Sign in with
Sign up | Sign in

High-Detail Benchmarks

Crysis 3 Performance, Benchmarked On 16 Graphics Cards
By

We're skipping over the Medium detail preset and going straight to High. In addition, we're enabling the Lens Flares option and setting Motion Blur to Low, with 8x AF and 2x SMAA added to the mix for texture filtering and anti-aliasing.

At 1680x1050 with these detail settings, the Radeon HD 7770 is too slow for playable performance, and the GeForce GTX 650 Ti barely averages more than 30 FPS. The rest of the cards manage the load much better, achieving at least 30 FPS minimums and averages in excess of 40 FPS.

For one reason or another, the Radeon HD 7950 with Boost is able to achieve a higher minimum frame rate than the rest of the cards.

A look at the frame rates over time shows how those average and minimum frame rates came to be. It's nice to see that the GeForce GTX 650 Ti remains above 30 FPS for the majority of this demanding benchmark.

Frame time variance is acceptable across the board. It's interesting to note that the GeForce GTX 670 incurs a higher average variance than most of the other cards.

Surprisingly, the higher resolution doesn't affect our results all that much. The GeForce GTX 650 Ti gets pushed slightly below 30 FPS, but the rest of the cards remain playable.

We've already seen the mysterious bottleneck that pulls performance down at the end of our run to 33 FPS or so, so we don't really learn anything new from the frame rate over time chart.

The frame time variance remains below 20 ms at this setting.

Ask a Category Expert

Create a new thread in the UK Article comments forum about this subject

Example: Notebook, Android, SSD hard drive

Display all 8 comments.
This thread is closed for comments
  • 1 Hide
    mi1ez , 5 March 2013 19:23
    All I can see in those image comparisons is the GIF artefacts. Is there not a better way?
  • 1 Hide
    sam_p_lay , 5 March 2013 21:29
    mi1ezAll I can see in those image comparisons is the GIF artefacts. Is there not a better way?


    +1. Low/Med/High (or whatever) rollovers to switch between PNGs would be nice. Good to give the user the ability to switch between what they want at their own pace instead of waiting for their frame to come back around.
  • 2 Hide
    marshallbradley , 5 March 2013 21:35
    According to the last graph on the first page, at 8x MSAA the Min FPS is higher than the average FPS. Eh??
  • 0 Hide
    bemused_fred , 6 March 2013 00:22
    I can see this game appearing in a lot of benchmarking suites.....
  • 0 Hide
    woodscrews , 12 March 2013 09:43
    interesting, would have been nice if you benched the hardware on med settings too, and maybe tested the latest AM3 athlon x4 seeing how its exelent value at the low end right now
  • 0 Hide
    chriss000 , 13 March 2013 22:51
    There arent many who would shell out for a big system upgrade to play this game.
    It looks nice but not 200 quids worth.
  • 0 Hide
    doveman , 14 March 2013 11:36
    My TV only really supports 1280x720 so I'm playing at that with V.High Textures, 4xMSAA, 16xAF and most settings at V.High or High except Particles and Shadows I think as they were crippling my fps from 60 to 30 but now it's mostly at 60fps, with some occasional dips no lower than about 45fps.

    I'm only running a Phenom II X4 955 @ 3.6ghz.
  • 0 Hide
    chriss000 , 14 March 2013 16:33
    Yes Doveman, how reiivant are 1080p benchmarks when more than half of all games pc's
    are hooked up to cheap monitors or bedroom size tv's?
    My donated lcd only runs at 1020x768 ! Looks ok to my old low deff peepers tho.