The Division DX12 Patch Benchmarks

[Head1985]

 

[Bacon1]

http://www.gamegpu.com/mmorpg-/-onlayn-igry/tom-clancy-s-the-division-test-gpu.html

SLI doesn't work in DX12 so thats why min/max are lower on CPU testing, but its a huge boost for the weaker CPUs and anyone not with SLI 1080

 

[Spjut]

Awesome CPU results! I wish more sites would do thorough CPU tests, it's one of the key improvements possible in DX12. It's the same deal like when BF4 got Mantle support and almost all sites insisted on using highly overclocked i7 and ultra settings.

 

[richaron]

http://www.gamegpu.com/mmorpg-/-onlayn-igry/tom-clancy-s-the-division-test-gpu.html

*Snip pics*

SLI doesn't work in DX12 so thats why min/max are lower on CPU testing, but its a huge boost for the weaker CPUs and anyone not with SLI 1080

I can't read the OG source but I think those DX12 CPU results are much more than "SLI doesn't work in DX12"...

Edit: And I've always been suspect of the order "gamegpu" chooses with their graphs, andI've posted here about it. I told myself maybe they just aren't printing a decimal place and I'm imagining things.. But now it appears sometimes it's reverse alphabetical order: so if all things equal Intel or nV will be graphed above AMD? (Honestly curious as to their method).

 

[Headfoot]

the lack of improvement on the GPU front in DX12 goes to show how incredibly well optimized this game was in the first place. Really good visuals with solid performance even on DX11.

 

[Bacon1]

Awesome CPU results! I wish more sites would do thorough CPU tests, it's one of the key improvements possible in DX12. It's the same deal like when BF4 got Mantle support and almost all sites insisted on using highly overclocked i7 and ultra settings.

Yeah wish they would have tested a 1060/480 instead of SLI 1080 for the CPU results, would be interesting to see how an i3/i5 compare in dx11 vs dx12.

It is a very well threaded game already though in dx11.

And I've always been suspect of the order "gamegpu" chooses with their graphs, andI've posted here about it.

Yeah their graphing is bad. They add the min + avg on their graph length, instead of having the min take up part of the avg. So 31 min, 51 avg would look much larger than 13 min, 60 avg even though the 2nd one has higher avg overall. I don't know how they handle min either, if its 99% or single lowest or what.

 

[richaron]

Yeah their graphing is bad. They add the min + avg on their graph length, instead of having the min take up part of the avg. So 31 min, 51 avg would look much larger than 13 min, 60 avg even though the 2nd one has higher avg overall. I don't know how they handle min either, if its 99% or single lowest or what.

I'm not sure of the specifics either, so I hope someone can chime in. And not simply about the min + avg thing, I think that's almost obvious enough to slide (although it can be misleading especially considering below).

It's the fact they order by "min" rates first, but if min is equal it appears they list Intel or nV above AMD. It sounds paranoid so in the past I've guessed they're using another next decimal place in "min" (not published), but the graphs above appear to confirm my tinfoil hat theory that if mins are equal they will list Intel/nV above AMD (and I have noticed nV products listed above AMD products with equal "min"s, but AMD has higher "avg" fps).

Obviously if you ordered your graphs by "min" fps first, if you have a tie you should then order those tied objects by "avg", but it appears they don't bother doing this. They simply list Intel/nV products higher on their graphs when they have equal "min"s even though the AMD product may have a higher "avg" fps.

I would be happy to be proven wrong in this, but I've seen it enough times to feel confident ranting about it...

 

[Bacon1]

They simply list Intel/nV products higher on their graphs when they have equal "min"s even though the AMD product may have a higher "avg" fps.

I know for the higher resolutions they tend to keep the items in the same order as they were in the 1080p, or used to anyway, I haven't bothered to try to keep track of their terrible graphing system lol