Thief

The pairing of Intel's Core Ultra 7 265K with NVIDIA's GeForce RTX 2060 yields a functional, though not exceptional, experience in the 2014 stealth title Thief. Measured data shows average frame rates spanning from 142 FPS at 1080p Low down to 31 FPS at 4K Ultra. Across the four tested settings (Low, Medium, High, Ultra) and three resolutions (1920×1080, 2560×1440, 3840×2160), the combo exhibits a typical scaling pattern dominated by the GPU, achieving a modest frame count that ranks 2927th out of 3720 tested configurations—or the 21.3rd percentile. The CPU and GPU individually sit in the 94th and 58th percentiles respectively, suggesting the processor is far more capable than the graphics card at this workload. The following sections dissect the measured rows to determine ideal settings, component roles, and scalability behavior.

Settings Recommendations

Among the preset options with reported measurements, the Medium setting offers the most balanced performance envelope across all three resolutions. At 1080p Medium, the system produces 108 FPS with a measured minimum of 91 FPS and a maximum of 124 FPS; at 1440p it yields 74 FPS (minimum 63, maximum 85); and at 4K it delivers 43 FPS (minimum 37, maximum 50). These values keep every resolution comfortably above the 30 FPS baseline, and they are only slightly lower than the corresponding High presets (82, 56, 33 FPS). More importantly, the Medium preset is the only setting for which minimum and maximum values are measured, making it the most predictable option in terms of frame pacing. In contrast, Low presets deliver significantly higher raw performance—142 FPS at 1080p, 97 at 1440p, 57 at 4K—but the lack of reported minimum and maximum values means we can infer that the consistency is likely lower, though we cannot confirm from the dataset. Ultra presets, meanwhile, reduce the frame rate to 78 FPS at 1080p, 54 at 1440p, and 31 at 4K, which is only 6–9 FPS lower than High except at 4K where it dips 2 FPS below the already narrow 33 FPS of High. Because the Medium setting maintains at least a 37 FPS minimum even at 4K, it maximizes the likelihood of a fluid experience without sacrificing high resolutions. Consequently, the measured data strongly indicates that Medium is the best preset for this combination, offering an average FPS that is at least 25% higher than Ultra at every resolution while providing a guaranteed floor that does not interrupt play.

FAQ

Q: What is the average FPS at 1920×1080 with High settings?

A: The average is 82 FPS.

Q: How does the combo rank among all tested configurations?

A: The Intel Core Ultra 7 265K + RTX 2060 combo ranks 2927th out of 3720 total combos, which corresponds to the 21.3rd percentile—meaning it beats only 793 other configurations in Thief.

Q: What is the minimum FPS measured at any resolution?

A: The lowest recorded minimum is 37 FPS, occurring at 3840×2160 on Medium settings; the maximum at that preset is 50 FPS, giving a narrow range.

Q: How does the GPU's benchmark score compare to its nearest rivals?

A: The RTX 2060 has a Passmark G3D score of 14,111, putting it 0% faster than the GeForce GTX 580 (score 15,283) and 0.1% faster than the Radeon 680M (15,270). It is also 0.6% faster than the RTX 3050 OEM (15,199) and 0.8% faster than the Radeon RX 7600 (15,171).

Q: What is the CPU's multi-core performance in Cinebench R23?

A: The Core Ultra 7 265K scores 35,850 points in Cinebench R23 multi-core and 2,020 in the single-core test.

Q: How much L3 cache does the CPU have?

A: The CPU has 30 MB of shared L3 cache, with 192 KB L1 per core and 3 MB L2 per core.

GPU Role

The NVIDIA GeForce RTX 2060 is clearly the limiting component in this combination. Its 6 GB GDDR6 memory on a 192-bit bus provides 336.0 GB/s of bandwidth, and at the 4K resolution the frame rates are already the lowest—averaging 31 FPS on Ultra and 33 on High. The scaling of average FPS with resolution is a direct indicator of GPU-bound behavior: from 1080p to 4K, the average drops to about 40% of its 1080p value for every preset. For example, from 1080p Low (142 FPS) to 1440p Low (97) to 4K Low (57), which is exactly a 1.86× and 2.49× pixel count increase relative to 1080p? The consistent ratio of about 1.4× for each step down to 4K is typical of a compute-limited GPU where the rendering workload scales with pixel count. The RTX 2060's 58th percentile among all GPUs and its Passmark G3D score of 14,111 (which ties its nearest rival at 0% difference) indicate it is a mid-range card in absolute terms. This aligns with the Thief results: while the CPU can easily push 142 FPS at 1080p (a high-rate environment for a 2014 title), the GPU cannot deliver a 60 FPS experience at 1440p High (54 FPS) or 4K above 33 (High). The 6 GB buffer may also affect the level of detail, but the lack of min/max data on most settings means we cannot separate frame-time spikes from bandwidth limitations. What is clear is that any attempt to exceed Medium at higher resolutions hits the GPU's capacity, since the average FPS collapses once the GPU is maxed out.

