Rise of the Tomb Raider DX11 vs. DX12 Review
Rise of the Tomb Raider has recently received a new patch which adds DX12 API support, in addition the patch adds NVIDIA VXAO Ambient Occlusion technology, however just under DX11. In this evaluation we will find out if DX12 is beneficial to the gameplay experience currently and how it impacts certain GPUs.
Rise of the Tomb Raider has become one of those games where forward looking GPU technologies are tested and utilized to better the gameplay experience. Rise of the Tomb Raider was already released with some groundbreaking GPU graphics effects that pushed GPU hardware to the limits. Now, a new patch has recently been released that offers the absolute latest support in API and new graphics effects. This game (with the new patch on March 11th), along with Hitman (also released on March 11th) are the first two AAA games to support the new DX12 API.
The developers working on Rise of the Tomb Raider are actively pushing the boundaries of graphics features and support even though the game has been officially released since January 28th. New patches are adding new features. We evaluated this game in our Rise of the Tomb Raider Video Card Performance Review on February 15th after a couple of patches had been released. These patches did improve performance and image quality in the game.
We then followed up with a Rise of the Tomb Raider Graphics Features Performance evaluation with another patch under the game's belt.
New Patch with DX12 Support
Released on March 11th was patch #5 bringing version 1.0.638.6. This latest patch on March 11th was a big one for the game, it introduced three important things into this game that weren't there before. There is now DX12 API support, there is now NVIDIA VXAO technology, and there is now a built-in benchmark in the game. You can see that our version above is actually 1.0.638.8; a hotfix was released early on for an issue you can read here, we are using the latest version.
Added support for DirectX 12 - DirectX 12 is a new advanced graphics API that on the right hardware can offer far better performance.
Adds NVIDIA VXAO Ambient Occlusion technology. This is the world’s most advanced real-time AO solution, specifically developed for NVIDIA Maxwell hardware. (Steam Only)
Added a new benchmark feature to allow easy comparison of performance on different systems, as well as at different graphics settings.
More information about how DX12 is implemented in this game can be read from the dev blog here.
Enabling DX12 is as easy as ticking a checkbox, if it is unchecked the game will run in DX11. Therefore, we can easily make gameplay comparisons.
Today we are going to focus on the DX12 API aspect of this game, in a follow-up we will look at VXAO image quality and performance.
We are using the latest video card drivers possible from AMD and NVIDIA. We are using AMD Crimson Edition 16.3.1 Hotfix Beta and NVIDIA GeForce 364.51 WHQL. After we installed each driver, and then checked the supported feature level API in DirectX Diagnostics and GPUz we found something rather interesting we will point out.
In the screenshots above check out the DirectX feature level support. On the AMD Radeon R9 Fury X (and all Fury or 300 series) the highest feature level support is 12.0 (12_0.) However, on NVIDIA GeForce GTX 980 Ti (and all Maxwell GPUs) the highest feature level is 12.1 and 12.0 (12_1 and 12_0.) It appears that Maxwell currently has 12.1 exposed and AMD does not yet. Probably a very minor nitpick, but worth noting.
GPU Testing Must Change?
Now that DX12 API support is finally here some things are going to change. Honestly, for us, it isn't a huge change since we've been evaluating GPUs this way now for more than a decade. Simply put there are no framerate over time capture tools like FRAPS that currently support the DX12 API. FRAPS does not support the API and there are no other third party tools that currently do, even NVIDIA's FCAT does not. This means we cannot record actual real world gameplay framerates in DX12 currently.
In fact, we can't even see the active FPS in the game as we play it under DX12. We have truly arrived at a time now where the framerate being exposed is entirely removed from the gameplay experience. This moment now that we have arrived in is the moment we were preparing for since 2003 when we changed our evaluation methodology.
Currently we can only completely rely on the highest playable settings of the game in DX12 and compare cards based on in-game quality features we are able to turn on.
In order to show and focus on the highest playable settings between DX11 and DX12 we have worked up a table format change to make comparing features easier. You will be able to look across a table at all the in-game quality settings and easily see which video cards allowed higher settings than the others.
In games that have a built-in benchmark mode made by the game developer, like Rise of the Tomb Raider now has we can use that built-in benchmark mode to replace what would have been our apples-to-apples section. It must be understood that the built-in benchmark mode in any game is an on-the-rails pre-recorded demo from the game developer and always has the potential for foul play via NVIDIA or AMD specifically optimizing performance just for those sequences.
However, we are going to show you this performance as it is programmed in by the game developer and it is something you can run yourself on your own game to compare to our results. The canned benchmark should not be considered as the primary focus on performance. In our highest playable settings we play through the game in real-time trying different levels, maps and areas to make sure the settings we chose are playable throughout the game.
Certainly, it will be interesting to see how the canned benchmarks compare to the our real world gameplay. In the past we have seen where these have been spot-on, and we have seen where these have been very different.