Roblox

The pairing of an Intel Core i5-13400F with an NVIDIA GeForce RTX 5090 D in Roblox produces extraordinary frame rates that place it at the top of the database, ranking first out of 3010 tested combinations. The measured results reveal a system where the GPU is rarely the limiting factor, and the performance ceiling is instead defined by the CPU's single-thread throughput and the game's inherent engine constraints. This analysis breaks down the scaling behavior, component roles, and optimal settings based strictly on the measured data.

Resolution Scaling

The FPS drop from 1080p to 4K is dramatic but leaves the system with headroom that most hardware cannot approach. At 1080p with Low settings, the combo averages 776 FPS, which falls to 518 FPS at 1440p and further to 293 FPS at 4K. That represents a 62% reduction in frame rate from 1080p to 4K at Low settings, indicating that while resolution scaling has a significant impact, the absolute numbers remain exceptionally high.

The scaling pattern shifts when moving to higher settings. At Ultra, the 1080p average sits at 507 FPS, dropping to 338 FPS at 1440p and 191 FPS at 4K. The percentage drop from 1080p to 4K at Ultra is 62%, nearly identical to the Low preset's scaling. This consistency suggests that the game's rendering workload scales linearly with pixel count across the settings spectrum, with the RTX 5090 D's massive compute resources absorbing the additional fill-rate demands without introducing a bottleneck.

What is more revealing is the gap between settings at a fixed resolution. At 4K, the difference between Low (293 FPS) and Ultra (191 FPS) is 102 FPS, a 35% reduction. At 1440p, that same Low-to-Ultra gap widens to 180 FPS (518 vs 338), and at 1080p it reaches 269 FPS (776 vs 507). The widening gap at lower resolutions points to the CPU becoming more involved in the performance equation. The i5-13400F's boost clock of 4.60 GHz and its 10-core, 16-thread configuration handle the game's logic and draw calls, but as resolution drops and GPU headroom expands, the CPU's limits become visible in the diminishing returns at higher settings.

The data indicates that 1080p is where the CPU starts to cap performance, particularly at Low and Medium settings where the GPU could theoretically push even higher. The 776 FPS at 1080p Low versus 518 FPS at 1440p Low shows a scaling factor of 1.5x for a 2.25x pixel increase, which is actually better than linear scaling — a sign that the GPU is still dominant. However, the 4K results, where the RTX 5090 D's 32 GB of GDDR7 memory and 1.79 TB/s bandwidth come into play, show a more predictable GPU-bound behavior.

GPU Role

The RTX 5090 D is a monster of a graphics card, with a 92.2 billion transistor count on a 750 mm² die built on TSMC's 5 nm process. Its specifications include 21760 shading units, 680 texture mapping units, and 176 raster output pipelines, paired with a boost clock of 2407 MHz. The memory subsystem is equally imposing: 32 GB of GDDR7 on a 512-bit bus, delivering 1.79 TB/s of bandwidth. These numbers translate into a Passmark G3D score of 44065, placing the GPU in the 92nd percentile of all tested graphics cards.

In Roblox, the GPU's role is to render simple geometry and apply shading effects, which is not particularly demanding by modern standards. The measured FPS at 4K Ultra (191 FPS) shows that even at maximum settings and highest resolution, the RTX 5090 D is nowhere near its limits. The card's average benchmark score of 77712 puts it just 1.1% ahead of the AMD Radeon RX 6650M XT in the nearest rivals list, though that comparison is based on synthetic workloads rather than actual game performance.

The GPU's clock behavior is consistent across resolutions, with the base clock of 2017 MHz and boost of 2407 MHz providing stable performance. The 575 W TDP and dual-slot design indicate a power-hungry card, but the measured FPS data shows that Roblox does not push the card into thermal or power throttling territory. The 4K Medium results, which include a min FPS of 227 and max of 307, show a tight frame time distribution that suggests consistent GPU utilization without spikes.