How This Combo Ranks

The combination’s 2927th place (out of 3720) places it in the lowest ~21% of all tested systems for Thief. This raw ranking is surprising when compared to the underlying component performance: the CPU sits at the 94th percentile overall (average benchmark 70,879, with nearest rivals including the Intel Xeon 6511P at -0.2% and the Intel Core i7-14700KF at +1%), while the GPU ranks in the 58th percentile. The discrepancy arises because the CPU is far more powerful than what the GPU can utilise at the resolutions and settings tested; thus, the ranking is effectively a reflection of the GPU’s capability. The rank is also influenced by the game’s dated engine, which may not be able to leverage the CPU fully. Nonetheless, the relative standing among tested combos is consistent with the GPU’s own percentile placing: the GPU at 68% percentile, but the combo ranking is lower because only 21.3% of systems are slower, which suggests that most test systems pair the RTX 2060 with a weaker CPU. In contrast, a 94th percentile CPU is an outlier in such a low-tier GPU pairing, so the final rank is dragged down by the GPU exactly as the average FPS across resolutions indicates. The percentile data reinforces that those wishing to optimise a similar setup would achieve more gain by improving the GPU rather than the CPU—the GPU’s percent rank (58) is exactly the bottleneck.

Resolution Scaling

Rendering resolution has a dramatic and consistent impact on frame rate. Moving from 1080p to 1440p at the Medium preset decreases average FPS from 108 to 74, that is, to 68.5% of the original; again, stepping to 4K reduces it further to 43, which is 39.8% of the 1080p value. For the Low preset, the same steps yield 142→97→57 FPS, to 40.0% of the 1080p figure. At High, 8382→56→33, also to 40.2% of the original. All four presets show nearly identical scaling—each successive resolution merge drops to about 60% and then 60% of that, actually lo — so the pixel-level workload is the limiting factor. This uniform FPS reduction across all settings is a classic sign of a GPU-bound pipeline: the CPU has more than enough throughput to generate frames up to 142 FPS (1080p Low) without limiting the GPU, so the GPU’s output limit is what determines game frame rates. The scaling ratios are also close to the inverse of pixel count (from 1920×1080 to 3840×2160, four times the pixels, but only 40% of the original FPS), indicating that memory bandwidth rather than pixel throughput may also contribute; however, the measured data cannot distinguish between bandwidth and shading effects. Notably, the worst-case (4K Ultra) averages 31 FPS, which is just above the generally considered playable threshold, while the 57 FPS at 4K Low is the most playable 4K setting. Therefore, the resolution scaling closely follows the GPU’s rendering limit, reinforcing that the RTX 2060 is the only component that governs frame rate outside of the 1080p high-FPS zone.

Measured FPS Breakdown

The following table encapsulates all measured average, minimum, and maximum FPS values for the Intel Core Ultra 7 265K with the RTX 2060 across the four resolution and preset combinations:

| Resolution | Settings | Avg FPS | Min FPS | Max FPS |

|------------|----------|--------|---------|---------|

| 3840×2160 | High | 33 | - | - |

| 3840×2160 | Low | 57 | - | - |

| 3840×2160 | Medium | 43 | 37 | 50 |

| 3840×2160 | Ultra | 31 | - | - |

| 2560×1440 | High | 56 | - | - |

| 2560×1440 | Low | 97 | - | - |

| 2560×1440 | Medium | 74 | 63 | 85 |

| 2560×1440 | Ultra | 54 | - | - |

| 1920×1080 | High | 82 | - | - |

| 1920×1080 | Low | 142 | - | - |

| 1920×1080 | Medium | 108 | 91 | 124 |

| 1920×1080 | Ultra | 78 | - | - |

The table shows the only data provided: the Medium preset alone records minima and maxima, while all other presets have only an average. The complete dataset includes no frametimes beyond the 1440p middle point that could reveal stutter spikes, so all non-Medium presets are assessed on the basis of their average.

CPU Role

The Intel Core Ultra 7 265K is a 20-core, 20-thread chip based on the Arrow Lake architecture, fabricated on a 3 nm process. It operates with a base clock of 3.90 GHz and a boost of 5.50 GHz, and its 125 W TDP sits within a conventional power range for desktop flagships. Its memory support of DDR5 alongside 30 MB of shared L3 cache and dual-channel 102.4 GB/s bandwidth underscores that this is a top-tier part. Its benchmark performance places it in the 94th percentile of all CPUs, with a Cinebench R23 multi-core score of 35,850 and single-core of 2,020; in Cinebench R20, multi-core yields 20,918 and single 2953. These numbers vastly exceed the demands of a 2014 release like Thief, whose typical worker distribution (if any) would rarely saturate two or three cores. The measured in-game frame rates that are uncapped at 1080p (142 FPS Low) indicate that the CPU does not constrain the system at even the highest frame rates produced here; otherwise, theFPS would stay roughly constant even when the resolution drops. Instead, the progressive 1080p→1440p→4K performance drops follow the pixel count, which is the GPU’s domain. The fact that the combo’s * ranking is much lower than the CPU’s percentile proves that performance is not a function of CPU power. Rather, the Core Ultra 7 265K is paired with a GPU whose average benchmark score (15,290) is only 58th percentile, so the CPU spends most of its time waiting for the graphics pipeline. Theories that ultra-responsiveness at higher resolutions would require a powerful CPU are moot in this instance because the CPU is abundant. Including the full benchmark results of the CPU, the projections (like Cinebench R25 708) show single-thread power far above the typical gaming requirement; thus, the CPU is a non-limiting factor for Thief at all measured settings.

In summary, the tech sheet tells a clear story: the Core Ultra 7 265K is an excellent processor, but paired with the RTX 2060 in a game like Thief, it is the GPU that dictates exactly how fast the game runs. The recommended Medium preset offers the balanced performance, and the measurable performance drops during resolution scaling all point to a single GPU-bound constraint. For any further improvements, the data suggests moving to a stronger graphics card is the most efficient path, since the CPU parameters are already far beyond any conceivable bottleneck at 1080p/1440p.