Overpower

Overpower, an action title from August 2023, presents a fascinating case study in system bottlenecking when pairing a 10th-generation Intel Core i5-10400F with the flagship NVIDIA GeForce RTX 5090. The benchmark data reveals a configuration where the GPU’s immense raw power is consistently held in check by the CPU's 6-core/12-thread Comet Lake architecture, resulting in a performance profile that is surprisingly flat across resolutions. This combination ranks 635th out of 3713 tested combos, placing it in the top 17.1% of all configurations, yet the measured FPS numbers suggest the RTX 5090 is far from being the limiting factor in most scenarios.

Resolution Scaling

The measured FPS data shows an unusual pattern: performance drops are minimal when moving from 1080p to 4K. At High settings, the average frame rate only falls from 137 FPS at 1920x1080 to 131 FPS at 2560x1440, and then to 125 FPS at 3840x2160. That represents just an 8.8% reduction from 1080p to 4K, which is remarkably small for a GPU with the RTX 5090's specifications. This flat scaling curve strongly suggests the CPU is the primary constraint, as a GPU-bound scenario would typically show a much steeper decline—often 30-40% or more—when pixel count quadruples.

The Low settings data reinforces this interpretation. At Low, the system achieves 168 FPS at 1080p, 162 FPS at 1440p, and 153 FPS at 4K. The delta between 1080p and 4K is only 15 FPS, or 8.9%. When graphics settings are lowered, the GPU has even less work to do, yet the frame rate barely improves, confirming that the Intel Core i5-10400F's 6 cores, with a base clock of 2.90 GHz and boost up to 4.30 GHz, are saturating the render pipeline. The data implies that at 1080p Low, the CPU is delivering its maximum achievable frame rate of around 168 FPS, and no amount of GPU headroom can push beyond that.

Interestingly, the scaling from Medium to Ultra settings at each resolution shows a consistent pattern. At 4K, the system drops from 128 FPS at Medium to 112 FPS at Ultra, a 12.5% penalty. At 1440p, it goes from 135 FPS at Medium to 123 FPS at Ultra, an 8.9% drop. At 1080p, the figures are 140 FPS at Medium and 128 FPS at Ultra, a 8.6% reduction. This consistent behavior across resolutions indicates that the CPU overhead per frame is relatively fixed, and the GPU is handling the increased graphical load without becoming the bottleneck until Ultra settings. The data suggests that the RTX 5090's 32 GB of GDDR7 memory and 1.79 TB/s bandwidth are more than sufficient for this title, even at 4K Ultra.

GPU Role

The NVIDIA GeForce RTX 5090 brings formidable specifications to this pairing: 21,760 shading units, 170 RT cores, and 680 tensor cores running on the Blackwell 2.0 architecture at a 5 nm process node. Its boost clock of 2407 MHz, combined with a 512-bit memory bus delivering 1.79 TB/s of bandwidth, should theoretically crush any modern game. The GPU's FP32 performance of 104.8 TFLOPS and a pixel rate of 423.6 GPixel/s place it in the 92nd percentile of all GPUs, with an average benchmark score of 79,842.

However, the measured FPS in Overpower tells a different story. The RTX 5090's nearest rivals in synthetic benchmarks—the NVIDIA Tesla P100 PCIe 16 GB (deltaPct 0.3%) and Tesla P100 PCIe 12 GB (deltaPct 0.6%)—are data-center cards with far lower gaming performance, yet this pairing with the i5-10400F cannot differentiate itself in actual gameplay. The GPU's role here is underutilized; at 1080p Low, the system achieves 168 FPS, which is only 50% higher than the 112 FPS at 4K Ultra. A fully GPU-bound scenario would show a much wider spread, often 2-3x between those extremes.

The VRAM capacity of 32 GB is irrelevant in this context, as the data shows no memory-related stuttering or frame drops that would indicate capacity limits. The 4K Medium settings show a min FPS of 109 and max of 147, with an average of 128, indicating the GPU is still capable of maintaining playable framerates even when the CPU is struggling to feed it. The GPU's 92nd percentile ranking among all GPUs is not reflected in the measured results, which are far more modest due to the CPU bottleneck. This suggests that for Overpower, the RTX 5090's role is to provide headroom for future titles or higher settings, not to maximize frame rates in this specific game at this CPU pairing.

