Thief

The AMD Ryzen 7 5800XT and AMD Radeon RX 6600 pairing delivers measurable frame rates in Thief (2014) through twelve benchmark configurations, spanning three resolutions (1080p, 1440p, 4K) and four presets (Ultra, High, Medium, Low). Measured data ranks this combination at position 2892 out of 3720 tested combos for the game, placing it in the lower quartile of performance pairs. The CPU's single-thread index (3,535 in PassMark) and the GPU's modest 8 GB VRAM capacity define a system that behaves predictably across settings, but the raw numbers reveal sharp trade-offs between image quality and fluidity.

Settings Recommendations

The measured FPS table provides a clear tiering of playable presets. At 1920x1080, every preset sustains an average above 78 FPS, with Ultra and High delivering 82 and 78 respectively, Medium reaching 108, and Low topping out at 142. For a 60 Hz display, Ultra already offers a comfortable margin, but for 120 Hz monitors, Medium still doubles the refresh threshold. Moving to 2560x1440, the performance landscape narrows: Low averages 98 FPS, Medium 74, High 56, and Ultra 54. At this resolution, 60 FPS is only achievable with Low or Medium, while High and Ultra fall short. At 3840x2160, the situation worsens dramatically: Medium produces an average of 44 FPS, High 33, Ultra 32, and Low just 57. No preset reaches 60 FPS at 4K; only Low comes close, so the optimal preset for 4K is clearly Low, albeit with a lower fidelity experience. For a balance of visual quality and smoothness, Medium at 1440p offers the best measured compromise, posting 74 FPS average with a minimum of 163 and maximum of 85 — a range that keeps frame times consistent. Lowering to 1080p and selecting Ultra delivers 78 FPS, which is reasonable, but the data suggests Medium provides headroom for demanding scenes without sacrificing perceived quality. Ultimately, the recommended preset depends on the target resolution: 1080p Ultra for ‑60 FPS, 1440p Medium for ‑60 FPS, and 4K Low for ‑57 FPS.

GPU Role

The Radeon RX 6600 is built on the 7 nm RDNA 2 architecture with an 8 GB GDDR6 memory buffer on a 128-bit bus, delivering 224.0 GB/s of bandwidth. Its base clock is 1626 MHz with a boost of 2491 MHz, and a game clock of 2044 MHz. Omitting the CPU for a moment, the GPU's raw compute resources—1792 shading units, 120 texel mapping units, and 64 ROPs—translate to a pixel fill rate of 159.4 GPixel/s and a texture rate of 279.0 GTexel/s. Floating-point throughput is rated at 8.928 TFLOPS (FP32). In the context of Thief, a 2015-era title, the GPU's VRAM capacity and bandwidth are ample for the textures and effects that appeared in the measured presets. At 1080p, the GPU achieves strong average FPS across all presets; but the primary constraint is the GPU's pixel throughput, which is evidenced by the steep drop in FPS when resolution increases from 1080p to 4K, as discussed in the resolution scaling section.

Functionally, the 8 GB VRAM is sufficient for the game, since the 4K-High and 4K-Ultra presets are not limited by memory. In each case, the GPU's core clock and memory clock (1,750 MHz, 14 Gbps effective) work in tandem, but the scaling of FPS with resolution reveals a GPU-bound scenario. The GPU's proximity to its nearest rivals in aggregated benchmarks—such as the 0% delta with the NVIDIA Quadro K6000, -0.1% relative to the RTX 4050 Mobile, and -0.3% against the Tesla K20m—shows that the RX 6600 sits at a narrow performance envelope. Yet in Thief, the absolute frame rates indicate that the RX 6600 delivers playable performance only at 1080p and 1440p Medium/Low, while 4K remains outside a 60 FPS band. The core's 124 shading units and 28 RT cores are theoretical, but Thief does not require ray tracing, so the GPU's fixed-function units are exercised minimally; the bottleneck is instead the per-FLOP throughput and memory bandwidth.

