Thief

The AMD Ryzen 9 5950X and NVIDIA GeForce RTX 5070 combination ranks 352nd out of 1,642 tested combos in Thief, placing it in the top 21.5% of all systems evaluated for this title. This is a strong showing for a pairing that pairs a 2020-era 16-core processor with a 2025 Blackwell GPU, and the percentile placement suggests the combo is well-matched for the game’s demands. The CPU’s 93rd percentile ranking among all processors, combined with the GPU’s 83rd percentile among all graphics cards, indicates both components are above-average performers in their respective fields. The average benchmark scores tell a similar story: the Ryzen 9 5950X posts an average score of 49,062, which is within 0.2% of the Intel Core i5-14600K’s 49,166, and within 0.6% of the Intel Core i5-14600KF’s 49,367. The NVIDIA GPU’s average score of 41,687 is effectively tied with the AMD Radeon Pro 5300’s 41,610, differing by just 0.2%.

How This Combo Ranks

The 352nd position out of 1,642 combos is a meaningful result because it places this system comfortably in the upper quartile of all tested configurations for Thief. The deltaPct values from the nearest rivals reinforce that neither component is a bottleneck outlier. The Ryzen 9 5950X’s average benchmark score of 49,062 trails the Intel Core i5-14600K by only 0.2%, while it leads the Intel Xeon Gold 5318H by 0.7%. This near-parity with modern mid-range Intel parts suggests the 16-core AMD processor is not holding the GPU back in any meaningful way. On the graphics side, the RTX 5070’s average score of 41,687 sits between the NVIDIA Tesla M40’s 41,897 (0.5% higher) and the GeForce RTX 3090’s 41,441 (0.6% lower). The RTX 5070’s 83rd percentile among all GPUs means it outperforms roughly 83% of graphics cards in the database, which is a robust position for a stealth title like Thief that leans more on CPU overhead than raw rasterization.

The data shows that this combo’s rank is not a fluke of one resolution or preset. Across the 12 measured data points, the system never dips below 59.1 FPS at any resolution or settings level, which confirms the balanced nature of the pairing. The CPU’s 16 cores and 32 threads, clocked at a base of 3.40 GHz and boost of 4.90 GHz, provide ample headroom for Thief’s engine, while the GPU’s 6,144 shading units and 80 ROPs handle the rendering load without creating a persistent frame-time bottleneck. The result is a combo that ranks higher than roughly 78.6% of the 1,642 tested systems, placing it in the same performance tier as configurations with more recent CPUs but comparable GPU horsepower.

GPU Role

The NVIDIA GeForce RTX 5070 brings 12 GB of GDDR7 memory on a 192-bit bus, delivering 672.0 GB/s of bandwidth. This memory configuration is critical for Thief, as the game’s high-resolution textures and shadow maps can consume significant VRAM at Ultra settings. The GPU’s boost clock of 2512 MHz and base clock of 2325 MHz are moderate for the 50-series, but the Blackwell 2.0 architecture compensates with 48 RT cores and 192 tensor cores, which are not heavily utilized in this 2014 title. The FP32 performance of 30.87 TFLOPS is more than sufficient for Thief’s DirectX 11-era rendering paths, and the 201.0 GPixel/s pixel rate ensures that even at 4K, the fill-rate demands are met without strain.

In the measured data, the GPU shows a clear scaling pattern that indicates it is the primary performance driver at higher resolutions. At 4K Ultra, the combo produces 59.1 FPS, while at 4K Low it reaches 112.2 FPS, a difference of 53.1 FPS that is almost entirely attributable to GPU load. The 12 GB VRAM capacity is not exceeded at any tested setting, as evidenced by the absence of min-FPS anomalies or stutter-inducing memory swaps. The GPU’s 83rd percentile ranking, while lower than the CPU’s 93rd, is still strong enough to avoid being the weak link in this pairing. The passmark G3D score of 29,137 and the 3DMark Steel Nomad DX12 score of 5,077 both indicate a card that performs well above the median, and the RTX 5070’s nearest rival, the GeForce RTX 3090, is only 0.6% behind in average score, suggesting that this 50-series card punches at or above the level of a previous-generation flagship.

Settings Recommendations

The measured FPS data provides a clear picture of which preset offers the best balance of visual quality and performance for this combo. At 1080p, the Ultra preset delivers 134.4 FPS, which is 7.3 FPS lower than High’s 141.7 FPS but still well above the 144 Hz refresh rate that many gamers target. The Medium preset at 1080p produces 152 FPS, and Low reaches 169.8 FPS, but the visual degradation from Ultra to Medium is significant in Thief’s shadow-heavy environments, making the 17.6 FPS gain from Ultra to Medium less compelling than it might be in other titles. For 1440p, the Ultra preset drops to 106.3 FPS, while High produces 106.6 FPS, a negligible difference that suggests the game’s Ultra settings do not add much visual fidelity over High at this resolution. The Medium preset at 1440p yields 139.2 FPS, which is 32.9 FPS higher than Ultra, making it the sweet spot for high-refresh 1440p monitors.

