Thief

The AMD Ryzen 7 5800X and NVIDIA GeForce RTX 2060 SUPER combination yields measurable frame rates in the stealth title Thief (released on 2014-02-24). Across four quality presets (Low, Medium, High, Ultra) and three resolutions (1080p, 1440p, 4K), the data shows a clear pattern: the GPU, not the CPU, defines the playable ceiling at higher resolutions. At 1080p the combo averages between 86 FPS (Ultra) and 155 FPS (Low), while at 4K the same presets produce averages from 35 to 63 FPS. The RTX 2060 SUPER is the limiting component at every configuration, with the Ryzen 7 5800X providing ample headroom.

Resolution Scaling

The measured frame rates at each resolution demonstrate an abrupt performance cliff when moving from 1920×1080 to 3840×2160. For High settings, the average FPS is 90 at 1080p, 62 at 1440p, and 36 at 4K. Low settings show the same pattern: 155, 107, and 63 FPS respectively. Medium settings follow with 118, 81, and 48 FPS; Ultra settings yield 86, 59, and 35 FPS across the three resolutions.

The relative drops between adjacent resolutions are nearly identical for every quality preset. From 1080p to 1440p, each preset loses roughly the same absolute amount, and the drop from 1440p to 4K is more severe in absolute terms but proportionally similar. For the Medium preset, the 1440p figure of 81 FPS sits below the 1080p average by 37 FPS, and 4K proceeds to 48 FPS, a further 33 FPS drop. High averages lose 28 FPS going from 1080p to 1440p, then 26 FPS to 4K. Ultra averages drop from 86 to 59 to 35, and Low from 155 to 107 to 63.

The uniformity of these declines points to a GPU‑bound scenario. The CPU in this pair, an 8‑core 16‑thread Zen 3 part with a boost clock of 4.70 GHz, does not change with resolution, so the payload shifts entirely onto the graphics card. The fact that the FPS loss between each step is almost constant—not suddenly accelerating—indicates that the RTX 2060 SUPER is being saturated in a regular and predictable way as the pixel count increases. The 4K numbers are particularly telling: every preset except Low average 48 FPS or lower, meaning that at 4K the card can barely maintain a playable rate even on Low.

GPU Role

The NVIDIA GeForce RTX 2060 SUPER is based on the TU106 chip, built on TSMC’s 12 nm process. It features 2,176 shading units, 136 texture mapping units, and 64 RO buddy. There are also 34 ray-tracing cores and 272 tensor cores, though a 2014 title like Thief does not engage those units. The card runs a base clock of 1470 MHz and boosts up to 1650 MHz, with 8 GB of GDDR6 memory on a 256‑bit bus, delivering a memory transfer rate of 448.0 GB/s. The memory clock is specified as 1750 MHz, which results in a 14 Gbps effective rate.

These specifications position the card at the 62nd percentile among all GPUs. Its performance in synthetic tests mirrors that standing: it scores 3DMark Steel Nomad 2011, Geekbench OpenCL 76957, and Vulkan 77402. The PassMark G3D score is 16462, and the card’s fractional FP32 throughput is 7.181 TFLOPS. Neighboring GPUs in the benchmark database include the NVIDIA GeForce RTX 3060 Mobile (delta 10% against 10), the AMD Radeon Pro 5700 (delta –0.5%), and the AMD Radeon RX 460 (delta –1.5%), all of which show that the RTX 2060 SUPER sits exactly in that performance tier.

In Thief’s measured data, the 8 GB VRAM does not appear to be a direct cap; the frame rates die primarily due to shading and fillrate constraint. On High at 4K, the average 36 FPS and the min 36 is not far from the 40 FPS of Medium, suggesting the card’s fixed function output is the binding factor. The RTX 2060 SUPER’s 256‑bit bus and 448.0 GB/s of bandwidth also help at 1440p, where it still delivers 62 FPS on High, but at 4K the number of raster primitives overwhelms the 105.6 GPixel/s peak pixel rate.

CPU Role

The AMD Ryzen 7 5800X, a Vermeer processor, is built on the Zen 3 architecture. It packs 8 cores and 16 threads, with a base clock of 3.80 GHz and a boost clock of 4.70 GHz. The L3 cache is 32 MB, and the CPU communicates on an AM4 socket with PCIe 4.0 support (20 CPU‑based lanes). It has a TDP of 105 W and an active production status. The chip’s benchmark suite positions it high among all desktop processors: it holds an average benchmark score of 29123 and sits in the 81st percentile of all CPUs.

In multi‑thread workloads, the 5800X shows undeniable strength. Cinebench R23 multi‑core score is 15476, while the single‑core score is 1574.5. In Geekbench, it scores 11036 multi‑core and 2032 singlecore. PassMark results confirm the same: multithread 27732, single‑thread 3445, and an overall average of 29123. Its nearest rivals in the database—the Intel Core Ultra 5 135H (0.1% higher), the Intel Xeon Phi 7210 (0.4% lower), the Core i5-13500HX (0.5% lower), and the Ryzen 5 5600XT (0.6% higher)—are all around this level.

