In a rather interesting observation originally made by NerdTech, it seems that Intel CPUs scale slightly better with NVIDIA RTX 30 series GPUs compared to the Radeon RX 6800/6900 lineup and the opposite is true for the Ryzen 5000 CPUs. We checked a few benchmarks from some reputable sources and the same seems to hold true across most games. Although there is no dearth of reviewers using the RTX 3080/3090 to test the 11th Gen Rocket Lake-S lineup, we could only find one reliable source using the Radeon RX 6800/6900 for the CPU benchmarks:

Let’s have a look at the Borderlands 3 benchmark at 720p. As you can see, ComputerBase’s benchmark sees the 11th Gen CPUs namely the Core i5-11600K, Core i7-11700K, and Core i9-11900K (and the older 10900K) take the lead with the Ryzen 5000 processors following shortly behind. CB used the NVIDIA RTX 3080 for the review.

Looking at Igor’s benchmark of the same game at the same resolution using the AMD Radeon RX 6900 XT, you can see that the averages are notably higher across the board. Furthermore, the Ryzen 500 lineup including the lowest-end Ryzen 5 5600X is faster than every Rocket Lake-S chip with even the 11900K being within an inch of it.

At 1080p, the deltas are almost negligible between the various CPUs but the 11th Gen Core lineup moves to the top of the rug, at least in CB’s review. In Igor’s review, the AMD Ryzen 9 5900X and 5950X are the fastest parts, followed by the Core i7-11700K and 11900K. The Ryzen 7 and 5 come in next with the Core i5-11600K at the end. This is primarily the result of a GPU-bound scenario so don’t take these too seriously. However, the general trend is the same: The Ryzen 5000 CPUs are faster with Radeon GPUs while the Intel chips tend to perform better with NVIDIA parts.


Shadow of the Tomb Raider is one game that scales really well with CPU core counts and yet we see the same behavior once again. The Ryzen 5000 lineup (including the hex-core Ryzen 5 5600X is notably faster than the Core i5-11600K, 11700K, and 11900K with a Radeon RX 6900 XT while with an RTX 3090, the Rocket Lake-S parts are generally faster, both at stock and when overclocked.

It’s hard to tell whether this behavior is due to the BIOS of the Rocket Lake parts or the Re-Sizable BAR implementation with Radeon vs GeForce parts, or so other reason, but it’s certainly there, and it’s more than noticeable in some cases. It’d be really nice if someone could test it with both configurations. We’d do it but the complete lack of retail and review samples in our region makes that hard. Plus, it really looks like Intel doesn’t want to give us a sample, so that’s there too.