Now that we know what we’re comparing, it’s time to see how these CPUs perform when clocked to the same frequency.
From the results we can clearly see that the “Skylake” micro-architecture used in the Coffee Lake i3 8350K has an advantage, even when its clock speed is the same as Ryzen. In cinebench the Skylake core is 13% faster in both multi-threaded and single-threaded scenarios. Moving on to SuperPi, the Intel CPU core manages to complete the 1M calculations of pi on a single thread 16% faster than the AMD core. In Geekbench 4 we see the i3 pull ahead by up to 23% in multicore and 19% in single core.
Opening up 3DMark, I ran Firestrike, and both editions of Timespy to evaluate the performance difference. In Firestrike CPU Physics, interestingly, we see both CPUs produce the exact same score (within margin of error) but the i3 takes the lead by 17% in both Timespy and Timespy Extreme. Moving on to Unigine Superposition, I ran two benchmarks one at a CPU bottle-necked scenario of 720P Low and one with 1080p Medium to see if both CPUs could keep the GTX 1070 Ti fed. In 720p Low, we see the quartet of Skylake cores generate a massive 42% lead over the quad-core Zen, but both CPUs can easily keep the scene GPU bound when the settings rise to 1080p medium.
To start off the gaming tests, we see the i3 8350K pull ahead of the Ryzen by 45% at the lowest settings the game allows. In Sleeping Dogs, the gap is lower at 27% but the i3 is still far ahead. Ashes of the Singularity is a great example of a well-threaded game using the latest DX12 API, and the Intel part continues its lead here with a result that is 25% higher than the Ryzen 3 1200.
What does this mean?
Clearly, the Intel part is faster one a per clock basis in the tests I have used, and especially in games. This advantage could be due to the relative immaturity of the “Zen” architecture and the extensive optimisations for Intel processors over many years. Since Skylake has been with us in some form or another since 2015, it is a well known quantity. That said, I was quite surprised by just how far ahead the Intel core is in these tests. But one thing I will underscore is, for the vast majority of gaming experiences, the difference is purely academic as Ryzen, when it matters, can easily push more than a hundred FPS without problem.
After the initial run of tests it became clear that the Skylake cores in the i3-8350K have a slight edge in IPC over the Zen cores of the Ryzen 3 1200, but are quite close in non-gaming tests. That said, it came to my attention that the performance difference in Cinebench was quite a bit larger than what AMD initially stated with the launch of Ryzen. The difference between Zen and Kaby Lake was a mere 6.5% according to AMD. Although I re-tested many times to come to my results, I decided to set the systems up for a re-match, using the same Crucial DDR4 memory modules in both systems running at 2400 MHz.
In addition, I performed some more benchmark tests with popular software to add further insight. Taking into consideration some feedback I received, I also tested three more games at higher resolutions and settings that would be realistic for these types of CPUs.
Addendum – conclusion
In the Cinebench retest, the gap is slightly closer at 11%, but still greater than AMD’s claim. I wonder if this is something to do with the configuration of the Ryzen 1200’s silicon. It would seem that PCPerspective also has similar results.
Corona Render is a benchmark where CPUs are tasked with Ray-Tracing a predefined image, and the time taken to complete this is measured. The i3-8350K continues its advantage, completing the benchmark a whole minute faster than the R3 1200 at the same frequency. This translates into 16.6% higher Rays traced per second.
CPU-z’s built-in benchmark is a handy tool to measure the performance of a processor. In this test we see the two processors produce very similar results in multi-core, with the Skylake cores coming out a mere 3.8% higher. In single-core, however, the Ryzen chip falls behind by 16.7%. This is surprising, and seems to indicate that Ryzen has superior multi-core scaling, at least in this bench.
The results in the gaming benchmarks follow the trend with the Coffee Lake part producing higher frames per second, but the Ryzen 3 1200 produces a perfectly playable result nonetheless.
To conclude this update, I’d like to put into perspective that the closest competitor to the Ryzen 3 1200 in the Coffee Lake series, is the quad-core i3-8100. This part is locked to 3.6 GHz on all cores and features a cut-down L3 cache of only 6MB. The Ryzen 3 1200 can routinely reach 3.8-3.9 GHz with a little overclocking and the extra frequency would help to mitigate the slight deficit in IPC.
In addition, if it were not for AMD’s Ryzen, we wouldn’t even have quad-core i3s. And with Intel seemingly unable to squeeze any more performance per clock out of their architecture, AMD has plenty of time to close that gap. But for now, the Blue team has the performance Crown.