Steve Smith talks about the primary differences between the synthetic benchmark and the real-world gaming performance we all see, and how the difference is generated.
Episode #7-23 released on February 4, 2017
For those enthusiast, out there like me, we may ask ourselves periodically why our computers perform better in those benchmark applications than in real game situations, and today, I will explain why this is so.
First, let's talk about benchmark engines. Those engines are nothing like game engines. While, they are based on or use game engines, they tend to force the highest possible settings on the computer to rate it against other computers. This is a good way to determine how the overall build will, in theory, do against other computers. These engines are usually developed with everything in mind, from different kinds of drives, graphics cards, CPUs, etc. Even with different types of configurations in mind, like SLI, and Crossfire. That is why we can use a benchmark to determine if the build is good or bad, in theory.
Second, and why do I say in theory? Game engines make it easier to make games, and you could easily presume that a series of games made with the same game engine, should perform relatively the same. This, however, can be far from reality. Differences in the way the game is ultimately coded, number of draw calls, number of players, AI characters, etc. can change the way the games perform in different situations, and from one game to the next. Plus, the very fact that developers must code support for SLI and Crossfire means that some higher end computers may suffer from a lack of performance simply because the developer didn't code for their configuration.
Third, since benchmarks often test with maximum settings applied, and most gamers may not apply the maximum settings, the game itself may run faster or slower depending on many different user configurable and default settings that are applied.
Therefore, benchmarks only serve has an accurate metric of comparing a computer to another, not a completely accurate means of testing for game performance. Especially, if your computer rig falls outside the normal configuration that game developers code for.
Host : Steve Smith | Music : | Editor : Steve Smith | Producer : Zed Axis Dot Net