How This Combo Ranks

The combo rank of 635 out of 3713 tested combinations places this system in the 82.9th percentile of all configurations tested. This is a strong overall position, but it masks the significant imbalance in the pairing. The Intel Core i5-10400F sits in the 68th percentile of all CPUs with an average benchmark score of 14,185, placing it near rivals like the Intel Xeon 6756E (deltaPct 0.2%), Intel Core 7 160UL (deltaPct -0.3%), AMD EPYC 7552 (deltaPct 0.5%), and AMD Ryzen 3 7320C (deltaPct -0.6%). These rivals are a mix of enterprise and low-power parts, highlighting that the i5-10400F is a mid-range CPU from 2020, not a high-end part.

The GPU, by contrast, ranks in the 92nd percentile of all GPUs, with an average benchmark score of 79,842. Its nearest rivals include the aforementioned Tesla P100 variants and the AMD Radeon RX 6850M XT (deltaPct 1.1%) and AMD Radeon Pro Vega 64X (deltaPct -1.4%). The discrepancy between the CPU's 68th percentile and the GPU's 92nd percentile is the root cause of the bottleneck observed in the FPS data. This combo's rank of 635 suggests that many other configurations—likely those with more balanced CPU-GPU pairings—outperform it despite having weaker individual components.

The data implies that the combo rank is being dragged down by the CPU's limitations. If the RTX 5090 were paired with a more capable processor, the rank would likely be much higher, given the GPU's raw power. The fact that this combo still manages to land in the top 17% of all tested configurations confirms the RTX 5090's brute force, but it also highlights a significant missed opportunity for performance. The measured FPS at 1080p High (137 FPS) is not far off from what a much cheaper GPU might achieve with a better CPU, making this pairing inefficient for Overpower specifically.

FAQ

Q: Why does the FPS drop so little from 1080p to 4K in this game?

A: The data shows only an 8.8% drop in FPS at High settings (from 137 to 125 FPS) between 1080p and 4K. This flat scaling indicates the Intel Core i5-10400F's 6-core/12-thread configuration is the limiting factor, not the RTX 5090, as a GPU-bound scenario would show much larger performance penalties at higher resolutions.

Q: Is the RTX 5090 being fully utilized in this configuration?

A: No, benchmark results indicate the GPU is underutilized. At 1080p Low, the system hits 168 FPS, while at 4K Ultra it manages 112 FPS—a difference of only 50%. The RTX 5090's 92nd percentile GPU ranking is not reflected in these results, which are capped by CPU performance.

Q: How does this combo rank compared to other tested configurations?

A: This combo ranks 635th out of 3713 tested combinations, placing it in the top 17.1% of all systems. However, this rank is modest given the RTX 5090's capabilities, and the data suggests the CPU's 68th percentile ranking is holding the system back.

Q: What is the best resolution for this system in Overpower?

A: The data shows that 4K Ultra still delivers an average of 112 FPS, which is playable. Since the FPS drop from 1080p to 4K is minimal, the system is best used at 4K to take advantage of the RTX 5090's 32 GB GDDR7 memory and 512-bit bus, without sacrificing much performance.

Q: Does the VRAM capacity of 32 GB matter for this game?

A: The measured FPS data shows no signs of memory-related issues, such as stuttering or frame drops, even at 4K Ultra. The 32 GB capacity is more than sufficient for Overpower, and the game's performance is constrained by the CPU, not the GPU's memory subsystem.

Q: Why is the frame rate at 1080p Low (168 FPS) not much higher than 4K Low (153 FPS)?

A: This 15 FPS difference (8.9%) suggests the CPU reaches its maximum frame delivery capability around 168 FPS. With fewer pixels to render at 1080p, the GPU finishes its work faster but must wait for the CPU, which is why the frame rate ceiling is so consistent across resolutions.

Measured FPS Breakdown

At 1920x1080, the system delivers its highest frame rates. Low settings produce an average of 168 FPS, Medium averages 140 FPS with a range of 119 to 161 FPS, High averages 137 FPS, and Ultra averages 128 FPS. The spread from Low to Ultra is 40 FPS, or 23.8%, which is moderate but shows diminishing returns as settings increase—the jump from Medium to High is only 3 FPS, while High to Ultra is 9 FPS.

