Thief

# Thief — Intel Core i5-10400F + NVIDIA GeForce RTX 3050 4 GB

The benchmark data for this combination in Thief places the system at rank 1509 out of 1642 tested configurations, indicating a bottom-decile standing for this title. The measured frame rates across resolutions and presets reveal a system that is consistently GPU-limited, with the 4 GB VRAM capacity of the RTX 3050 acting as a significant constraint at higher settings. The CPU, a 6-core/12-thread Comet Lake part with a boost clock of 4.30 GHz, posts a 73rd percentile ranking among all CPUs, suggesting it has ample headroom for this 2014 stealth title. The data shows a clear pattern: resolution scaling produces steep performance drops, and the gap between Low and Ultra presets narrows as resolution increases, which is characteristic of a graphics card operating near its memory and bandwidth limits.

Resolution Scaling

The measured FPS data shows a steep and consistent performance decline as resolution increases from 1080p to 4K, demonstrating that the RTX 3050 4 GB is the dominant limiting component across all settings. At Low settings, the average frame rate drops from 110.6 FPS at 1920x1080 to 72.9 FPS at 2560x1440, a reduction of roughly 34%, and then falls further to 43.4 FPS at 3840x2160, which is approximately 60% below the 1080p figure. This scaling pattern is much steeper than what would be expected from a purely pixel-bound workload, indicating that memory bandwidth and capacity are playing a critical role.

At High settings, the same trend is visible but with a more compressed range: 62.7 FPS at 1080p, 41.7 FPS at 1440p, and 24.4 FPS at 4K. The drop from 1080p to 4K at High settings is about 61%, nearly identical to the Low preset drop, which suggests that the GPU is saturated regardless of the graphical load. This is a telling sign that the 4 GB frame buffer is being exhausted at higher resolutions, forcing the card into memory-bound behavior that limits its ability to scale with resolution.

The Medium and Ultra presets reinforce this interpretation. Medium produces 85.0 FPS at 1080p, dropping to 58.2 FPS at 1440p and 33.4 FPS at 4K. Ultra, the heaviest preset, yields 60.8 FPS at 1080p, 43.7 FPS at 1440p, and 25.0 FPS at 4K. Notably, the gap between Ultra and High at 1080p is only 1.9 FPS (60.8 vs. 62.7), which is surprisingly small for a full preset step. This near-identical performance suggests that at 1080p, the system is already hitting a ceiling that is not purely based on shading work, but rather on memory throughput. At 4K, Ultra (25.0 FPS) and High (24.4 FPS) are virtually indistinguishable, further confirming that the GPU's memory subsystem is the bottleneck.

The data indicates that the i5-10400F is not limiting performance at any resolution. The CPU's multi-threaded benchmark scores, such as 10297 in Cinebench R23 multicore and 12115 in Passmark multithread, are far beyond what a game from 2014 would require. The 3DMark 16-thread score of 4748 and 8-thread score of 3929 show strong scaling, meaning the CPU can feed the GPU adequately. Therefore, the resolution scaling curve observed here is almost entirely a function of the RTX 3050's capabilities, particularly its 128-bit memory bus and 192.0 GB/s bandwidth.

Settings Recommendations

Given the measured data, the optimal settings for this configuration depend heavily on the target resolution and desired frame rate. At 1920x1080, the High preset delivers 62.7 FPS, which is above the 60 FPS threshold for smooth gameplay, while the Ultra preset produces 60.8 FPS. The difference between these two is negligible at 1.9 FPS, so choosing High over Ultra is a rational decision to reduce GPU load without a perceptible change in smoothness. The Medium preset at 85.0 FPS offers a substantial 35% improvement over High, making it an attractive option for players who prioritize frame rate over visual fidelity.

At 2560x1440, the situation changes. High drops to 41.7 FPS, which is below 60 FPS and will feel noticeably less fluid. Medium offers 58.2 FPS, which is close to the 60 FPS target but not quite there. Low at 72.9 FPS is the only preset that provides a comfortable margin above 60 FPS at 1440p. For a stealth game like Thief, where precise timing and visual awareness matter, the Low preset may be the most pragmatic choice to maintain responsiveness, despite the visual compromise.

At 3840x2160, none of the presets achieve 60 FPS. Low produces 43.4 FPS, Medium 33.4 FPS, High 24.4 FPS, and Ultra 25.0 FPS. The data shows that even Low is far from playable at a high refresh rate, and the card's 4 GB VRAM is clearly insufficient for 4K rendering in this title. The recommendation for 4K would be to avoid this resolution entirely, as the measured frame rates are below 30 FPS for all presets except Low, and even Low (43.4 FPS) is marginal.

The Ultra preset offers the worst value in terms of performance per visual gain. At 1080p, Ultra is only 1.9 FPS slower than High, but at 1440p, Ultra (43.7 FPS) is actually 2.0 FPS faster than High (41.7 FPS), which is an anomaly in the data. This inconsistency suggests that Ultra may not be properly optimized or that the GPU is hitting a memory wall that makes preset differences irrelevant. The measured data suggests that High is the best "sweet spot" for 1080p, Medium for 1440p, and Low for 4K (if one must play at that resolution), but none of these are ideal.

FAQ

Q: Is the CPU or GPU the bottleneck in Thief with this configuration?

A: The data strongly indicates the GPU is the limiting component. The i5-10400F has a 73rd percentile ranking among all CPUs and scores 10297 in Cinebench R23 multicore, which is far more than Thief (a 2014 title) can utilize. The FPS scaling from 1080p to 4K shows a steep drop that correlates with GPU load, not CPU capability.

Q: What is the highest playable setting at 1080p?

