View Single Post
Old 08-18-10, 01:52 PM   #11
ragejg
nV News Alumni
 
ragejg's Avatar
 
Join Date: Sep 2002
Location: Finger Lakes of Hell, NY
Posts: 5,399
Send a message via AIM to ragejg
Default

... A Post Script

GTX 460 Performance using the Lost Planet 2 Benchmark

Keeping with the tradition of this review being heavy on MT Framework titles, I have decided to share current performance information for the EVGA GTX 460 SC 768MB and the Sapphire Radeon HD 5830 when running the new Lost Planet 2 Benchmark.

Set on the same fictional planet as the first game, Lost Planet 2 aims to please fans of the original with uniquely smooth gameplay and an imaginatively created environment featuring startlingly huge and odd-looking enemies. I personally remember the first couple times I played the first game, and the sheer scale and fearfulness of some enemies actually made my heart beat a little fast at times.

This benchmark is not a playable demo, but aims to show the viewer how their respective system can run the game. Like the first game, there are two separate code paths, and in this game's case it's DirectX 9 and DirectX 11. Upon first inspection it appears that this revision of the MT Framework Engine utilizes many of the same effects as Resident Evil 5, and the DirectX 11 code path adds user-adjustable degrees of hardware tessellation.

Tessellation on (DirectX 11):



Tessellation off (DirectX 9 or DirectX 11 with DX11 Features turned to Off):



As you can see, at least in some areas of the game the difference is dramatic. The difference is visible in other areas of the game as well, like in rendered water which is also tessellated, and some other alien creatures have tessellated features of their bodies.

Now we're going to take a look at the performance of the EVGA GTX 460 SC 768MB when running this benchmark, as well as the performance of its ATI competitor, the Sapphire Radeon HD 5830. The system is the same as what was used in the review (Phenom II x3 at 3.5ghz, 4gb DDR2-800), but there are a few differences:
  • New Vista Ultimate Install
  • HD 5830 now uses Catalyst 10.7
  • GTX 460 is running at optimally overclocked frequency of 840/3900
  • HD 5830 is running at optimally overclocked frequency of 875/4800 (vs 800/4000)

Since this IS a demanding game I thought it best to test these cards in overclocked states to obtain the best possible framerate for the demanding resolution of 1920x1080. All settings are at their highest except where denoted in chart titles, and no AA or AF is used, as performance would be dragged down too far.

GTX 460 AND HD 5830 used highest quality texture filtering. Nvidia card used no optimization as it's not used by default (and most people don't use it), and the ATI card made use of the standard mode of Catalyst AI. I don't personally see Catalyst AI as any sort of "cheating" optimization, since in standard mode there's little to no image quality difference, and the PC hardware enthusiast community in general prefers using the standard mode. I didn't use the advanced mode due to image quality and over-optimization concerns, as well as suspicions of additional CPU usage. So am I not comparing apples to apples? It is such a slippery slope, so what I believe I'm doing is running both cards in the modes most preferred by their respective users.

Test A

The primary purpose of Test A is to give an indication of typical game play performance of the PC running Lost Planet 2. (i.e. if you can run Mode A smoothly, the game will be playable at a similar condition). In this test, the character�'s motion is randomized to give a slightly different outcome each time.

Test B

The primary purpose of Test B is to push the PC to its limits and to benchmark the maximum performance of the PC. It utilizes many functions of Direct X11 resulting in a very performance-orientated, very demanding benchmark mode.


Results




Benchmark data:




Benchmark data:




Benchmark data:




Benchmark data:


(ADDED 8/20/10) DirectX 9 Performance:

I had inferred that DirectX 9 performance in this benchmark would basically be the same as running the DX11 codepath with DirectX11 features turned off. I recently ran the DX9 codepath to see if this was true:


Benchmark data:


Given the amount of activity and the sheer volume of visuals being shoved through the monitor at you, both cards don't seem to perform all that bad, especially given the fact that they're both not high-end, and the system they're in isn't really high-end either. But clearly, the win on this goes to the GTX 460 in DX11 mode. It has less of a problem delivering high amounts of tessellation, and overall (in this specific scenario) delivers performance around 10-40% better than the HD 5830.

It is worth mentioning that if you're not a fan of the tessellation effects in this benchmark/game, the DX9 path may be a viable alternative that will afford you some more FPS. It is interesting to note here that the HD 5830 takes the lead when running the benchmark in this path.


This benchmark is just another example of the GF100/GF104 doing what it does best - tessellation. It is clear that NVIDIA has put this feature high on their priority list, and as more and more games come out that use tessellation, ATI Radeon products will have a hard time keeping up in games that use full DX11 features until they develop something that can compete better on this specific feature.

It is worth mentioning, however, that this benchmark was originally made available at NVIDIA's website, which may indicate that while NVIDIA has been able to make optimizations for this upcoming game, ATI has not. Since this is a benchmark that has been released two months before the actual game release, there is a possibility that performance differences for the final product might be a little different.



Last edited by ragejg; 10-22-10 at 02:25 AM.
ragejg is offline