No part of the measured FPS suggests the CPU is a bottleneck. The tiny difference between 1080p and 1440p outputs across presets is wholly explained by the GPU, because even at the CPU‑lowest settings (1920×1080 Low gives 155 FPS) the processor has headroom, as its performance indices are far above any possible limitation in a title from that era. The consistency of the FPS ratios across all presets implies that the 5800X is not ever waiting on the GPU or vice; the 3.8–4.7 GHz range is more than enough to keep the GPU feed at all measured resolutions.

How This Combo Ranks

In Thief’s formal ranking, this CPU‑GPU combination placed at rank 2591 out of 3720 possible combinations. That position puts it near the middle‑lower of the tested fleets. A rank of 2591 means that 2589 combinations performed better in this game, and only 1129 performed worse. The data available is a measured value (dataIsMeasured: true), so these numbers are drawn from real benchmark runs.

The combination’s rarity stems from the GPU’s percentile (62) being far lower than the CPU’s percentile (81). While the CPU alone is in the top twentieth percentile, the GPU’s performance relative to the current field pulls the overall rank down. This is typical for a pairing where the GPU is an older, end‑of‑life product and the CPU is a more recent multitasking champion. In Thief, the measured FPS yields produced by this pair are respectable but not class‑leading; the GPU’s own tier, represented by the RX 460 and the 3060 Mobile, is similarly placed.

Measured FPS Breakdown

The full measured dataset gives the exact frame rates for each setup combination.

1920×1080:

• Low: 155 avg FPS (no min/max data)
• Medium: 118 avg, with a minimum of 100 and a maximum of 136
• High: 90 avg
• Ultra: 86 avg

2560×1440:

• Low: 107 avg
• Medium: 81 avg (min 69, max 93)
• High: 62 avg
• Ultra: 59 avg

3840×2160:

• Low: 63 avg
• Medium: 48 avg (min 40, max 55)
• High: 36 avg
• Ultra: 35 avg

The Medium preset is the only one with min/max values recorded, so the 48 FPS average at 4K includes a range of 40 to 55 FPS and the 118 FPS at 1080p spans 100 to 136. The other presets provide only the pure average, so the exact frame pacing is unknown, but the averages are reliable measures.

For the player targeting 60 FPS, the 1080p Ultra frame rate of 86 FPS satisfies that easily. At 1440p, High gets to 62 FPS, which is just above 60, while Ultra is only 59 FPS—a marginal difference. At 4K, no preset exceeds 63 FPS, and only Low reaches it. Medium is 48, High drops to 36, and Ultra is 35.

Settings Recommendations

Given the measured averages, the optimal preset depends on the display resolution the user plans to use.

For 1920×1080 with a 60 Hz display, Ultra at 86 FPS provides a smooth threshold while maintaining maximum quality. The difference between High (90 FPS) and Ultra (86 FPS) is only 4 FPS, so Ultra is the better choice for visual fidelity. For users of high refresh rate monitors, Low yields 155 FPS, but even Medium at 118 FPS will saturate most 120/144 Hz panels.

At 1440p, the borderline between playable and not sits close to the High/Ultra boundary. High averages 62 FPS, which is just above 60, while Ultra drops to 59. If a stable 60 FPS is mandatory, choosing High is the safest recommendation. Medium at 81 FPS offers more headroom and still looks good, but High delivers 62 while retaining the extra detail.

At 4K, the CPU‑GPU pair can only produce a smooth experience on the lowest preset. Low delivers an average of 63 FPS, which is adequate to maintain a 60 FPS V‑sync, though the minimum is not reported. Medium sinks to 48 FPS, and both High and Ultra fall into the 35–36 FPS range, which is generally considered difficult for continuous motion. The data suggests that at 4K, a user must either accept a lower quality preset or reduce the display resolution.

The graphics settings have a distinct effect. At 1080p, moving from Low to Medium adds only a 37 FPS drop, while Medium to High costs another 28 FPS, and High to Ultra only 4 FPS. At 1440p, the drop from Low to Medium is 26 FPS, then 19 FPS from Medium to High, and 3 FPS from High to Ultra. At 4K, the pattern is 15 FPS from Low to Medium, 12 from Medium to High, and 1 from High to Ultra. This means that the Ultra preset is a marginal improvement over High, especially in 1440p/4K, while the jump from Low to Medium is significant in all resolutions. Therefore, for the best practical experience, selecting Medium or High rather than Ultra gives a high observable FPS boost at a minor graphical cost. Given the 2031 ranking, for a 1080p all‑out gamer, Ultra is a no‑brainer; for 1440p, High; for 4K, Low.

In summary, the Ryzen 7 5800X delivers a frame‑proved wide, but the RTX 2060 SUPER always stays the limiting factor. The measured FPS gives a clear sawtooth curve: at 1080p the card can drive Ultra well, at 1440p High is the maximum for 60 FPS, and at 4K only Low enables near‑60. The 8 GB VRAM and 448.0 GB/s bandwidth are sufficient at this game’s load, but the rasterization output dominates the outcome.