The Talos Principle PC Benchmark Performance Nikolas Nikolaou December 14, 2014 Benchmarks A puzzle game isn’t something you usually benchmark, but as The Talos Principle is a creation of Croteam, the guys behind the Serious Sam series, this performance analysis was inevitable. The Talos Principle is a first person or third person (that’s up to you) philosophical narrative puzzle game. You play the main character, a mechanical robot, trying to find the path to enlightenment, guided by Elohim a deitic voice. If you’ve played Portal or Portal 2 you’ll get the feel for the game and progress through the laser beam, fan and box puzzles, collecting your sigils (tetris blocks) fairly quickly. The game uses the Serious engine and as mentioned is developed by Croteam and published by Devolver Digital. While not the most demanding engine, it’s 3 years old afterall, it’s still impressive and blends well with the free roaming puzzle solving scope of the game. Progressing into our benchmark performance analysis they’ve added a handful of advanced options as well as a built-in benchmark. Native SLI and Crossfire support is not present, so we will have to stick to single GPU results for the time being. Benchmark Components The Talos Principle Benchmark System CPU i7 3930K 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: 344.75 WHQL, AMD: Catalyst Omega 14.12 Testing Methodology The Talos Principle has a very unique built-in benchmark that runs a predefined 15,30,60 or unlimited seconds run. You can find the benchmark in the first game menu. It runs through the puzzles in first person mode, exactly as if you were playing the game. Its 100% representative of game performance. I used a 60 second run on all tests as seen in the video above. After completion you can find a very detailed report in the Log folder of your The Talos Principle game install directory. If you run multiple benchmarks they will be logged in the same file. There’s quite alot of info in the log file from game errors, system info aswell as benchmarks results. The – benchmark results – section is what will be discussed in the benchmark charts and will be similar to the below lines : 03:17:24 INF: Duration: 60.0 seconds (2877 frames) 03:17:24 INF: Average: 48.0 FPS (49.3 w/o extremes) 03:17:24 INF: Extremes: 112.3 max, 27.0 min 03:17:24 INF: Sections: AI=5%, physics=1%, sound=1%, scene=77%, shadows=15%, misc=1% 03:17:24 INF: Highs: 351 in 5.8 seconds (60.3 FPS) 03:17:24 INF: Lows: 515 in 13.4 seconds (38.5 FPS) 03:17:24 INF: 30-60 FPS: 97% 03:17:24 INF: > 60 FPS: 3% For all minimum and maximum FPS I used the Highs: and Lows: while for average FPS the first Average: that contains extremes. The developers have added a big array of graphical detail settings in the GPU settings menu, for those that would like to customize their experience. I prompted to go for the highest possible settings in the performance tab with everything turned up to Ultra, while in the graphics options menu I enabled Triple buffering, set FOV to 90, Multi-Sample AntiAliasing OFF and the Max 3D Rendering MPIX to the relevant resolution that was tested. The Talos Principle 1920×1080 Benchmark Both our benchmark graphic cards the GTX970 and R9-290 perform very well with a minimum of 76.5 FPS and averages just over 100 FPS. The GTX970 has a slight 5% lead over the R9-290 but nothing noticeable in terms of performance. The Talos Principle 1920×1080 Screenshot 2560×1440 Benchmark At 2560×1440 both cards are over 64 FPS minimum and 79 FPS on average frame rate. The R9-290 has a slight lead of 1-4 FPS, nothing extraordinary though. The Talos Principle 2560×1440 Screenshot The Talos Principle 4K Benchmark At 4K we see the AMD R9-290s memory bandwidth kick-in producing better frame rates across the min max and averages. The game as seen in the charts is not very demanding. Our benchmark graphic cards the GTX970 and R9-290 perform impressively across all resolutions with the GTX970 leading in the 1920×1080 benchmark by a small margin while the R9-290 outperforms in both the 2560×1440 and 4K resolutions. The game is playable and without any noticeable stutter or framedrops across all resolutions. The Talos Principle 4K Screenshot CPU Usage CPU Usage in the game is limited to 4 threads as seen above in the screenshot from MSI Afterburner. A lower end 2 core, as per system requirements should do the job while a 4-core is ideal. VRAM Usage Using the Nvidia GTX970 to test VRAM usage, the game uses roughly 2-2.3GB video ram on the 1080p,1440p and the 4K resolution. The game will be playable on most if not all graphics cards by just lower the settings. Total System RAM Usage The total system RAM usage is very low for The Talos Principle. With 3GB memory you should be able to run all resolutions and presets without any issues. Conclusion The open world element while being linear when solving puzzles, does add a very unique experience to the game. It’s imposing how they were able to combine ruins, sci-fi and philosophical musings all in one game while keeping the experience unique and enjoyable. I loved the puzzles and the way the Elohim (the godly voice) got into my head. It might have it’s roots from the original Portal but in my opinion stands as the closest any game developer could get to Valve’s approach to the puzzle game genre. The Talos Principle had no performance issues whatsoever. The graphics might not be “next-gen” but aren’t shabby, the overall gameplay and trouble-free experience covers more than enough to please most if not all gamers. Note: The new NVIDIA GeForce 347.25 WHQL driver now has support for SLI. If you have any questions about the benchmark please ask in the comment section below and I will be glad to answer. mstrjidi Gtx 980 SLI Amd 9370 8core @ 4.715 32gb ram @1600 2560×1080 Sabertooth 990 FX r2.0 Evga 1000w 500gb evo ssd – Initially, using default settings on talos, getting both significant frame drops and stuttering. Was actually very surprised and disappointed how noticeable the performance drop was.(especially while on elevators and near “level portals”.) Spent more time than anticipated tweaking settings to get game to run smoothly. Nikolas Nikolaou Thats strange, I run GTX970s in SLI when I usually game (only 1 GPU works in Talos for now) didn’t seem to have any problem with 4K on the highest settings with AA off. Are you running the 344.75 WHQL or the latest beta 347.09 drivers ? Mike Chen Not familiar with NV settings, but the Crossfire works well as per to the devs’ suggestion. (http://steamcommunity.com/app/330710/discussions/0/620696522087842086/) My setup: HD7850 (2GB, @950/1200) x2, driver 14.12 AMD FX6350 @ 4.2GHz 4GB DDR3-1600 x2 Asrock 990FX Killer Fatal1ty Antec 450W Classic 128GB Sandisk Extreme SSD + 1TB WD Black 1920×1080, all setting ultra except Vsync off. Average: 78.0 FPS (79.3 w/o extremes) Extremes: 148.2 max, 26.8 min Sections: AI=11%, physics=2%, sound=2%, scene=50%, shadows=30%, misc=5% Highs: 699 in 6.4 seconds (109.1 FPS) Lows: 731 in 12.6 seconds (58.0 FPS) 30-60 FPS: 8% 60 FPS: 92% Nikolas Nikolaou Yeap, like most games you can hack it with Nvidia Inspector. Thanks for the info. http://freecad-tutorial.blogspot.com/ kwahoo “While not the most demanding engine, it’s 3 years old afterall” Nope. http://www.croteam.com/technology/ Nikolas Nikolaou Thanks for commenting . I meant to say it’s a heavily modified version of the original engine, I didn’t word it properly I guess.