The 4K Ray Tracing Problem That CPUs Actually Control
Ray tracing at 4K resolution is one of the few gaming scenarios where GPU bottlenecks are so severe that CPU differences should, in theory, disappear entirely. The graphics card is working so hard pushing millions of rays through complex scenes that the processor sitting next to it rarely gets a chance to become the limiting factor. And yet, when you drop an AMD Ryzen 7 9800X3D and an Intel Core Ultra 7 265K into the same test rig with the same GPU, the numbers still diverge in ways that matter to anyone spending serious money on a high-end build.
This comparison uses both chips paired with an RTX 4090 across a range of titles that stress ray tracing workloads heavily – Cyberpunk 2077 with Path Tracing enabled, Alan Wake 2, and Control Ultimate Edition. All tests run at native 4K without upscaling to keep the results honest. The 9800X3D runs on an X870E board with DDR5-6000 in EXPO mode, while the 265K sits on a Z890 board with DDR5-6400 in XMP mode – both configurations representing realistic enthusiast setups rather than cherry-picked overclocks.
Cyberpunk 2077 Path Tracing: Where the Gap Appears
Path Tracing in Cyberpunk 2077 at 4K is arguably the most punishing ray tracing workload available on consumer hardware right now. Even an RTX 4090 struggles to hold 60fps without DLSS Quality mode, and both CPUs spend most of the benchmark in GPU-limited territory. At the benchmark sequence through the city center, the 9800X3D averages around 54fps while the 265K comes in at 51fps – a gap small enough to be practically invisible during play but consistent enough across multiple runs to be real rather than noise.
The more revealing number is 1% lows. The 9800X3D holds its floor at around 44fps during the benchmark’s most complex segments, while the 265K dips to approximately 39fps in the same moments. That five-frame gap in minimum performance is where the 3D V-Cache architecture earns its premium. Cyberpunk’s path tracing workload generates enormous amounts of shader data and BVH traversal information that benefits directly from the 9800X3D’s 96MB of L3 cache sitting directly beneath the compute die. The 265K’s Lion Cove cores are genuinely fast at single-threaded workloads, but they cannot compensate for the cache-bandwidth advantage when the GPU is feeding data back to the CPU as quickly as it does in path-traced scenes.
Alan Wake 2 and Control: Different Stories, Same Winner
Alan Wake 2’s Full Ray Tracing mode tells a slightly different story. The game’s scene complexity is high but its CPU communication overhead per frame is lower than Cyberpunk’s path tracing implementation, which means the gap between the two processors narrows. The 9800X3D averages around 61fps at 4K Full RT, and the 265K is only about two frames behind at 59fps. At that level of parity, most players would never notice the difference on a 4K display.
Control Ultimate Edition with High Ray Tracing at 4K flips the script slightly in terms of framing, but not in terms of outcome. Control’s RT implementation stresses the GPU less than the other two titles, which means more CPU headroom is available – and when CPU headroom opens up, the 9800X3D’s architectural advantages become more visible again. Average framerates here are higher across both chips, sitting in the mid-80s, but the 9800X3D maintains a consistent 5-7fps lead across the benchmark sequence.
What connects all three titles is consistency. The 265K is not performing poorly – it is a genuinely fast processor, and in rasterization workloads it trades blows with the 9800X3D regularly. But across every ray tracing test at 4K, the Intel chip never wins. The margin shifts between games depending on how much the specific RT implementation involves CPU-side BVH management and shader streaming, but the direction of the gap stays the same throughout all three titles tested here.
This is worth understanding at a mechanical level. Ray tracing in modern engines is not purely a GPU task. The CPU is responsible for setting up acceleration structures, handling scene graph updates for dynamic objects, and managing driver-level communication about which rays to cast. Games with highly dynamic scenes – moving NPCs, destructible environments, complex lighting rigs – create more CPU work per frame than static or semi-static scenes. The 9800X3D handles that additional work with less latency because the data it needs is more likely to be sitting in cache rather than being fetched from main memory.
Thermals and Power During RT Loads
Both chips run hotter under sustained ray tracing loads than they do in rasterization benchmarks, though for different reasons. The 9800X3D draws less power overall – sitting around 120W during the Cyberpunk path tracing benchmark – but its 3D V-Cache stack creates thermal density challenges that make cooling choices more consequential. A 240mm AIO is technically sufficient for the 9800X3D, but a 280mm or 360mm keeps temperatures comfortably below the chip’s thermal limits during extended sessions.
The 265K draws considerably more power during the same workloads, pushing past 180W in sustained ray tracing scenarios, and its performance depends meaningfully on that power budget being available. Builds using entry-level 850W PSUs will be fine, but the 265K’s total platform power draw – CPU plus a 4090 – peaks high enough that anything below 850W becomes a concern. The 9800X3D’s lower power consumption gives it an edge in builds where efficiency matters alongside raw performance.
Who Should Buy Which Chip
If 4K ray tracing is a primary use case – not just an occasional toggle, but the actual daily gaming mode – the 9800X3D is the cleaner choice. Its lead is not so dramatic that it makes the 265K look bad, but it maintains that lead consistently across every title tested, and its lower power draw is a genuine bonus in a system already stressed by an RTX 4090’s thermal output.
The 265K makes more sense in a build where gaming splits time with content creation, video encoding, or software development. Its hybrid architecture handles multi-threaded workloads more efficiently than the 9800X3D, and outside of cache-sensitive gaming scenarios it trades performance closely enough that the gaming disadvantage feels like a reasonable trade. For a pure gaming rig pointed at 4K RT, though, there is no test result in this comparison that argues for choosing Intel over AMD.
One factor that sometimes gets lost in head-to-head CPU coverage is platform trajectory. AMD’s AM5 socket has a roadmap that extends through future Ryzen generations, giving the 9800X3D a longer upgrade path without a board change. Intel’s Z890 platform is newer and will support future Arrow Lake and Panther Lake desktop chips, but Intel’s historical track record on socket longevity makes that promise harder to bank on. For buyers who plan to swap CPUs before replacing the entire platform, that distinction matters as much as any benchmark number. Those who’ve already been exploring AMD’s current stack at different resolutions may want to check out our coverage of the same matchup at 1080p with ray tracing enabled to see how the gap shifts when the GPU is under less pressure.
The 9800X3D’s advantage in 4K ray tracing is real, repeatable, and rooted in an architectural feature – 3D V-Cache – that Intel has no direct answer to in its current desktop lineup. Whether AMD’s next-generation chips keep that advantage or whether Intel closes it with a competing cache solution will define the next chapter of this matchup, but as of this generation, the cache wins.
