Yesterday I finally wrapped up a few days worth of testing for the latest Battlefield title. The focus was mainly on GPU performance, so I stuck with my Core i7-6700K test rig for the most part. The DirectX 11 results had the RX 480 and GTX 1060 performing pretty close to one another though the GeForce graphics card was a few frames faster.
I only took a quick look at the DirectX 12 performance because what I had seen so far wasn’t that impressive and reports from others at the time seemed to be inline with my own findings. The AMD graphics cards delivered about the same performance while the Nvidia cards were a bit slower.
A few of you were quick to point out that a few highly rated outlets such as DigitalFoundry, found the RX 480 to be quite a bit faster when using DX12. With my Core i7 test rig overclocked to 4.5GHz this wasn’t the case. At 1440p I got the same performance of 67 to 68 fps on average.
Since finishing my initial testing, AMD also released the Crimson 16.10.2 driver which they accidently forgot to pass along to my little channel. Anyway last night around 10pm my AMD contact was still awake and he passed the driver along, what a champ! I immediately got testing and I am about to show you what I found. For this video I am just going to focus on the GTX 1060 and RX 480 along with the Core i7-6700K and Core i5-6600K. Since we won’t be doing a huge amount of testing for this one I will just show you the gameplay footage which has been captured using the external Elgato HD60S. Let’s get to it…
First up we have the 6700K test machine which is clocked at 4.5GHz. With the tank stationary at the start of our test we see that the 1060 is a few frames faster than the 480 using DX11. However using DX12 the 480 is around 5 or so frames faster than the 1060. Interestingly the 480 looks to be delivering much the same performance using DX12 as it did DX11. The difference being that the GTX 1060 is slower. Let’s just watch this footage till the end and then move onto the Core i5 results…
Here we have the same system with a quick CPU swap. Now armed with the Core i5-6600K, which I might add isn’t overclocked, let’s have a look at the performance. At the beginning of this test it is interesting to note that the performance is almost exactly the same as what we saw with the 6700K. In my initial benchmark I noted that when briefly testing with the 6600K it was just few frames slower than the overclocked 6700K, and once again here we can see that this is indeed the case.
Again we see that for the most part the RX 480’s performance seems much the same using DX12 and DX11. Meanwhile the 1060 is up to 10 fps slower using DX12.
Here is a direct comparison between the RX 480 running DX11 and DX12 in our test. Really at no stage is there more than a 2fps difference. Sometimes DX11 was faster and sometimes DX12 was, which makes sense since I originally found no real difference here. These results were recorded using a heavily overclocked Core i7 processor so again let’s check the Core i5 6600K results.
Even with the stock 6600K the results seem to be much the same, with a few exceptions. Towards the end of the benchmark where things really kick off the RX 480 takes a serious performance hit using DX12 and this contradicts what some others have reported. Whereas the RX 480 never drops below 83 fps using DX11 we see it drop down to 51 fps very briefly using DX12 while quite a bit of time was spent in the 60’s. So I am not sure what to make of that.
Moving on to the final comparison I would like to make let’s see how the GTX 1060 using DX11 compares to the RX 480 using DX12. Initially the 1060 can be seen sitting 5 to 7 fps ahead of the RX 480. Throughout the test the two do trade blows, though for the most part the 1060 does stay ahead. At the end of the test when things get busy the 1060 never dipped below 78 fps whereas the RX 480 often drops down into the 60’s. This is by no means bad performance from the RX 480 but it was at times 10fps slower despite using DX12.
Okay so I am not sure if this has helped clear anything up or just created more questions. I do stand behind my initial DX11 test and strongly believe that to be 100% accurate, if I didn’t I would amend the video. I admit more DX12 testing needs to be done and I plan to tackle this more in the future. Unfortunately for now this is the best I could do with the time I have available. As many of you know I am going in for ACL surgery next week and before then I have to cover the GTX 1050 and 1050 Ti graphics cards for you guys.