Resolution Scaling

The data reveals a clear nonlinear relationship between resolution and average FPS. At 1080p, the measured averages for Low, Medium, High, and Ultra are 142, 108, 82, and 78, respectively. When resolution increases to 1440p, the average for Low falls to 98, Medium, and high/ultra to 56 and 54. The 1080p-to-1440p scaling ratios are: Low 142→98 (a 31% drop), Medium 108→74 (31%), High 82→56 (32%), Ultra 78→54 (31%). From 1440p to 4K, the drops are: Low 98→57 (42%), Medium 74→44 (41%), High 56→33 (41%), Ultra 54→32 (41%). The consistency of the percentage drop across all settings is striking: transitions from 1080p to 1440p reduce performance by about 31%, while 1440p to 4K reduces it by about 41%. Since 4K has 2.25 times the pixels of 1440p, the 41% throughput drop suggests that the GPU is not purely pixel-bound; rather, the limited memory bandwidth (224 GB/s) and the intrinsic fill-rate capabilities become the constraining factor as resolution rises. At 4K, even on Low, the GPU barely manages 57 FPS, whereas at 1440p Low it reaches 98. This scaling profile indicates that the GPU is the primary limiting component at all resolutions above 1080p, as the CPU's horsepower is unlikely to bottleneck a low-resolution task with such high core count and clock speed.

How This Combo Ranks

In the overall database of measured combinations for Thief, this processor+GPU pair holds a rank of 2892 out of 3720, meaning approximately 78% of tested combos deliver higher average frame rates in this game. This percentile—rank 2892/3720—places the system solidly in the lower performance tier, reflecting that the RX 6600 is a mainstream GPU from 2021, while the Ryzen 7 5800XT is a high-end CPU from 2024. The GPU's relative standing in its own benchmark pool is 63rd percentile, while the CPU is in the 81st percentile. That imbalance suggests that the CPU is not the limiting factor; the GPU's mid-range classification is the decisive parameter. Comparing to immediate CPU rivals from the benchmark data, the Ryzen 7 5800XT scores nearly identical to the Ryzen 7 7840H (0% difference), slightly higher than the Ryzen 7 8845HS and 7840HS (+0.3% each), and just below the Ryzen 5 9600X (−0.6%). None of these differences are material; the CPU is effectively neck-and-neck with its peers. In contrast, the GPU is surrounded by older professional cards like the Quadro K6000 and Tesla K20m, which are not current gaming GPUs, underscoring its mid-range positioning. In Thief, this combo ranks at the bottom of the tested spectrum, confirming that the GPU's 63rd percentile and the game's high demand at high resolution lock the system out of the upper echelons.

CPU Role

The Ryzen 7 5800XT, built on the Zen 3 (Vermeer) architecture with 8 cores and 16 threads, carries base and boost clocks of 3.80 GHz and 4.80 GHz, respectively. Its cache hierarchy includes 64 KB L1 and 512 KB L2 per core, share 32 MB of L3, and supports DDR4 memory at 51.2 GB/s through a dual-channel configuration. In Thief, a 2014 single-player stealth title, the CPU's involvement is most pertinent in frametime consistency and minimum FPS. Since the measured FPS lists min values for Medium settings only (for example, 92 at 1080p Medium, 63 at 1440p Medium, 37 at 4K Medium), the CPU's influence is subtle. The data indicates that at 4K Medium, the minimum FPS is 37, which is 16% below the average, indicating rendering malt spikes. Thief's CPU demands are modest; a set of eight physical cores with high boost is, in all probability, never the apex, as evidenced by the CPU's 81st percentile in CPU benchmarks. The CPU's scores—3,535 single-thread PassMark, 2,039 multi-thread, and 2,021 R20 single core—validate it as a strong single-thread performer, yet the game's frame rates from the measured data drop linearly with resolution, a behavior that cannot be attributed to the CPU. Instead, the CPU's role is to feed the GPU and process game logic; its high single-thread speed ensures that framerates in preset tests are limited by the GPU, not the processor. The fact that at 1080p Ultra the average FPS is 78, and 4K Ultra is 32, while the CPU's single-thread rating remains constant, further confirms that the CPU is not the bottleneck in Thief; the GPU is.

