Dying Light Benchmark Performance Nikolas Nikolaou January 29, 2015 Benchmarks Zombie survival games seem to have become a craze lately. Today we are taking a look at the PC performance in our Dying Light benchmark. Dying Light is a first person open world apocalypse-themed survival game, developed by Techland (the studio behind the Dead Island series) on the Chrome 6 Engine and published by Warner Bros Interactive Entertainment. I’m not into Zombie games, but Dying Light seemed very interesting both gameplay and visual wise watching pre-release game footage. Dying Light provides a singleplayer , co-op and a 4v1 player Be The Zombie mode, similar to what we have in Evolve. The game starts out with cutscenes and a very lengthy tutorial in the singleplayer campaign. Dying Light has a lot of free roaming and parkour involved which needs a little getting use to, especially mastering jumping and clinging in the open world across terraces. Melee weapons like axes, pipes and just about anything you can find are your basic arsenal, guns are scarce and hard to find, even a good few hours in. The more you progress the more you unlock abilities through each of the individual three skill trees. Theres a unique day-night cycle, where preparation is done in the daylight to kill off zombies that become more aggressive when the sunsets. Getting into the benchmark performance , Nvidia have released an optimised driver with a special SLI bits profile that you can update through Geforce Experience, but sadly with any resolution over 1080p the game performs worse than a single GPU. I decided not to incorporate any SLI/CFX benchmarks until both camps come out with better optimized solutions. Benchmark Components Dying Light Benchmark System CPU i7 3930K @ 3.8Ghz Motherboard Asus P9X79 Deluxe RAM Kingston Hyper-X 16GB 1600mhz HDD/SSD Crucial M550 256GB GPU Galax GTX970 EXOC 4GB , MSI R9-290 Gaming 4S Monitor Asus PB287Q 28″ 4K 60Hz 1MS PSU Corsair AX1200W OS Windows 8.1 Drivers Nvidia: 347.25 WHQL, AMD: Catalyst Omega 14.12 Testing Methodology One of the most demanding areas of the game is when you exit buildings into the open world. That’s where I decided to start out the 60 second run that can be seen in the video below. I went for a slightly different method of testing than our usual benchmarks, with presets varying for each resolution. None of the Nvidia Gameworks features were used as we are also testing with an AMD GPU. All settings where turned Off – Anti-Aliasing included – other than the options featured below, while slider options where left as is other than view distance at 25% (1/4). 1080P – Texture Quality High, Shadow Map High, Foliage High and Ambient Occlusion ON 1440P – Texture Quality Medium, Shadow Map Medium, Foliage Medium and Ambient Occlusion ON 4K – Texture Quality Low, Shadow Map Low, Foliage Low and Ambient Occlusion ON Dying Light 1920×1080 Benchmark Dying Light 1920×1080 Screenshot 2560×1440 Benchmark Dying Light 2560×1440 Screenshot Dying Light 4K Benchmark Dying Light 4K Screenshot It’s clear that the AMD drivers are holding back the R9-290 as we don’t have an optimised version out yet. The GTX970 performance is acceptable at the settings I used for each resolution, while not cranked up they are at least providing playable frame rates across all resolutions – 4K included. I mentioned previously that I set the view distance slider to 25%, which is the main reason we are getting higher FPS than what most users are reporting across various forums. While with some settings you get minimum to no increases, view distance reduction provides a boost of over 50% in most cases. CPU Usage A lot of users have mentioned how one or more of their cores seem to be spiking to 100% and random stutters occuring. With the i7-3930K I tested, I did notice increased CPU usage on one core and heavy usage (30%+) across all threads, but didn’t encounter any form of stutter. The minimum requirements by Techland specify a i5-2500 and a i5-4670K as recommended, which looks about right, but the problem seems to lie in the optimization. The game does utilize multithreading but also needs good single core performance which is somewhat counterproductive, especially for users that don’t overclock and/or have higher-end 6 or 8 core CPUs. VRAM Usage I clearly want to state before you look at the chart, that VRAM usage might seem somewhat off. This hasn’t got anything to do with the game itself, but with how I used High settings at the HD resolution, Medium at 1440p and Low at 4K. Total System Ram Usage Looking at the relevant resolutions and the total system memory usage it is clear that the game does need over 4GB. The memory requirements initially were 4GB minimum and 8GB recommend , but have now changed to 8GB and 16GB respectively for no severe bottlenecks. Conclusion While I did enjoy the gameplay across all three of the modes (especially playing the Zombie), the nuances to some of the better PC ports available seem to be mostly performance related. While I did fiddle with the settings to get a playable and stable frame rate, the game didn’t feel that responsive when interacting with different objects. There is some noticeable pop-in in some areas that is mostly from surrounding texture LOD, not NPCs like we had in AC Unity. The visuals while impressive aren’t what you would be waiting for with such a performance hog. I want to state that most users will not have this high-spec and well rounded of a system, and will probably encounter performance issues. This has nothing to do with your system but clearly the way Dying Light has been ported. I hope that issues will be resolved with a optimization patch and better drivers from both Nvidia and AMD. If needed I might redo the benchmark at a later point in time adding SLI and Crossfire. See Pricing Update: Patch 1.2.1 has lowered LOD (maximum LOD is 55% of previous value). CPU usage is better spread between the cores. SLI is still problematic. Update: Patch 1.4.0 has lowered system memory usage as well as increased frame rate across the board. If you have any questions about the benchmark please ask in the comment section below and I will be glad to answer. trAnwhiz Why did you stop posting frametime results? I think we’ve come to know that it’s a very important statistic, sometimes more than just framerate, and especially when testing the GTX 970. Nikolas Nikolaou If the AMD driver was out I would, but performance is not on par without them. Here are the frametimes for the GTX970 but my settings are custom for each res (1080 high no AA, 1440p Med, 4k Low) as mentioned above. My GTX970s are getting RMAed on Monday btw. trAnwhiz Thanks, it doesn’t look like the GTX 970 gets a great experience. Are you RMAing them because of the 3.5GB memory thing? Nikolas Nikolaou Yes, I was using them in my main rig ( because of the low heat output) and as I play @ 4K and like eye candy. I was getting strange stutter and in general a bad experience over the 3.5GB mark. The R9-290s @ 4K are so much better and smoother. trAnwhiz Damn, nvidia focked up, I’m playing on 2560×1440, don’t know if I should return them (if my retailer will even accept it). Maybe there’s a chance for some sort of improvement with a driver update, I doubt that’s possible, though. Nikolas Nikolaou The only sort of fix would be for them to limit the card to the 3.5GB with no access to the slower 512mb imo. You should be fine for now or if/when VRAM hits 3.5GB+ lower the settings slightly.