Welcome Stranger to OCC!Login | Register

Cryostasis: Sleep of Reason Performance Evaluation

ccokeman    -   June 16, 2009
» Discuss this article (13)

Testing:

Cryostasis is a PhysX heavy game that needs to have this option turned on to get the best visual effects. Unfortunately, this means added strain on the GPU for video cards that support this technology, and a massive performance hit when cards that do not support this technology are used with the PhysX effects enabled. Testing will consist of running the game through a five minute sequence that is a demanding section of the game and puts the GPU under strain from the opening sequence. I will test with the PhysX effects enabled, as that is the allure of the game in much the same way as Mirror's Edge. I will be using the OCC standard test bed to test the performance of each video card in resolutions ranging from 1280x1024 to 2560x1600. The drivers used for this test are the 185.85 from Nvidia, and the Catalyst 9.5's from ATI.

Testing Setup:

Comparison Video Cards:

 

Settings:

  • Vertical Sync: Off
  • Shader Model 4.0
  • Hardware PhysX: On
  • Texture Resolution: High
  • Normal Maps: Medium
  • Specular Maps: Medium
  • Shadows: Medium
  • Motion Blur: On
  • Camera Motion Blur: Off
  • Water Reflections: Off
  • Water Caustic: Off
  • Anisotropic Filtering: On
  • Anti-Aliasing: On

 

 

 

Higher Scores = Better

 

When Cryostasis is played as it is meant to be, you can see what kind of strain the system is put under, when even the top card in each manufacturer's food chain is pretty much beat down at 2560x1600. Even when running a GTS 250 to take care of the PhysX calculations, the single GPU Nvidia cards deliver unplayable frame rates. From 1280x1024 up to 1920x1200, the GTX 295 and GTX 285 are playable without the aid of the GTS 250, while the 275 and 260 above 1680x1050. The ATI cards just get beat down because all of the PhysX work is done on the CPU, making for a totally unplayable game with a performance that ranges between 8 and 12 FPS. To be fair, if you change just three settings (PhysX Effects, Anti-Aliasing and Anisotropic Filtering), the ATI cards deliver playable frame rates up to 1920x1200 at the expense of the visual effects.

 

In an effort to see what the CPU usage is for this game, I ran through a timed run and exited from the game to check the usage levels of the cores in Task Manager under the Performance tab. I checked with both the Nvidia cards and the ATI cards while PhysX was enabled. What I saw was that with PhysX enabled on the Nvidia cards, the CPU usage appeared to load two cores substantially more than the rest. When the ATI cards were run, the usage loaded two cores a bit higher than the Nvidia cards, but the rest of the cores look to be loaded heavier than when run with the Nvidia cards - showing the CPU doing more "work" with the ATI cards installed. While this did not look to be anywhere near a full load on four cores, the fact that all 8 threads are showing higher levels using the ATI cards shows that the game does hit more than just 2 cores.

 

 




  1. Introduction
  2. Closer Look (Continued)
  3. Closer Look (Continued)
  4. Testing: Setup & Results
  5. Conclusion
Random Pic
© 2001-2014 Overclockers Club ® Privacy Policy

Also part of our network: TalkAndroid, Android Forum, iPhone Informer, Neoseeker, and Used Audio Classifieds

Elapsed: 0.0244410038