Battlefield 1 DirectX 11 & 12 Performance Analysis
Battlefield 1 – 1080p, 1440p and 4K DirectX 12 Benchmarks
Well this is a big disappointment, as we see the DirectX 12 bring down the overall FPS of almost every card at the top of the list by around 10FPS. However, almost every other card from about the Fury X down saw equal performance or a few minor FPS gains. It’s clear that DirectX 12 is bringing better improvements for older hardware, so it’s advisable to test both on your hardware to see which runs better, rather than just picking one and sticking with it.
The GTX 1080 dropped a few frames down here too, although, for the most part, we’re seeing small improvements on the rest of the board. While overall average frame rates are often a tiny bit lower, the minimum frame rates have increased, which means a smoother gaming experience overall.
This is where things get pretty interesting, as the GTX 1080 has lost performance in DX12 so far, but once we push to 4K, it actually gains around 8FPS minimum and average, while all the most cards suffer further losses in average frame rate, but a small increase in minimum FPS.
Would’ve been interesting if you’d done the test vs. a system with only an i5, for instance.
With a GTX 1080 (just upgraded from 980ti) I ALWAYS try to avoid DX12 and opt for DX11. Because I tend to lose up to 10fps just by using DX12….
the only upside to using DX12 is that I won’t get that stutter every 3-4 Seconds like you’ll get by using DX11 in processor-heavy games like GTA5, for example.
At the end of the day, when hardware and software don’t match up right, you’ll get crap like that^^^ (i.e. DX12 -> CUDA [nvidia] // DX12 -> SM Unit [AMD])
Because DirectX is written by Microsoft, and we all know how great Microsoft is at coding…. /s
The only way to remedy these trade offs is for AMD or Nvidia to write their own respective API’s. but that will never happen
“The only way to remedy these trade offs is for AMD or Nvidia to write their own respective API’s. but that will never happen”
AMD partially did this with Mantle, now Vulkan. I think that counts.
The dx12 api may give you some gpu saving on getting ride of some dx11 implicit barriers and cache flush plus usually improve driver madness on heavy traffic in regards to texture streaming that could hitch on a dx11 implementation. BUT, it is an api to improve cpu, it does not really matters to test at 4K, you would have better indicators if you run at 720p on mid range CPUs.