Thief

The combination of the AMD Ryzen 9 5950X and the AMD Radeon RX 6700 XT was evaluated in Thief (2014) across four presets (Low, Medium, High, Ultra) and three resolutions (1080p, 1440p, 4K), with all measurements directly observed. The collected dataset shows a wide spread of average frame rates, from 41 FPS at 4K Ultra to 171 FPS at 1080p Low, indicating that both CPU and GPU resources are engaged differently depending on the resolution and quality preset.

Resolution Scaling

The measured FPS data reveals how the workload shifts when resolution changes. At Low settings, the average frame rate falls from 171 FPS at 1920×1080 to 128 FPS at 2560×1440, a loss of 43 FPS, and then falls further to 75 FPS at 3840×2160, a further drop of 53 FPS. At Ultra settings, the corresponding sequence is 103 FPS, 70 FPS, and 41 FPS — the 1440p result is about 33 FPS lower than 1080p, and the 4K result is another 29 FPS lower. This monotonic decline across all presets confirms that resolution is the dominant scaling factor. For example, at 1080p High, the average is 108 FPS; at 1440p High, it is 74 FPS; and at 4K High, it is 43 FPS. The difference between 1080p and 1440p is 34 FPS, while the difference between 1440p and 4K is 31 FPS, showing a slightly smaller marginal loss when moving from 1440p to 4K compared to the 1080p to 1440p transition for High. At Medium settings, the 1080p average is 141 FPS, the 1440p average is 97 FPS, and the 4K average is 57 FPS — the 4K result is the only preset among those measured that carries a sub-60 average. The scaled behavior suggests that the GPU becomes increasingly responsible for the frame pacing as the pixel count grows, since the penalty for moving to 4K is roughly a 60% reduction in FPS relative to 1080p when considering Low settings (171 FPS to 75 FPS). To be precise, the reduction from 1080p to 4K at Low and Ultra is 96 FPS and 62 FPS, respectively. The overall pattern is a consistent decline with no stepwise anomaly; the data does not show a sudden inversion or a point where the CPU takes over, which would appear as a dramatic drop in scaling across resolutions.

GPU Role

The Radeon RX 6700 XT is the primary rendering component here, and its specifications align with the measured FPS patterns. It carries 12 GB of GDDR6 memory on a 192-bit bus, yielding a bandwidth of 384.0 GB/s. The GPU clock fitting: base 2321 MHz, game clock 2424 MHz, and boost 2581 MHz. Particularly at 1080p, the FPS numbers are high enough (e.g., 171 at Low, 141 at Medium, 108 at High, 103 at Ultra) to suggest the GPU itself is not the limiting factor at that resolution when using the 5950X as the CPU. The gap between 1080p and 1440p is moderate (about 20–40% when comparing median), but the gap on 4K is much smaller in absolute FPS, and at 4K the Ultra preset offers only 41 FPS — a drop of 62 FPS from the 1080p Ultra average.

The GPU’s benchmark data corroborate its class: it holds a percentileVsAllGpus score of 72, with an average benchmark score of 27,425. Its nearest competitor, the NVIDIA GeForce RTX 4070 Mobile, has an average score of 27,435, exactly the same (0% deltasPct), while the RTX 3090, a higher-end desktop card, is only 0.5% faster (average 27,565 vs 27,425). The RTX PRO 4000 Blackwell is 1.1% slower, showing the GPU is near-par with high-end mobile and. In contrast at 4K Ultra, the same GPU config gets only 41 FPS, which is less than a quarter of the 1080p Low number. This behavior points to a GPU with enough capacity for 1080p and 1440p but not for 4K maxed out. VRAM is unlikely to be a limit at these resolutions (the game’s requirement is unknown here, but 12 GB is a sizeable amount), yet the raw compute (13.21 TFLOPS FP32, 165.2 GPixel/s pixel fill) may still limit 4K workloads. When the 5950X is paired with this GPU, the reciprocal scaling indicates that the GPU thermal or power design (230 W TDP) is the barrier to higher 4K FPS. The measured 41–75 FPS at 4K are exactly the sort of numbers that fall within the GPU’s native ceiling, while the CPU’s strength remains untapped at that resolution.

Settings Recommendations

Based on the measured FPS alone, the user can choose a preset according to their target. The data shows that at 1080p every preset yields an average above 100 FPS (High 108, Medium 141, Low 171, Ultra 103). At 1440p, the Low and Medium produce 128 and 97 FPS, while High and Ultra yield 74 and 70 FPS. At 4K, only Low can achieve an average above 60 FPS (75 FPS); Medium gives 57, High 43, and Ultra 41. Thus, for a game played at 4K, the measured rows indicate that changing from Low to Medium reduces the average from 75 FPS to 97 FPS if moving to 1440p, but at 4K the Low preset is the sole option crossing 60 frames per second. Nor do the data contain any preset that slows below 40 FPS without the user manually adjusting other settings; the entire measured range is confined between 41 and 171 FPS. The likely sweet spot for visual quality versus smoothness is likely the High at 1440p (74 FPS) or Medium at 4K, if the user is willing to accept 57 FPS—a 24% reduction from Low’s 75 FPS. Yet since the pack does not include any frame-time percentile or consistency data (only average FPS, except for a single min/max at 4K Medium), precise specification of stable framerate is not possible. The safest stance is that measurable performance scales linearly with the preset order: Low always offers the highest FPS, Ultra the lowest, with Medium and High in between at every tested resolution. Data from the tables indicate that preset changes affect FPS by roughly 12–20 FPS per step at 1080p, and by about 15–18 FPS per step at 4K.

