Low End PC Performance Guide: Tomb Raider
The newest game from the Tomb Raider franchise – the 10th in the franchise, a reboot telling how Lara Croft became Tomb Raider and doesn’t carry any of the storylines from the previous games. This is the fourth Tomb Raider game developed by Crystal Dynamics, and they used a heavily modified and updated Crystal engine, the engine they used in Tomb Raider Underworld and Guardian of Light. As a result, the game now supports DX11 features like hardware tessellation, super-sample anti-aliasing, and the new TressFX developed by AMD – a hair physics effect rendering a realistic hair animation.
We’re gonna take a look at how this game will perform on low-end systems and possible ways how to improve its performance.
Test System and Requirements
|Test System||Minimum requirements|
|Processor||Intel Celeron G550 2.6 GHz Dual-core||Dual core CPU: AMD Athlon64 X2 2.1 Ghz (4050+), Intel Core2 Duo 1.86 Ghz (E6300)|
|Memory||4GB DDR3 1600MHz||2 GB|
|Video Card||AMD Radeon HD 7750 1GB DDR5 nVidia GeForce 9600GT 512MB DDR3||DirectX 9 graphics card with 512Mb Video RAM, AMD Radeon HD 2600 XT, nVidia 8600|
|Driver / Patch version||AMD Catalyst 13.2 Beta 7 nVidia Forceware 314.07||1.0.722.3 Survival Edition|
|Operating System / DirectX||Windows 8 64-bit, Windows 7 SP1 64-bit / DX 11||Windows XP Service Pack 3, Windows Vista, Windows 7, Windows 8 (32bit/64bit)|
The requirements are not that high and probably older systems can still manage to play the game with good performance, but we have to assume that it will only be on “Low” settings.
All tests were done with the following components and settings
- Intel Celeron G550 2.6 GHz Dual-core
- 4GB DDR3 1600MHz
- AMD Radeon HD 7750 1GB DDR5
- Windows 7 SP1 64-bit
- Normal preset, 1920×1080 (1600×900 for individual settings tests)
- Fraps 3.5.9 for recording average and minimum frames per second; 60 seconds and twice for each test.
Built-in Benchmarking Tool vs Actual Gameplay
I was curious if the built-in benchmarking tool really represents the performance of the game, so I picked “actual” gameplay to compare with. The level is from Helicopter Hill to Windmill Campsite.
At Low, Normal, and Ultimate presets, the two benchmarking process displayed different results, most especially at Low preset with 14fps difference. Only at Ultra preset where the two processes displayed almost identical results with 21 and 22 frames per second. I also noticed that at Ultra and Ultimate presets, the actual gameplay exhibited faster performance than using the Built-in benchmarking tool.
With these uneven results, I decided to use the actual gameplay for the rest of the tests. I am not saying that the built-in benchmarking tool is unreliable, it’s just I’m more comfortable using the gameplay for benchmarking. And besides, It’s the “actual gameplay”.
You can read a more detailed article here from our partner site comparing benchmarking using built-in tools against actual gameplay.
Image Comparison and Performance
Texture quality affects the sharpness and crispiness of texture on wood, stones, soil, and clothes. The differences between each settings are noticeable and I recommend texture quality to be set on ultra since there are no performance hit when changing between settings.
Texture Filtering makes the texture clearer. The higher the value, the farther the texture will become clearer. The performance is the same across each setting until you set it to 16x where the performance dropped by 1, but even that is negligible. I recommend to set this at 8x.
TressFX improves the look and the movement of hair, but took away some frames per second. For mid-range setup, I recommend hair quality to be set on normal.
Turning on any anti-aliasing make the edges smoother, the higher the value the better but will also take away some frames per second. FXAA improves the quality without losing performance. I recommend FXAA to be used on this setting.
Turning off shadows will give you additional frames per second but will make the scene unrealistic. I recommend to set this on normal.
Shadow resolution setting improves the overall quality of shadows. With low, shadows are blurry. Setting it to normal and high makes the shadows sharper. The quality between normal and high are indistinguishable but high has better performance. I recommend to set this on High.
Level of Detail
Level of Detail controls how objects and terrain would look like. The higher the setting, objects and terrain are more detailed. I recommend Medium for this setting.
The reflections on water don’t turn off and I couldn’t recognize any differences among values (Off, High, Normal) and their performances are all the same. Since this setting doesn’t make any changes, we can turn this off.
Depth of Field
Depth of Field makes the distant objects blurry and fuzzy, but only on Normal. Turning it to High turns it off. We can turn this off. It’s a personal preference, other like the blurry effect, others don’t.
Screen Space Ambient Occlusion
This setting adds shadows on objects. Performance hits are minimal but I recommend to set this on Normal.
Post Processing adds additional lighting effects. Lamps and fire produce more light, thus creating more shadows. The performance hit is minimal and I recommend to turn this on.
Tessellation is more noticeable on rocks. The performance hit is a little high and I recommend to turn this off.
High Precision puts more pixels on a given space or object making the object more detailed, but in this case we didn’t see any visual difference and the performance are all the same.
Based on quality/performance observation, I formulated our Custom preset to get the best visual quality possible while still having enough frames per second to play the game well.
|Texture Filter||Bilinear||Anisotropic 4x||Anisotropic 8x||Anisotropic 16x||Anisotropic 16x||Anisotropic 8x|
|Level of Detail||Low||Normal||High||Ultra||Ultra||Medium|
|Depth of Field||Off||Normal||Normal||Ultra||Ultra||Off|
Our Custom preset looked like another preset with visual quality comparable to Ultra preset with less smoke and less detail on objects, but the performance is way better than Ultra and a little better than Normal preset. We improved the performance without losing much visual quality. Knowing which settings to turn on and off is really helpful, especially for those who doesn’t have monster gaming rigs.
Video Card Performance
Beginning at Low Preset, the GPU usage and memory begin to rise as you switch to a more demanding preset. Ultra and ultimate preset will use 100% of your GPU’s processing power and you will need more than 1GB of video memory.
You can play the game up to high preset on AMD 7750 and only with normal preset at 1366×768 on nVidia 9600GT. Also notice that you don’t have access to ultra and ultimate preset if you are using nvidia 9600GT since it is not a DX11 capable video card.
The game doesn’t utilize the 100% of the CPU, but having lower clock speed than minimum, which is 2.1GHz, will not give you a satisfying experience. Also, having only 1 core of the CPU, even at 2.6GHz, will make the game too slow to play.
Having more system memory will improve the overall performance of the game – higher frames per second, faster game launching and faster loading level.
Operating System Performance
Windows 7 and 8 performed equally on frame per second. When loading a level, Windows 7 is faster by 5 seconds but Windows 8 is faster when launching the game.
The game looks really good and alive even at normal preset. The developers provided us with detailed and customizable settings to cover almost every type of gaming PC. The HD 7750 performed very well with normal and custom preset at 1920×1080 resolution while the Nvidia 9600GT performed well with low preset. Low budget processors are also capable of playing the game even at lower clocks. What I delighted the most is its memory performance. Performances, on frames per second and on loading and launch time, are getting better with bigger system memory.
Overall, Tomb Raider doesn’t need a monster setup to play and enjoy the game. It is truly a wonderful game from the story, gameplay, presentation, and graphics. If the game is really a great game, bad graphics won’t make it a bad game.