3DMark Vantage vs. Games
Recently Tom's Hardware posted a large set of data on today's video cards and their performance in wide variety of different games as part of their ATI Radeon HD 4870 review. There has always been discussion about synthetic benchmarks and their validity when compared to the results from benchmarking actual games. When trying to figure out what card is best for a specific game, testing that game is the obvious answer, but when determining the overall performance across wide range of games the answer is less clear. We decided to crunch some numbers and compare 3DMark Vantage scores (from Futuremark's ORB database) with the results obtained by Tom's Hardware while benchmarking a large set of games. Does a synthetic benchmark hold up against real game benchmarking? Hardware review sites often shun synthetic benchmarks, or include them with disclaimers stating that the reader should really concentrate on the results from actual games. But should they? Sure, there are many good examples where a synthetic benchmark and a specific game benchmark disagrees, but since nobody buys a video card just to play a single game, (well, except all those World of Warcraft junkies, but let's disregard that) so the interesting factor is the average performance you get with different games. Real World vs. SyntheticHere's a graph comparing the average framerate of a number of video cards while running the following games.
Resolution and other settings used for the graph were as close to the Performance Preset of 3DMark Vantage as possible. This is a very diverse set of titles with many different engines - each with their own bottlenecks that ultimately dictate the performance.
This graph shows the average performance of each card in games (average fps) and the average 3DMark Vantage score (Performance Preset). The line on the graph shows the expected performance of each card based on 3DMark score alone. As you can see, for a synthetic benchmark it's pretty much spot-on - the relative average performance of each card while running actual games is very close to what can be interpolated from the ORB database. The largest error is with the GeForce GTX 260 - the average framerate in games included is less than what could be predicted by just looking at the 3DMark Vantage score average for the card - in other words, 3DMark Vantage slightly "over-scores" this card based on this comparison. I'd chalk this up as a minor error most likely caused by the relatively small number of 3DMark Vantage results submitted with the card. Even if you ignore the excuses, the result is still within a reasonable margin of error, giving you a very good idea of the expected gaming performance.
Tags |
 
 
See if your PC can handle the latest games:
 
 |
||||||||||||||||||||||||||||||||||||||
 |