How This Combo Ranks

The rank for this CPU/GPU pairing in Thief is 1894 out of 3,720 tested combos (measured data). That places this configuration just below the middle of the overall distribution—approximately the 51st percentile from best (since 1 – 1894/3720 ≈ 0.491). To contextualize, the 5950X alone has a percentileVsAllCpus of 91, meaning it is in the top 9% of all CPUs, and the RX 6700 XT is at percentileVsAllGpus of 72. Yet in this game, the two-partCombined rank lands at the halfway mark, pointing to a combination that does not fully exploit the CPU’s strong standing. The rank is derived from the measured FPS at all exact, but since the game is a title with a 2014 release date, its engine may not be sensitive to the flagship thread count of the 5950X. Moreover, the combined performance is not evenly matched. If we compare to the GPU’s closest rivals (Radeon RX 6700 XT vs RTX 3090: -0.5% average score), the GTX 3090 likely produces higher FPS in this title, but those are not part of this data set. What the rank 1894/3720 shows is that many other CPU+GPU pairs achieve better frame rates in Thief. The gap between the CPU’s standing (91st percentile) and the combo’s standing (just below 50th) suggests the GPU is the primary bottleneck in this particular pairing—unlikely the same with a faster GPU would rise much higher.

CPU Role

The Ryzen 9 5950X is built on AMD’s Zen 3 architecture, featuring 16 cores, 32 threads, with base clock 3.40 GHz and boost 4.90 GHz, drawing 105 W TDP. In this package, the CPU contributes its computing power, but in Thief, the frame rate does not scale in proportion to the CPU’s capability. The multi‑core benchmark, such as Cinebench R23 multicore (26,017 points) and PassMark multithread (45,424), show a high level of raw perform, but the game’s per‑perspective graphs show that at 1080p (where the CPU has more influence, the FPS better) the high card reaches 171 FPS. Single-thread 3DMark score 944, C23 single-core 1,614, and PassMark single-thread 3,470) indicate strong IPC. The nearest CPU rivals lean to the 5950X: the Intel Core Ultra 5 235HX averages 52,073 points vs the 5950X’s 51,947 (delta -0.2%), the EPYC 8124P is -0.3%, the i9-13900F –0.4% (favourable) and the i7-14700 –0.7%. These deltaPcts are in the CPU’s synthetic benchmarks. In Thief’s FPS depth, the 5950X’s 192K, 32GB memory bandwidth (1) provides enough headroom for even the oldest titles, but this game is not likely to use all 16 threads. Since the 4K results stay low even when FPS are low, the CPU is not the limiter at 4K; rather the GPU. At 1080, the high FPS (100+) show that the 5950X is sufficient. The pattern is consistent across all settings: 1080p Low 171, 1080p High 108, while 4k Low 75, everything points to the GPU. Therefore, resources that the CPU supplies (16 cores / 32 threads, 64 MB L3) are underused. Still, one cannot conclude that the CPU is not relevant—the measured FPS at 1440p High (74) versus 4k High (43) shows that the CPU does not limit the 1440p result; hence the CPU can provide the required IPC to irritate the GPU at that degree.

FAQ

Q: What is the average FPS for this combination at 1080p with Ultra settings?

A: 103 FPS.

Q: At which resolution and preset does the FPS fall below 50?

A: 4K Ultra (41 FPS) and 4K High (43 FPS) are the two sub‑50 cases.

Q: How much FPS does the Low preset give at 1440p compared to 4K?

A: 128 FPS at 2560×1440 and 75 FPS at 3840×2160 — the 4K figure is 53 lower.

Q: Which preset gives the maximum average FPS at 4K?

A: Low, with 75 FPS, is the only 4K preset above 60 FPS.

Q: Does the rank of this combo better than half of the tested systems?

A: No, the rank is 1894 out of 3,720, meaning roughly 1,826 systems perform slower, so it is marginally slower than the median (the halfway point would be 1,860).

Q: What are the minimum and maximum FPS observed for 1080p Medium?

A: Minimum 120 FPS, maximum 163 FPS.

Measured FPS Breakdown

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

|---|---|---|---:|---:|---:|

| 3840×2160 | High | 43 | — | — |

| 3840×2160 | Low | 75 | — | — |

| 3840×2160 | Medium | 57 | 48 | 66 |

| 3840×2160 | Ultra | 41 | — | — |

| 2560×1440 | High | 74 | — | — |

| 2560×1440 | Low | 128 | — | — |

| 2560×1440 | Medium | 97 | 82 | 112 |

| 2560×1440 | Ultra | 70 | — | — |

| 1920×1080 | High | 108 | — | — |

| 1920×1080 | Low | 171 | — | — |

| 1920×1080 | Medium | 141 | 120 | 163 |

| 1920×1080 | Ultra | 103 | — | — |

The distinct pattern is plain: at 1080p, all presets maintain a triple‑digit average; at 1440p, the Low, Medium, and High presets remain above the 70‑FPS mark, with Ultra at exactly 70; and at 4K, only the Low preset reaches 75 FPS, while Medium, High, and Ultra fall to 57, 43, and 41 FPS, respectively. This table also shows that the only frame‑time data available (min/max) appear at Medium settings for each resolution, with the largest spread at 1080p Medium (120–163) and the smallest at 1440p Medium (82–112). At 4K Medium, the min/max is 48–66, indicating that even within the same preset the variance is substantial. These numbers serve as a direct reference for scalability and for any user who wishes to compare a specific setting.