Moving to 2560x1440, the averages are 162 FPS at Low, 135 FPS at Medium (with min/max of 115/155), 131 FPS at High, and 123 FPS at Ultra. The step from Low to Medium is a significant 27 FPS drop, but from Medium to High is just 4 FPS, and High to Ultra is 8 FPS. The 1440p results are nearly identical to 1080p, with only a 6 FPS difference at Low (168 vs 162) and 5 FPS at High (137 vs 131), reinforcing the CPU bottleneck.

At 3840x2160, the system still maintains strong performance: 153 FPS at Low, 128 FPS at Medium (with min/max of 109/147), 125 FPS at High, and 112 FPS at Ultra. The 4K Low result is only 15 FPS lower than 1080p Low, and 4K Ultra is only 16 FPS lower than 1080p Ultra. The Medium settings show the most variance at 4K, with a 38 FPS range between min and max, indicating some occasional CPU-induced stuttering. Across all resolutions, the pattern is clear: the i5-10400F caps performance around 150-168 FPS at Low, and the RTX 5090 only becomes the limiting factor at Ultra settings, where the frame rates drop to 112-128 FPS depending on resolution.

CPU Role

The Intel Core i5-10400F, with its 6 cores and 12 threads, is the clear bottleneck in this pairing. Its base clock of 2.90 GHz and boost clock of 4.30 GHz, combined with 12 MB of shared L3 cache, provide adequate single-thread performance for many titles, but Overpower appears to scale with CPU capability beyond this point. The CPU's 3DMark scores reveal its threading characteristics: 1,350 for 2 threads, 2,560 for 4 threads, 3,929 for 8 threads, and 4,748 for 16 threads. The scaling from 2 to 4 threads is 89.6%, but from 4 to 8 threads it's only 53.5%, and from 8 to 16 threads just 20.9%, indicating diminishing returns beyond 4 cores.

In Cinebench R23, the multicore score of 10,283 and single-core score of 1,451 show a balanced part, but the single-core performance is what matters most for gaming. The Geekbench single-core score of 1,420 places this CPU in the 68th percentile of all CPUs, which is respectable but not exceptional for a 2020 release. The PassMark single-thread score of 2,541 reinforces this mid-range positioning. The CPU's nearest rivals in average benchmark score are the Intel Xeon 6756E (14,163, deltaPct 0.2%) and AMD Ryzen 3 7320C (14,277, deltaPct -0.6%), both of which are not gaming-focused parts.

The data shows that the CPU's 12 threads are not being fully utilized in Overpower, as the frame rate is consistent across resolutions, suggesting a single-thread or low-thread-count limitation. The 3DMark 2-thread score of 1,350 is particularly telling—it represents only 28.4% of the 16-thread score, indicating that the CPU's dual-core performance is relatively weak compared to its multi-core capabilities. This is likely why the system cannot exceed ~168 FPS in any scenario, regardless of the GPU's massive headroom.

Settings Recommendations

Given the measured data, the optimal settings for this configuration depend on the user's resolution and performance priorities. At 1920x1080, the data shows that High settings (137 FPS) offer only 9 FPS less than Low (168 FPS) but likely provide significantly better visual quality. The jump to Ultra (128 FPS) costs another 9 FPS, which may not be worth it for the visual gain, especially since the 1080p Medium settings (140 FPS) are nearly identical to High.

At 2560x1440, the same pattern holds: High (131 FPS) is a good balance, only 7 FPS below Medium (135 FPS) and 8 FPS above Ultra (123 FPS). The 1440p Medium setting shows a min FPS of 115, which is still playable, but High does not have a recorded min FPS, suggesting it may be more stable. For most users, 1440p High would be the recommended sweet spot, delivering smooth gameplay with no noticeable compromise.

At 3840x2160, the data shows that all settings are playable, from Ultra (112 FPS) to Low (153 FPS). The 4K Medium settings have a min FPS of 109, which is above the commonly accepted 60 FPS threshold for smooth play, and the average of 128 FPS is excellent. Since the CPU bottleneck is so pronounced, there is little performance penalty for choosing Ultra (112 FPS) over High (125 FPS) at 4K—only 13 FPS. Given the RTX 5090's 32 GB VRAM and high bandwidth, 4K Ultra is the most sensible choice for this configuration, as it maximizes visual fidelity while still maintaining a high average frame rate. The overall recommendation from the data is to prioritize resolution over settings, as the CPU will cap performance regardless, and the GPU has ample headroom to handle Ultra settings at any resolution without significant frame rate losses.