We've completed two tests with each of our three games - a raw FPS test using various iterations of Fraps-timed runs and also Fraps-gathered data relating to frame times. The frames per second is hopefully self-explanatory - the higher the better. However, with frame times, lower is better, as it's the time it takes to render each frame. The average frame time is just that - the average time it took to render frames throughout the benchmark. The 1 percent / 99th Percentile is perhaps a more important figure, though, as it shows the frame rendering time you can expect for 99 percent of the time. Anything less than 40-50ms is generally considered fine and probably won't be noticeable in games, but concerns were raised over the Pentium G3258's poor performance here, so we felt it was important to at least cover it, even if not in masses of detail.

Frame Rates

There's a very short answer here and that is that there's essentially no difference in FPS between the Core i3-7350K and our makeshift Core i5-7400, so any bottlenecking due to two cores vs four cores will be happening either in different games or with much more powerful graphics cards, if at all. The only exceptions were a slightly faster result for the Core i3 in The Witcher 3: Wild Hunt and a single frame per second slower result in Deus Ex: Mankind Divided, which is the only game we tested in DX12.

Frame Times

The frame times were a little more interesting but still a somewhat moot point. The makeshift Core i5 did have slightly lower (better) results in a couple of areas, but it was slight and simply wouldn't be noticeable in games. Admittedly, even the 0.1 percent times - the larger frame times - were very similar and rarely ventured above 30ms for either CPU. Again, admittedly, the Core i3 did stray above 40ms twice and the makeshift Core i5 did not, but these were single frames in minute-long benchmarks. We should note that as Fraps doesn't work with Deus Ex: Mankind Divided in DX12 mode, we had to switch to DX11 to obtain the frame time data for both setups.

Conclusion

We think this proves, fairly conclusively, that while the Core i3-7350K only has two cores, the massive frequency advantage you can achieve over a similarly priced multiplier-locked Core i5 not only makes up for the core deficit but makes it faster in several instances too. It was quicker in Cinebench, which was actually a surprise given how it loves cores and threads, as well as PCMark 8's photo editing test, which is also heavily multi-threaded.

The video encoding test is perhaps the stand-out benchmark, though, as it represents a program that will only make significant use of up to two cores, which is still what vast swathes of programs out there adhere to, unfortunately. In games, there's very little difference, both in terms of frame rates and frame times, with just two tiny niggles in single frames across nearly three minutes of benchmarking for the Core i3 - not worth worrying about.

Again, though, we're not saying that the Core i3-7350K is the best CPU out there or the best value. The Core i5-7600K is considerably more powerful - when it's overclocked it would smash the Core i3 into next week in anything multi-threaded, and if you can stretch to the extra £70 or so then you should. Equally, if none of the Kaby Lake or Z270 offerings take your fancy and you find an older generation locked quad-core for a few notes less than the Core i3-7350K's price tag, then again it's worth considering, as is opting for a locked Core i5 and cheaper non-Z270 motherboard. However, there are some decent Z270 boards out there for just £114 now, and we've had personal experience with Asus's Prime Z270-P, which overclocked as well as any board we've tested so far.

For those on an even tighter budget, a locked Core i3 is the way to go, but we hope this paints a clearer picture of why the Core i3-7350K is actually a decent CPU and should be considered over a locked Core i5, at least if you're up for a bit of overclocking.