Hitman DirectX 12 Internal Benchmark Tool
First I'll start with the built in benchmark tool that is provided. Usually benchmark tools aren't accurate and doesn't give a good representation of actual gameplay. We will see how accurate the benchmark tool is when compared to my Real Time Benchmarks™. Let's take a look at the settings and the benchmark:
"Render Target Reuse" = Disabled
The first issue I noticed was that there is a setting named "Render Target Reuse"which is a feature only accessible from the DirectX 12 API. This settings had massive issues with the gameplay and caused tons of micro-stutter. High frame times also occured when this setting is enabled. These issues definietly hurts the benchmark results. The graphical settings are at the highest gamers can push them and the "Render Target Reuse" is set to Disable as you can see.
Check out the benchmark below:
Fury X - Hitman Internal Benchmark Tool
3285 frames
28.97fps Average
4.83fps Min
65.35fps Max
34.51ms Average
15.30ms Min
206.97ms Max
Those aren't the worse stats. but there's more that isn't seen. As I mentioned above there was tons of micro-stutter. The Frame Time was all over the place. There was only 3285 frames rendered as well. This is definietly not a good experience and I urge all Fury X users to "Enable" this feature or set it to "AUTO". This setting definitely shouldn't be used and I've decided to leave it set as Enabled isntead of AUTO just in case. Let's see what happens when I enabled the feature.
"Render Target Reuse" = Enabled
Fury X - Hitman Internal Benchmark Tool
4081 frames
36.02fps Average
5.24fps Min
65.61fps Max
27.76ms Average
15.24ms Min
190.95ms Max
As you can see there are huge gains across the board. There were more frames rendered and the frame rate & rendered frames increased by 24%. The Frame Time is much better. The benchmarks was much more pleasant with much less micro-stutter and issues found with the the Render Target Reuse Disabled. So once again I highly suggest that AMD Fury X users keep the Render Target Reuse Enabled.
With that out of the way lets looks at how the internal benchmark performance with different settings and resolutions.
Hitman Internal Benchmark Tool GPU Results - DirectX 12
| 1920x1080 | 2560X1440 | 3200x1800 | 3840x2160 |
| 10112 frames | 8164 frames | 4976 frames | 4081 frames |
| 89.43fps Average | 72.30fps Average | 43.96fps Average | 36.02fps Average |
| 8.79fps Min | 10.14fps Min | 8.30fps Min | 5.24fps Min |
| 151.52fps Max | 115.43fps Max | 83.03fps Max | 65.61fps Max |
| 11.18ms Average | 13.83ms Average | 22.75ms Average | 27.76ms Average |
| 6.60ms Min | 8.66ms Min | 12.04ms Min | 15.24ms Min |
| 113.83ms Max | 98.66ms Max | 120.48ms Max | 190.95ms Max |
As you can see the stock Fury Xperforms very well in DirectX 12. The minumum FPS are misleading. THe internal benchmark tool has some issues that it needs to work out. When I benchmarked actual gameplay I never got FPS mins shown above. Next I'll modify a few settings to see how much performance I can squeeze out at 4K when lowering some settings. I came up with this:
| 3840x2160 |
| 4997 frames |
| 44.27fps Average |
| 8.80fps Min |
| 71.11fps Max |
| 22.59ms Average |
| 14.06ms Min |
| 113.67ms Max |
I lowered the LOD to High, completely disabled AA, and dropped Shadow Maps to High. The game still looks fantastic and I increased my FPS by nearly 23%. The rendered frams increased by 22.44%.
I'll be sure to update this post with the DX11 results in the coming days. I'll also be sure to add overclock settings soon as well. Now lets see how well the Fury X performs in DirectX 12 when using my own benchmark program. I had a few issues with DX12 and benchmarking the GPU output, but those issues have been resolved. I spent most of the day programming and ensuring that everything was accurate and working properly. Everything is looking good. Just as a reference I include something that I like to call "FPS Min Caliber".Check the next page for a short explination and my Real Time Benchmarks ™