Date:
Monday , March 21, 2011
Author:
Mark Warner
Editor:
Brent Justice
Google +1

Crysis 2 Multiplayer Demo Video Card Performance

Crytek released the first Crysis 2 demo a few weeks ago, giving anxious gamers a first look at the graphics and performance its "new" engine brings to the table. We've tested the demo extensively with a handful of video cards, and we're here to share our experiences with you.

Introduction

A multiplayer demo for the highly anticipated Crysis 2, sequel to 2007's technical masterwork Crysis was released last February the 28th. The commercial version is set to release on March 22nd, but we wanted to go ahead and give our readers our impression of how this demo performs on a variety of mid-to-high-end video cards.

For this preview, we are going to use four video cards: an NVIDIA GeForce GTX 580 and GTX 570, and an AMD Radeon HD 6970 and HD 6870. These four video cards give us a price range from about $200 to about $500.

Article Image

For the NVIDIA GeForce GTX 580 and GTX 570, we used NVIDIA's ForceWare beta version 267.24. For the AMD Radeon HD 6970 and HD 6870, we used AMD's WHQL certified Catalyst 11.2 driver package.

Crysis 2 MP Demo Performance

Article Image

The Crysis 2 multiplayer demo is DirectX 9 only. Thus, there are no new DX10 or DX11 features in the demo. There is no PhysX support, and there are only a handful of graphics settings. On the game's login screen, there is a yellow disclaimer which reads "This demo is for evaluation purposes only and is not necessarily representative of the final features and quality of the software". We sincerely hope that that is the case. The first Crysis featured a nice quantity of graphics options with which to tune the game's performance. We do hope that Crysis 2 is more PC gamer friendly than this demo appears to be. After logging in and navigating to the graphics options menu, we find this:

Article Image

The "System Spec" setting has three options: Gamer, Advanced, and hardcore (in order from lowest to highest). That's all there is. Disappointing, no? There is no convenient way to tweak any individual settings. There are a few functional command line parameters to disable motion blur, change the FOV setting, and so on, but that is beyond the scope of this article. When the full game comes out, we hope there will be more options to explore. For now, the above is what we have to work with.

Testing this demo proved to be a challenge. It lacks a dedicated or 'listen' server option, so we were unable to roam around any levels on our own. Our only option was to connect to live internet games and play against other people. Generally, we avoid that scenario in our testing, as the results we get from online games are a lot less predictable and less consistent than we like to see. In this game, however, it simply wasn't an option. Our only option was to use live online games. The problem with that is threefold: First, we could not be guaranteed a reserved spot on any given server, so varying network responsiveness factored into gameplay. Second, we were not directly in control of which of the two levels or game modes we could play. Third, we were definitely not directly in control of how long each game session lasted. Some games lasted a scant 3 minutes, and others lasted much, much longer.

The basic problem with using an online-only first-person-shooter as a performance analysis tool is that it is too unpredictable. That said, we were able to play a lot of this game, and we can definitely offer some feedback on how it performs. To test the Crysis 2 MP demo, we played several matches with each video card. We were sure to test both game modes on both levels for a minimum of four completed matches per video card. In practice, it came out to more like about 15 to 20 matches per video card. While we played, we kept an eye on the FRAPS FPS counter to make sure we weren't dipping below 35FPS. For this particular demo, we found that gameplay was best when the average framerate was about 40, and did not dip below about 35. These numbers are not absolute requirements, but rather more like guidelines.

Article Image

Predictably, the NVIDIA GeForce GTX 580 finished testing at the head of the class. The GTX 580 was able to play the Crysis 2 MP demo at 2560x1600 on "Hardcore". The GeForce GTX 570 and the AMD Radeon HD 6970 exhibited nearly identical performance, and both were playable at 2560x1600 with the in-game setting set to "Advanced". The Radeon HD 6870 could not handle 2560x1600 even at "Gamer", so we turned it down to 1920x1200. At that point, the HD 6870 was playable with the game at "Hardcore".

It should be noted that traditional anti-aliasing does not function in this demo. Of course, the game does not actually have its own AA setting, but even attempted to force MSAA, either from the AMD Catalyst Control Center or from the NVIDIA Control Panel, produced no results. In spite of that, we never really felt like we were missing traditional MSAA in this game. There is actually some edge smoothing going on, which is presumably built into the engine. It does not look like traditional AA, but rather some shader-based smoothing operation. In addition to that, the post-processing going on in this game is quite extreme. Between the motion blur, the HDR lighting, and aforementioned shader-based edge-smoothing, aliasing was reduced to the degree that it was simply not a problem for us.