Why Graphics Settings Matter When Benchmarking CPUs!
In my "Should You Buy a Ryzen 7 9700X or 7800X3D?" video, I discovered that Hardware Unboxed were conducting their CPU testing at Ultra graphics settings, which resulted in a very small performance difference between the 9700X and 7700X. At the time, I pointed this issue out to Hardware Unboxed, but they chose to ignore it … up until very recently, when they started including 1080p medium results as well. This is strange, because it shows that Steve now understands his mistake, but for some reason refuses to address it properly. The issue becomes clear when you look at the impact of graphics settings on performance. If we look at Call of Duty Black Ops 6, and compare the 1080p performance at different graphics settings, you can see that at high settings, the game is clearly GPU bound, whereas at low settings, the game is clearly CPU bound, which results in a massive 42% increase in performance. Running the same benchmark at medium settings helps, but the game is still heavily GPU bound, so it makes absolutely no sense to use medium settings to compare the relative performance of CPUs. To emphasize this point, if we compare two different CPUs in the same game, you can see that, not only does the difference in performance between the CPUs significantly reduce as the graphics settings are increased, but you would also get a wildly different conclusion as well. This is why it’s important to heavily load CPUs when testing them, and why you can’t trust the CPU benchmarks shown by Hardware Unboxed.
Blackbird PC Tech
Why Graphics Settings Matter When Benchmarking CPUs!
In my "Should You Buy a Ryzen 7 9700X or 7800X3D?" video, I discovered that Hardware Unboxed were conducting their CPU testing at Ultra graphics settings, which resulted in a very small performance difference between the 9700X and 7700X. At the time, I pointed this issue out to Hardware Unboxed, but they chose to ignore it … up until very recently, when they started including 1080p medium results as well. This is strange, because it shows that Steve now understands his mistake, but for some reason refuses to address it properly. The issue becomes clear when you look at the impact of graphics settings on performance. If we look at Call of Duty Black Ops 6, and compare the 1080p performance at different graphics settings, you can see that at high settings, the game is clearly GPU bound, whereas at low settings, the game is clearly CPU bound, which results in a massive 42% increase in performance. Running the same benchmark at medium settings helps, but the game is still heavily GPU bound, so it makes absolutely no sense to use medium settings to compare the relative performance of CPUs. To emphasize this point, if we compare two different CPUs in the same game, you can see that, not only does the difference in performance between the CPUs significantly reduce as the graphics settings are increased, but you would also get a wildly different conclusion as well. This is why it’s important to heavily load CPUs when testing them, and why you can’t trust the CPU benchmarks shown by Hardware Unboxed.
1 month ago | [YT] | 44