dx12 for AMD with differing systems. I can only think a recent patch has screwed this up or msaa is really tanking performance.

Even here if Nvidia were running in dx11 they are only around 15% faster when matching the cards. Kapp's OC 1060 is over 50% faster than a RX480 OC.

https://www.techpowerup.com/reviews/MSI/GTX_1070_Gaming_Z/23.html

http://cdn.videocardz.com/1/2016/09/10_TOTALWAR.png

Kaapstad · 30 Sep 2016 at 10:42

TheRealDeal said:
It's not that either as Guru are running max detail at 4k and the Amd cards still show up pretty well. There is no cpu bottleneck here or at least it's comparable to 1080p max with 8xaa.

http://www.guru3d.com/articles_pages/zotac_geforce_gtx_1080_arcticstorm_review,17.html

Nope they are not using max settings.

If you use max settings you get a big difference between the Pascal Titan and GTX 1080 even at 1080p.

If you turn the settings down they score about the same as they are both CPU bottlenecked.

TheRealDeal · 30 Sep 2016 at 10:44

Kaapstad said:
Nope they are not using max settings.

If you use max settings you get a big difference between the Pascal Titan and GTX 1080 even at 1080p.

If you turn the settings down they score about the same as they are both CPU bottlenecked.

It says Max detail which to me is everything bar aa. It's at 4k any how so that takes away a cpu bottleneck.

Kaapstad · 30 Sep 2016 at 10:48

TheRealDeal said:
It says Max detail which to me is everything bar aa. It's at 4k any how so that takes away a cpu bottleneck.

Max settings include AA as in the OP of this thread.

TheRealDeal · 30 Sep 2016 at 10:50

Kaapstad said:
Max settings include AA as in the OP of this thread.

I know the settings off this thread. I am just trying to figure out what is happening with the AMD scores as they seem way off in comparison to all the review sites whether they run at 4k/1440p/1080p. It's not a cpu bottleneck as can be seen by the 4k results on other sites. Nvidia probably have the edge but not a 50% edge card per card. It's way off from what it should be.

Basically amd on sites were gaining 40-45% from DX12. That's not happening here which is why the gap is so big between Nvidia and AMD. Best guess is dx12 support has been borked in a recent patch.

Kaapstad · 30 Sep 2016 at 10:55

TheRealDeal said:
I know the settings off this thread. I am just trying to figure out what is happening with the AMD scores as they seem way off in comparison to all the review sites whether they run at 4k/1440p/1080p. It's not a cpu bottleneck as can be seen by the 4k results on other sites. Nvidia probably have the edge but not a 50% edge card per card. It's way off from what it should be.

The scores are what they are and the 480 and 1060 used the same PC with the same settings.

@1080p in the links you posted the scores are CPU bottlenecked. At higher resolutions it looks like AMD cards don't handle AA as well as the NVidia cards.

TheRealDeal · 30 Sep 2016 at 10:57

Kaapstad said:
The scores are what they are and the 480 and 1060 used the same PC with the same settings.

@1080p in the links you posted the scores are CPU bottlenecked. At higher resolutions it looks like AMD cards don't handle AA as well as the NVidia cards.

I am not denying the results Kaap. The Nano results prove that. I am just wondering why AMD are not gaining from Dx12 as if they were getting the 40% that other sites are showing results would look fine.

Kaapstad · 30 Sep 2016 at 11:06

TheRealDeal said:
I am not denying the results Kaap. The Nano results prove that. I am just wondering why AMD are not gaining from Dx12 as if they were getting the 40% that other sites are showing results would look fine.

I think I got it right the first time when I said they need to work on the drivers.

The RX 480 does not lack the power to do well so that just leaves software.

TheRealDeal · 30 Sep 2016 at 11:10

Kaapstad said:
I think I got it right the first time when I said they need to work on the drivers.

The RX 480 does not lack the power to do well so that just leaves software.

On this evidence you may be correct or the games Dx12 needs patching again. Amd's Dx11 in this game is pitiful as well.

Doom112 · 30 Sep 2016 at 13:59

TheRealDeal said:
I am not denying the results Kaap. The Nano results prove that. I am just wondering why AMD are not gaining from Dx12 as if they were getting the 40% that other sites are showing results would look fine.

DX12 is only helping AMD at CPU bottleneck situations at the moment.

TheRealDeal · 30 Sep 2016 at 14:07

Doom112 said:
DX12 is only helping AMD at CPU bottleneck situations at the moment.

It's not helping Amd at all any more. I have shown above that at 4k dx12 is still helping AMD so it's not all about Cpu bottleneck. When dx12 came out for the game the gains were 40%+. There are no gains now, infact dx12 is slower.

http://cdn.videocardz.com/1/2016/09/10_TOTALWAR.png

Check the link above. Add 15% on to the Nvidia frames as that seems to be the hit for the 10 series in this game. As you can see all the way up to 4k the Amd cards would still be doing decent even with the added 15% percent for Nvidia in dx11.

Amd have either screwed up in the drivers or a patch has taken away the dx12 benefit for AMD.