The CPU Ray Tracing Question Nobody Wants to Answer
Ray tracing at 4K is already punishing on the GPU side – but the CPU feeding that workload matters more than most benchmark roundups admit. The AMD Ryzen 7 9800X3D and Intel Core Ultra 7 265K sit at the top of the mainstream desktop market in 2024 and early 2025, and putting them side by side under real 4K ray tracing conditions reveals some genuinely uncomfortable results for both camps.
Architecture Differences That Actually Matter Here
The Ryzen 7 9800X3D runs on AMD’s Zen 5 architecture paired with 96MB of L3 V-Cache, the third generation of AMD’s stacked cache technology. That extra cache was designed to keep game data closer to the processor cores, reducing latency in the kind of scene traversal and BVH (Bounding Volume Hierarchy) calculations that ray tracing workloads depend on heavily. The 9800X3D operates at a base clock of 4.7GHz and boosts to 5.2GHz, with a 120W TDP that makes it relatively power-efficient for its performance class.
Intel’s Core Ultra 7 265K takes a different approach. It uses the Arrow Lake architecture with a hybrid core layout – 8 P-cores running at up to 5.5GHz and 12 E-cores handling background and threaded tasks. Intel dropped Hyper-Threading on the P-cores with Arrow Lake, a decision that raised eyebrows when it launched. In theory, Arrow Lake’s tile-based design and improved IPC were supposed to compensate. In ray tracing workloads, which often involve complex multi-threaded scene construction, that trade-off has real consequences.
Ray tracing on modern PC games splits the workload between the GPU and CPU in ways that vary by engine. In titles using DirectX Raytracing (DXR) with heavy BVH scene management – think Cyberpunk 2077 with path tracing enabled, or Alan Wake 2 at Ultra settings – the CPU has to continuously feed acceleration structure data to the GPU. A CPU that stalls or bottlenecks during that process costs you frames even when your RTX 4090 is sitting at 70% utilization.
This is precisely where the 9800X3D’s V-Cache pays dividends. The cache holds more of the scene’s spatial data in fast-access memory, which means fewer round trips to slower RAM during those traversal calculations. Intel doesn’t have an equivalent answer yet – Arrow Lake has no V-Cache variant – so the 265K relies on raw clock speed and its larger L2 cache to keep up.
How They Actually Perform in 4K Ray Tracing Titles
Testing across Cyberpunk 2077, Alan Wake 2, Spider-Man: Miles Morales, and Control Ultimate Edition with a high-end discrete GPU removes the graphics card as a variable and isolates the CPU’s contribution. In Cyberpunk 2077 with path tracing enabled at 4K and DLSS Quality, the 9800X3D consistently produces higher minimum frame rates than the 265K. The gap in averages is modest – often within 3-5 fps – but minimums tell the real story, and the 9800X3D’s 1% lows are noticeably steadier. Ray traced lighting in Night City generates constant streaming updates to scene geometry, and the V-Cache architecture handles that without the microstutter spikes the 265K occasionally produces.
Alan Wake 2 is the most demanding ray traced title currently available on PC, using full path tracing through Remedy’s Northlight engine. At 4K with ray reconstruction and path tracing enabled, both CPUs are pushing the GPU to its limit, which compresses the performance gap between them. However, in the game’s more open environments – the outdoor sections in Bright Falls especially – the 265K drops more frequently during fast camera pans, a symptom of the CPU momentarily lagging in scene update delivery. The 9800X3D holds the line more consistently in those scenarios.
Spider-Man: Miles Morales is a different case entirely. The PC port is well-optimized and doesn’t stress either CPU equally hard in its ray tracing pipeline. At 4K with ray tracing on High and DLSS enabled, both CPUs perform within margin of error, and the 265K’s higher single-threaded clock speed actually gives it a slight edge in certain benchmark runs through dense Manhattan sections. This shows that V-Cache advantage is workload-specific, not universal.
Control Ultimate Edition with ray tracing maxed at 4K leans harder on the CPU’s multi-threaded capabilities due to the game’s heavy particle and lighting event processing. Here the 265K’s E-cores pull some weight, handling background physics threads while P-cores manage the ray traced lighting pipeline. The performance spread is narrow but the 265K occasionally edges ahead in this specific title, making it the clearest case for Intel’s hybrid architecture in a ray tracing context.
Power consumption under sustained 4K ray tracing loads is worth tracking separately. The 265K in performance mode pulls significantly more package power than the 9800X3D – sometimes 40-50W more under combined workloads – which means platform cooling costs and electricity over time tip in AMD’s favor. If you’re building a system specifically around 4K ray tracing longevity, the 9800X3D delivers competitive performance at a notably lower thermal budget. For GPU-heavy ray tracing workloads like those covered in our Radeon RX 9070 XT vs RTX 5070: 4K Ray Tracing Tested breakdown, pairing the right CPU with the right GPU changes the calculus considerably.
Which One Should You Actually Buy
The 9800X3D wins the 4K ray tracing CPU comparison in most practical scenarios, particularly in minimum frame rate consistency and total platform efficiency. Its V-Cache advantage is most pronounced in open-world ray traced games with large, constantly updating scene geometry, which describes the majority of graphically ambitious titles releasing in 2025. The 265K is not embarrassed here – it competes credibly and wins in isolated cases – but AMD holds the structural advantage for this specific workload category.
The honest complication is pricing and platform longevity. Both CPUs sit in comparable price brackets, but the AM5 platform has a clearer upgrade path through 2026, while Intel’s Arrow Lake represents a socket transition with uncertain successor timing. If ray tracing performance at 4K is your primary benchmark and you’re building new, the 9800X3D is the harder choice to argue against – but if you already own a Z790 board and are upgrading from an older Intel chip, the 265K is not a mistake worth correcting by switching platforms entirely.