Measured FPS Breakdown

At 1920p1080, the measured averages are: Low 142 FPS, Medium 108 FPS (min 92, max 124), High 82 FPS, Ultra 78 FPS. The medium frame time range is tight (92-124), indicating stable performance. The jump from Ultra to Medium is +30 FPS (38%), and from Low to Ultra is 64 FPS (about 25% of total). At 2560p1440 (1440p), Low nin during 58, Medium 74 (63-85), High 56, Ultra 54. Here the medium range once again remains narrow, while ultra-ultra gap is only 2 FPS, suggesting resolution saturation. The 1080p Medium -> 1440p Medium drops from 108 to 74 (a 31% reduction), matching the scaling trend. At 3840p2160 (4K), low 57, medium 44 (37-50), high 33, ultra 32. The medium min/max 37-50 is a 35% spread, indicating sharper fluctuations at 4K. Even at low settings, 4K falls short of 60 average, while 1080p ultra sleeps well above that. All measurements come from a single setup; there are no min/max values for most configurations, limiting a full analysis of frame time stability. The data reveals a monotonic inverse relationship between average FPS and both preset quality and resolution, with the GPU core load reaching higher peaks as more shading work is demanded.

FAQ

Q: Is this combo able to run Thief at 4K with 60 FPS on any preset?

From measured results, none of the presets at 4). The highest average FPS is 57 at Low (Medium average 44, High 33, Ultra 32). So a 60 FPS average is not achieved at 4K under any quality level.

Q: Which resolution gives the best balance of visual detail and smoothness?

At 2560p1440p, the Medium preset delivers an average of 74 FPS with a minimum of 63, which is the only 1440p setting that securely stays above 60 FPS. The High and Ultra settings fall to 56 and 54 respectively. If 60 FPS is the hard requirement, only 1440p medium is viable, while 1080p can run all presets above 78.

Q: What is the CPU's contribution to the frames, given that it has 8 cores and 16 threads?

In the measured data, the CPU is not the limiting component. The consistent scaling of average FPS with resolution (31% dropout from 1080→1440, 41% from 1440→4K) clearly points to a GPU-bound situation. The Ryzen 7 5800XT's strong single-thread score (3,535 PassMark) ensures it does not bottleneck the RX 6600 in a game like Thief, which uses fewer than 8 threads.

Q: How does the GPU's VRAM capacity influence Thief's frame rates?

With 8 GB of GDDR6 on a 128-bit bus and 224 GB/s bandwidth, Thief's quality presets appear to stay within this memory limit, as no resolution shows a sharp drop beyond a linear ratio. The 4K (37 min on Medium) or 4K-Low (57 avg) still sustains memory-adjacent operation, so no memory out-of-core stalls are detected; the framerate drops are due to the GPU compute limits.

Q: In the overall database ranking, how does the combo's performance group compare in percentile?

The combo ranks 2892nd out of 3720, which puts it at roughly the 22nd percentile of the best combos (i.e., 22% of tested systems are slower). This places the system well below the upper half, not from the CPU side but due to the RX 6600's 63rd percentile in the GPU class. In Thief, a title with low overall demands, the GPU is the limiting factor.

Q: Are there any settings that provide 120+ FPS in Thief?

Yes—at 1920p1080 Low, the average FPS is 142, and Medium also exceeds 120 (108), but no other configuration reaches 120. At 1440p, the highest is Low with 98 FPS, so at 1440p the system achieves 60 FPS on Medium but skips 120.