The VRAM capacity of 32 GB is far beyond what Roblox requires, but it ensures that texture streaming and asset loading never become a bottleneck, even at 4K Ultra. The GPU's DirectX 12 Ultimate support and Vulkan 1.4 API compatibility are relevant for modern rendering paths, though Roblox's engine may not fully utilize these features. The Passmark DirectX scores (434 for DX9, 371 for DX11, and 219 for DX12) indicate that the card's performance varies by API, but the game's actual rendering workload keeps the GPU comfortably ahead of the CPU's output.

CPU Role

The Intel Core i5-13400F is a 10-core, 16-thread processor based on Raptor Lake architecture, built on Intel's 10 nm process with a 215 mm² die size. Its base clock of 2.50 GHz boosts up to 4.60 GHz, and it carries a 20 MB shared L3 cache alongside 1.25 MB L2 per core. The CPU's benchmark results include a Cinebench R23 multi-core score of 22604 and a single-core score of 3191, with a Passmark single-thread score of 3634. These numbers place the CPU in the 77th percentile of all processors.

In Roblox, the CPU's role is critical because the game is known for its single-threaded physics and game logic. The 3DMark single-thread score of 960 and the 2-thread score of 1880 provide insight into how the CPU handles the game's primary workload. The boost clock of 4.60 GHz is the key specification here, as higher clocks directly translate to higher frame rates in CPU-bound scenarios.

The measured data shows this CPU limitation clearly. At 1080p Low, the system hits 776 FPS, but at 1080p Ultra it drops to 507 FPS. The CPU is not the bottleneck at Low settings — the GPU is still feeding frames fast enough — but the scaling from Low to Ultra at 1080p shows a 35% reduction that is more pronounced than the 4K scaling. This suggests that at lower resolutions, the CPU's ability to process draw calls and game state updates becomes the limiting factor, particularly when higher settings add more objects and effects.

The i5-13400F's 10 cores and 16 threads provide ample parallelism for Roblox's multi-threaded aspects, such as asset streaming and audio processing, but the core game loop remains single-threaded. The Passmark physics score of 1437 and the floating-point math score of 60539 indicate strong computational capabilities, yet the game's reliance on single-core performance means the 4.60 GHz boost clock is what matters most. The CPU's nearest rival, the AMD Ryzen 9 6900HS, scores nearly identically in average benchmarks (25284 vs 25292, a 0% delta), but the Intel chip's higher boost clock gives it an edge in this specific game's workload.

How This Combo Ranks

This combination of the i5-13400F and RTX 5090 D ranks first out of 3010 tested combos in Roblox, placing it at the absolute top of the database. The ranking reflects the measured FPS across all resolutions and settings, with the combo's lowest average (191 FPS at 4K Ultra) still exceeding what most systems can achieve at any setting. The rank is a direct consequence of the GPU's overwhelming compute power combined with the CPU's strong single-thread performance.

The GPU's nearest rivals in synthetic benchmarks — the AMD Radeon RX 6650M XT (1.1% slower), RX 6850M XT (1.6% faster), and NVIDIA Tesla P100 variants (2.1-2.4% faster) — are not indicative of real-world Roblox performance. The RTX 5090 D's 92nd percentile ranking among all GPUs and the i5-13400F's 77th percentile among CPUs combine to create a system that dominates this game. The CPU's nearest rivals, including the Ryzen 9 6900HS and Ryzen 5 5600X3D, all score within 0.4% of each other in synthetic benchmarks, but the Intel chip's 4.60 GHz boost clock provides the necessary single-thread speed for Roblox's engine.

The combo's rank is particularly impressive given that the CPU is not a top-tier processor. The i5-13400F's 77th percentile means there are faster CPUs available, but none of them, when paired with this GPU, produce higher frame rates in Roblox according to the measured data. This suggests that the game's performance ceiling is reached with this CPU, and additional cores or cache would not yield meaningful gains. The GPU's 92nd percentile ranking is more relevant, as it ensures that even at 4K Ultra, the card has enough headroom to maintain high frame rates.

