Tom Clancy The Division - DX12 patch

Turns out that HBAO+ doesn't work under DX12. So my gains are invalid as they had HBAO+. It uses some random method instead, but still shows HBAO+ in options.

I'm gonna run my tests again.

 

So I did new runs on my system with AO on Ultra instead of HBAO+.

DX11:

Spoiler

DX12

Spoiler

Still getting nice boost.

 

Come again? http://www.guru3d.com/articles_pages/the_division_pc_graphics_performance_benchmark_review,6.html

 

So apparently PCSS shadows don't work either on DX12. So even more new runs:

DX11

Spoiler

DX12

Spoiler

So final verdict on benchmark is +6fps.

 

Dont you see when HH did that benchmark? Check some recent ones.

 

I won't be disrespectful to HH and look elsewhere. This is his site and I'm not going to doubt his work.

 

Sure thing. I would be happy if HH did a DX12 update benchmark but hes a busy man so im not asking for it.

Still, his last benchmark is 4 months old. Things change.

 

Feel free to supply a recent bench including the cards you mentioned from a reputable site.

 

I have no clue how amd is doing right now compared to nvidia on division but it wont hurt you (or HH) looking other reviews out of guru3d if they are more recent

His review was accurate and well done at that time but game had many patches and many drivers has come out since then.

Im sure HH will understand. Hes a cool man after all.

 

Posting a more recent test has nothing to do with respect, be serious.

 

feel free to post one, from a respectable site, if you find one. I quickly googled the division performance test and all the results were from March.

 

Run with Ultra Preset.

I ran the benchmark on my 980ti/5820K combo.
c1:
Graphic Setting - Ultra Preset

Spoiler

c1:
DX11 Setting

Spoiler

c1:
DX11 Results

Spoiler

c1:
DX12 Setting

Spoiler

c1:
DX12 Results

Spoiler

:nerd:

 

This was the last 480 HH tested. Definitely not ahead of the 980. And that it a slightly large gap to make up.

 

He said that the 290x was ahead of the 980 not the 480. still, wrong.

 

Did a run on 1080 too. Ultra presets with no vsync.

1080p

DX11 Avg 75,9 _-_ 99% 76,4

DX12 Avg 83,7 _-_ 99% 84,2

1440p
DX11 Avg 58,2 _-_ 99% 58.9

DX12 Avg 62.7 _-_ 99% 63.2

Biggest difference is min fps in both test.

 

Hilbert added to the graph, he didn't retest all those cards lol

 

It shows the latest 480 drivers at the time were not a huge step from the initial ones tested so the graph is still relevant. there were no huge improvements on AMD's side at the time of the last test Hilbert did. I would like to see the relive drivers though.

So your improvement from Hilbert's test was 2fps average. That kind of proves that Hilbert's table is accurate enough to use.

 

yes, but the 480 is still behind the 1060 and they were tested reasonably close to each other, time wise. That was half of this argument. I'd like to see a more up to day chart with all the cards re-tested but until I see one, I'll just go by what HH tested.

 

What are you, 5 years old?

Techspot has the 290X ahead at 1600P

They also use a 6700K, which everyone knows will help with the overhead over a 5960X due to increased IPC.

1080P has them nearly tied. Recent drivers have helped with CPU overhead in DX11, which is irrelevant anyway now, because ASYNC Compute is now used in DX12. 980 is left out in the cold on that one.

1440 290X would win as well:

http://i.imgur.com/kGS2bnN.png

1080P

http://i.imgur.com/oRCJsU7.png

 

For some reason it works today.
DX11
1075/1300. Which is what I use for most games.

1175/1500

DX12
1075/1300

1175/1500


Very small gains.
DX12 is the same as my DX11 overclock performance pretty much.
The game is smoother with freesync in DX12 vs DX11.