At 4K, the situation changes. Ultra produces 59.1 FPS, which is barely playable for a stealth game that benefits from smooth camera movement. High at 4K reaches 67.1 FPS, a modest improvement, but Medium jumps to 82.5 FPS, providing a much more comfortable margin above the 60 FPS threshold. Given that the difference between High and Medium at 4K is 15.4 FPS, and the visual impact of dropping from High to Medium in Thief is primarily in shadow resolution and texture filtering, the Medium preset at 4K represents the best experience for this combo. It delivers 82.5 FPS, which is 40.1% higher than the Ultra preset’s 59.1 FPS, while still maintaining acceptable image quality for a 2014 title. The data suggests that players with 1440p monitors should use Medium for maximum smoothness, while 4K users should avoid Ultra entirely and settle on Medium to stay well clear of the 60 FPS floor.

Measured FPS Breakdown

At 1920x1080, the combo exhibits a tight spread across presets. Low produces 169.8 FPS, Medium 152 FPS, High 141.7 FPS, and Ultra 134.4 FPS. The gap between Low and Ultra is 35.4 FPS, which is the smallest absolute range of any resolution, indicating that at 1080p the CPU is doing more of the heavy lifting relative to the GPU. The Ryzen 9 5950X’s 16 cores and 32 threads ensure that the game’s CPU-bound sections, such as the crowded city streets of Thief’s open hub areas, do not cause frame drops even at Low settings. The 1080p Low result of 169.8 FPS is the highest frame rate recorded across all 12 measurements, and it shows that the combo is never CPU-limited in this resolution, as even the most demanding preset stays above 134 FPS.

Moving to 2560x1440, the spread widens considerably. Low reaches 150.9 FPS, Medium 139.2 FPS, High 106.6 FPS, and Ultra 106.3 FPS. The interesting data point here is that High and Ultra are nearly identical, with a difference of just 0.3 FPS, which suggests that the Ultra preset’s additional effects are not GPU-bound at this resolution. The Medium preset, at 139.2 FPS, is 32.9 FPS higher than High, making it the clear choice for 1440p users who want to push refresh rates. The 4K results show the most dramatic scaling: Low produces 112.2 FPS, Medium 82.5 FPS, High 67.1 FPS, and Ultra 59.1 FPS. The drop from Low to Ultra at 4K is 53.1 FPS, nearly double the 1080p range, which confirms that the GPU is the limiting factor at this resolution. The 4K Ultra result of 59.1 FPS is the only measurement below 60 FPS in the entire dataset, and it underscores the recommendation to avoid Ultra at 4K.

Resolution Scaling

The FPS progression from 1080p to 4K reveals a clear GPU-bound scaling pattern. At Low settings, the combo delivers 169.8 FPS at 1080p, 150.9 FPS at 1440p, and 112.2 FPS at 4K. This represents a drop of 18.9 FPS (11.1%) from 1080p to 1440p, and a further drop of 38.7 FPS (25.7%) from 1440p to 4K. The percentage decline from 1080p to 4K at Low is 33.9%, which is less than the theoretical pixel count increase of 4x, indicating that the CPU is partially masking the GPU load at lower resolutions. At Ultra settings, the scaling is more pronounced: 134.4 FPS at 1080p, 106.3 FPS at 1440p, and 59.1 FPS at 4K. The 1080p-to-1440p drop is 28.1 FPS (20.9%), and the 1440p-to-4K drop is 47.2 FPS (44.4%). The total 1080p-to-4K decline at Ultra is 56.0%, which is much closer to the pixel-count increase, confirming that the GPU becomes the dominant bottleneck at higher resolutions and settings.

This scaling behavior indicates that the Ryzen 9 5950X is more than capable of feeding the RTX 5070 at 1080p, where the 16-core processor’s single-thread performance (3DMark single-thread score of 944) and multi-thread capabilities (Cinebench R23 multicore score of 38,607) prevent any CPU-side stalls. However, at 4K Ultra, the GPU’s 12 GB of VRAM and 672.0 GB/s bandwidth are taxed to their limits, resulting in the 59.1 FPS result. The fact that the 1080p-to-4K drop at Low is only 33.9% versus 56.0% at Ultra suggests that Thief’s higher presets introduce GPU-intensive effects that scale disproportionately with resolution. For this combo, the limiting component is definitively the GPU at 4K, while at 1080p the CPU has enough headroom that even the highest preset does not cause the frame rate to dip below 134 FPS. The data supports a conclusion that this pairing is best suited for 1440p Medium or 4K Medium, where the frame rates remain high enough to avoid any perception of sluggishness in Thief’s stealth gameplay.