Settings Recommendations

Based on the measured FPS data, the optimal settings depend on the target resolution and desired frame rate consistency. At 1080p, all settings produce playable frame rates, but the Medium preset offers the best balance. The Medium setting at 1080p averages 706 FPS with a minimum of 600 FPS and maximum of 812 FPS, providing a smooth experience with tighter frame pacing than the Low setting's 776 FPS average, which has no recorded minimum or maximum. The Ultra setting at 1080p drops to 507 FPS, which is still excellent but represents a 28% reduction from Medium without a proportional visual improvement.

At 1440p, the Medium preset again emerges as the recommended choice. It averages 471 FPS with a range of 401 to 542 FPS, which is sufficient for any high-refresh-rate monitor. The High preset at 417 FPS is close behind, but Medium offers a 13% improvement in average FPS. The Ultra preset at 338 FPS is playable but begins to approach the threshold where frame drops could become noticeable on very fast displays.

At 4K, the Medium preset is the clear winner. It averages 267 FPS with a minimum of 227 FPS and maximum of 307 FPS, providing a consistent experience that is well above the refresh rate of most 4K monitors. The High preset at 236 FPS is also viable, but Medium's additional 31 FPS headroom helps absorb scene complexity spikes. The Low preset at 293 FPS offers the highest frame rate but sacrifices visual quality that the RTX 5090 D is more than capable of handling. The Ultra preset at 191 FPS is the only setting that falls below 200 FPS, and while still smooth, it offers diminishing returns over Medium.

Across all resolutions, the Medium preset provides the best combination of frame rate and visual fidelity. The data shows that the gap between Medium and High is smaller than the gap between High and Ultra, suggesting that the Medium-to-High transition offers the most visual improvement per frame rate cost.

Measured FPS Breakdown

The complete set of measured results provides a detailed picture of performance across all tested configurations.

1080p (1920x1080):

• Low: 776 FPS average
• Medium: 706 FPS average, 600 FPS minimum, 812 FPS maximum
• High: 625 FPS average
• Ultra: 507 FPS average

1440p (2560x1440):

• Low: 518 FPS average
• Medium: 471 FPS average, 401 FPS minimum, 542 FPS maximum
• High: 417 FPS average
• Ultra: 338 FPS average

4K (3840x2160):

• Low: 293 FPS average
• Medium: 267 FPS average, 227 FPS minimum, 307 FPS maximum
• High: 236 FPS average
• Ultra: 191 FPS average

The only configurations with recorded minimum and maximum FPS values are the Medium presets at each resolution. At 1080p, the frame range spans 212 FPS (600 to 812), at 1440p it spans 141 FPS (401 to 542), and at 4K it spans 80 FPS (227 to 307). The narrowing range at higher resolutions indicates that the GPU's consistent frame delivery becomes more dominant, while at 1080p, CPU-side variations cause larger frame time fluctuations.

The scaling from Low to Ultra within each resolution shows consistent patterns. At 1080p, moving from Low to Ultra reduces FPS by 269 FPS (35%). At 1440p, the reduction is 180 FPS (35%). At 4K, the reduction is 102 FPS (35%). This consistent 35% penalty for maxing out settings suggests that Ultra adds a proportional amount of rendering work regardless of resolution, and the RTX 5090 D handles it with equal efficiency.

The resolution scaling at each fixed setting shows that 4K performance is roughly 38% of 1080p performance across all settings (191/507 at Ultra, 236/625 at High, 267/706 at Medium, 293/776 at Low). This consistent ratio indicates that the system is not hitting any unexpected bottlenecks as resolution increases, and the GPU's 1.79 TB/s bandwidth and 512-bit memory bus are sufficient to feed the 21760 shading units without starvation.