Performance Analysis
Let's begin by looking at the big picture. Here is a graph showing the average relative performance when using the Phenom II with one, two, three, and four CPU cores enabled:
The blue bar represents game performance, the green bar represents application performance, and the dark bar represents synthetic performance.
First off, let's look at the gaming results. We get the impression that, compared to a single-core CPU, there is a huge benefit to using at least a dual-core CPU when gaming. After that, increasing the number of CPU cores provides a small increase in performance, with triple-core CPUs providing the same performance as a quad-core processor.
When we scrutinize application performance, we see a more linear progression, suggesting that multithreaded apps are better-suited to take advantage of multiple CPU cores. Do keep in mind that certain applications showed no increase in performance, as they aren't threaded. So, the programs you run will directly influence the actual performance increase.
The synthetic results appear as slightly exaggerated application results, which bodes well for the argument that synthetic benchmarks are a useful tool in looking forward--here, measuring multithreaded performance.
It is interesting to note that once four CPU cores are used, both application and game performance run about twice as fast as they do on a single-core CPU.
Now that we've examined this, let's look at the average performance graph from the previous article, where we used the Core 2 Quad Q6600 as our test case:
In general, the results paint a similar picture, even though some of the details are a little different. The most notable change from the Phenom II results is the synthetic results, which seem far too optimistic in the Core 2 Quad tests. Application performance is similar on the whole, showing comparable increases with each CPU core that is utilized. Even the game results are fairly close. Therefore, we can say with some certainty that the split cache on the Core 2 Quad Q6600 didn't affect the results enough to invalidate them in the previous review.
The other notable tidbit we learned wasn't too much of a surprise: as the concurrent application benchmark demonstrated, even though triple- and quad-core CPUs might show nearly-identical results in a given benchmark, these results can drastically change when more than one application is run at the same time. If you have a quad-core CPU, you're more likely to be able to run an application in the background without as much of a performance penalty. Conversely, folks with fewer CPU cores at their disposal may want to pay attention to the apps running in the background and terminate them if they're participating in a CPU-intensive task, like a game.