A: At 1920x1080, the Ultra preset delivers 60.8 FPS, which is technically playable, but the High preset at 62.7 FPS is nearly identical. Since the difference is only 1.9 FPS, High is recommended as it provides the same experience with less GPU strain. Medium at 85.0 FPS offers a significant boost if frame rate is the priority.

Q: Can this system handle 4K gaming in Thief?

A: No. At 3840x2160, the best measured result is 43.4 FPS on Low settings. Medium drops to 33.4 FPS, and High/Ultra fall to 24.4 and 25.0 FPS, respectively. None of these achieve the 60 FPS target, and even 30 FPS is only met by Low. The 4 GB VRAM is insufficient for 4K in this game.

Q: Why is Ultra slower than High at 1080p but faster at 1440p?

A: At 1080p, Ultra (60.8 FPS) is 1.9 FPS slower than High (62.7 FPS). At 1440p, Ultra (43.7 FPS) is 2.0 FPS faster than High (41.7 FPS). This inversion suggests the GPU is hitting a memory bandwidth limit where preset complexity becomes less relevant than the frame buffer usage, causing inconsistent scaling.

Q: How does the 4 GB VRAM affect performance?

A: The 4 GB GDDR6 memory with a 128-bit bus and 192.0 GB/s bandwidth is the primary constraint. The data shows that increasing resolution causes disproportionate FPS drops, and the gap between Low and Ultra narrows at higher resolutions, both of which are classic signs of VRAM capacity or bandwidth exhaustion rather than compute limits.

Q: Is the RTX 3050 4 GB a good match for the i5-10400F?

A: For this game, the CPU has significant headroom, as evidenced by its high benchmark scores relative to Thief's requirements. The GPU is the limiting factor at all resolutions. In terms of balance, the CPU is overqualified for the GPU in this title, but the pairing is reasonable for other workloads, as shown by the CPU's 73rd percentile ranking.

Measured FPS Breakdown

At 1920x1080, the measured average frame rates are as follows: Low produces 110.6 FPS, Medium produces 85.0 FPS, High produces 62.7 FPS, and Ultra produces 60.8 FPS. The spread from Low to Ultra is 49.8 FPS, which represents a 45% reduction in performance when moving to the highest preset. The step from Medium to High costs 22.3 FPS (85.0 to 62.7), which is the largest single preset penalty, while the step from High to Ultra costs only 1.9 FPS, indicating diminishing returns at the top end.

At 2560x1440, the measured averages are 72.9 FPS for Low, 58.2 FPS for Medium, 41.7 FPS for High, and 43.7 FPS for Ultra. The Low to Ultra spread narrows to 29.2 FPS, a 40% reduction. Notably, the High to Ultra step shows an increase of 2.0 FPS, which is anomalous and suggests that the Ultra preset may be triggering a different rendering path or that the measurement variance is high. The Medium preset at 58.2 FPS is the closest to the 60 FPS target.

At 3840x2160, the measured averages are 43.4 FPS for Low, 33.4 FPS for Medium, 24.4 FPS for High, and 25.0 FPS for Ultra. The Low to Ultra spread is 18.4 FPS, a 42% reduction, which is proportionally similar to the other resolutions. The High and Ultra presets are effectively tied at approximately 24-25 FPS, confirming that the GPU's memory bandwidth is the bottleneck. The Medium preset at 33.4 FPS is the only setting that approaches playability, but it is still well below 60 FPS.

The data reveals a consistent pattern: the absolute FPS values drop by roughly 40-60% from 1080p to 4K across all presets, but the relative difference between presets shrinks as resolution increases. At 1080p, Low is 45% faster than Ultra; at 4K, Low is only 42% faster than Ultra. This is a subtle but important indicator that the GPU is becoming more memory-bound and less compute-bound at higher resolutions.

GPU Role

The NVIDIA GeForce RTX 3050 4 GB is the central constraint in this configuration. Its specifications—2048 shading units, 64 TMUs, 32 ROPs, and 16 RT cores—are modest for a 30-series card, but the most critical factor is the memory subsystem: 4 GB of GDDR6 on a 128-bit bus with 192.0 GB/s bandwidth. The card's base clock of 1545 MHz and boost clock of 1740 MHz, combined with a 90 W TDP, place it in the entry-level segment. The pixel rate of 55.68 GPixel/s and texture rate of 111.4 GTexel/s are sufficient for 1080p gaming, but the memory bandwidth is insufficient for higher resolutions.

The card's FP32 performance of 7.127 TFLOPS is not the limiting factor in Thief, as the game's graphics workload is relatively light by modern standards. Instead, the data points to VRAM capacity as the primary issue. Thief at 4K with High or Ultra settings likely requires more than 4 GB of frame buffer, causing the card to spill over into system memory or reduce texture streaming, which manifests as the severe FPS drops observed (24.4 and 25.0 FPS at 4K High/Ultra). The fact that Low at 4K (43.4 FPS) is nearly double that of High/Ultra suggests that the game is able to stay within the 4 GB limit at Low, but exceeds it at higher presets.

The GPU's 50th percentile ranking among all GPUs (with an average benchmark score of 0, indicating no synthetic tests were recorded) further contextualizes its position. The card is a mid-pack performer, but its 4 GB capacity is a known limitation that is exacerbated by modern game requirements. In Thief, this limitation is exposed at 1440p and 4K, where even Medium settings fail to maintain 60 FPS. The card's memory clock of 1500 MHz (12 Gbps effective) and 192.0 GB/s bandwidth are not sufficient to compensate for the capacity shortfall, leading to the steep resolution scaling observed in the data. The 8 nm process node and 8,700 million transistors on a 200 mm² die are irrelevant to performance here; the memory configuration is the decisive factor.