A well thought out test course
The editors of Hardware.Info know better than anyone that the battle between AMD and Intel is a sensitive one, especially since AMD once again plays a significant role in the market for desktop processors. That's why we put a lot of time into developing a test protocol for this article that is as fair as possible. For starters, we test with a lot more games than in regular CPU tests, but we have expanded this test in even more ways.
It all starts, of course, with putting together comparable test systems. That means we've looked less at what processor manufacturers say than usual, but more at what you'd probably buy if you put together a system around two of today's most popular processors: the AMD Ryzen 5 2600X and Intel Core i5 8600K. At the time of writing, the AMD is on average a couple tenners chaeper than the Intel, but those prices vary slightly.
Both systems have been equipped with a motherboard that is potent enough to run the processor at full power. Furthermore, we have provided each configuration with 16GB of DDR4 memory at 3000 MHz (with CL15 timings), a Samsung 850 Evo 1TB SSD, a Seasonic power supply and a NZXT Kraken X62 water cooler. For the OS we installed the latest version of Windows 10, with everything up to date for the Spectre and Meltdown patches.
We haven't talked about the video card yet. On both the AMD and Intel systems, we ran all the tests twice: once with an AMD Radeon RX Vega 56 and once with a Nvidia GeForce GTX 1070 Ti, two video cards that are generally equally as fast. Moreover, these video cards seem to us to be a logical combination with the processors tested.
|Processor||AMD Ryzen 5 2600X||Intel Core i5 8600K|
|Motherboard||ASUS Crosshair VII Hero||ASRock Z370 Extreme4|
|Memory||G.Skill Trident Z 16GB DDR4-3000 @ CL15-15-15-36|
|GPU||AMD Radeon RX Vega 56 (18.5.1) & Nvidia GeForce GTX 1070 Ti (397.93)|
|SSD||Samsung 850 Evo 1TB|
|Power supply||Seasonic Prime Titanium 650W|
|CPU cooler||NZXT Kraken X62|
|OS||Windows 10 April 2018 Update (1803) incl. Spectre & Meltdown patches|
Resolutions and settings
With regards to the resolutions and settings that we test at, we also looked at the most logical usage scenarios. That's why we left out Full HD on medium settings and Ultra HD on Ultra settings; for the former this hardware is actually much too fast, for the latter the hardware is just too slow. This means that we have carried out our tests on the following settings:
- Full HD (1920x1080) on ultra settings
- WQHD (2560x1440) on medium settings
- WQHD (2560x1440) on ultra settings
- Ultra HD/4K (3840x2160) on medium settings
As mentioned above, we selected twelve games to draw a conclusion on the gaming performance of the Intel and AMD processors. We tried to make a good mix of new and popular games, the different APIs (DirectX 11, DirectX 12 and Vulkan) and the different game genres, such as shooters, open-world games and racing games. In the end, we came up with the following titles:
- Assassin's Creed: Origins (DirectX 11)
- Battlefield 1 (DirectX 12)
- Destiny 2 (DirectX 11)
- Far Cry 5 (DirectX 11)
- Forza Motorsport 7 (DirectX 12)
- Ghost Recon: Wildlands (DirectX 11)
- GTA V (DirectX 11)
- Prey (DirectX 11)
- PlayerUnknown's Battlegrounds (DirectX 11)
- The Division (DirectX 12)
- Total War: Warhammer (DirectX 12)
- Wolfenstein II: The New Colossus (Vulkan)
In addition to the frame rates (fps), we also logged the individual rendering times of each frame for each test. If the CPU performance is a bottleneck, it is sometimes more apparent from the frame times than from the fps. There will not necessarily be fewer images per second shown, but they will appear less frequently on your screen and you will experience this as small hiccups (stuttering). Below the graphs you can see a table with the 99th percentile, i.e. the limit of the 1% slowest frames. The frame times are expressed in milliseconds; lower = better.