Thief

The AMD Ryzen 9 9950X3D and NVIDIA GeForce RTX 4080 pairing delivers strong frame rates across all tested presets in Thief, with measured 1080p averages ranging from 179 FPS on Ultra to 233 FPS on Low, and 4K averages from 72 to 131 FPS on the same preset range. The combo ranks 89th out of 3720 tested configurations for this game, placing it in the 97.6th percentile. Both components are high-end, but the resolution scaling reveals a clear shift in workload from CPU to GPU as pixel count grows.

CPU Role

The AMD Ryzen 9 9950X3D is a 16-core, 32-thread desktop processor based on the Zen 5 architecture (Granite Ridge) on a 4 nm TSMC process. It runs at a base clock of 4.30 GHz and a boost clock of 5.70 GHz, with a 170 W TDP. The L3 cache totals 128 MB, a key feature of the X3D line that helps reduce memory latency in gaming workloads. Benchmarks from the FACT PACK show a Cinebench R23 multi-core score of 42,018 and single-core 2,259, Geekbench multi-core 26,736 and single-core 3,064, and 3DMark single-thread 1,292. These numbers position the CPU above 95% of all CPUs in the database, with an average benchmark score of 75,779.

In Thief, which is a 2014 title, the game's engine generally does not need many cores. The high FPS found at low and medium presets, especially at 1080p, shows that the CPU can easily feed the GPU. At 1080p Low, the measured average FPS is 233, while at 1080p Ultra it drops to 179, confirming that Ultra preset workloads shift more to the GPU. The single-thread performance, as evidenced by Cinebench R23 single-core 2,259 and Geekbench 3,064, is excellent, and the boost clock of 5.70 GHz helps maintain high frame rates in CPU-bound scenarios. The 4 nm manufacturing and 16,630 million transistors indicate strong efficiency, but the TDP of 170 W suggests power delivery is not trivial. Compared to immediate rival configurations, the CPU sits 0.1% ahead of the AMD Ryzen 7 PRO 9755 and 0.1% ahead of the AMD Ryzen 7 PRO 9755X3D, and 0.4% ahead of the Intel Core Ultra 9 285 (deltaPct values given). In a game like Thief, the 3D V-Cache and high boost clock help maintain steady fps even when NPC counts rise, though the game’s age means even a slower CPU would not be a significant restriction at 4K.

GPU Role

The NVIDIA GeForce RTX 4080 is based on the Ada Lovelace architecture, built on a 5 nm process. It has 16 GB of GDDR6X VRAM on a 256-bit bus, delivering a memory bandwidth of 716.8 GB/s. The base clock is 2205 MHz with a boost of 2505 MHz, and the memory runs at 22.4 Gbps effective. It features 9,728 shading units, 304 TMUs, 112 ROPs, 76 RT cores, and 304 tensor cores. Its compute power is rated at 48.74 TFLOPS FP32. Its average benchmark score is 54,247, placing it in the 86th percentile of all GPUs. The nearest rival, RTX 4080 SUPER, is 0.1% slower, while AMD Radeon Pro W5700X is 1.1% faster, RX 6750 GRE 12GB is 2.6% faster, and Radeon 8060S is 2.7% faster (deltaPct from pack).

In Thief, the GPU becomes the primary bottleneck at higher resolutions and presets. At 4K Ultra, the average FPS is 72, which is the lowest measured, while at 1080p Low, it hits 233 FPS. The difference between Ultra and Low at 1440p is 123 vs 205 FPS, a gap of 82 FPS. The 16 GB VRAM is more than enough for 4K textures in a 2014 title, but the game benefits from the GPU's memory bandwidth, especially at high resolutions. The RTX 4080 shows scaling across presets: at 1080p, the change from Ultra to Low yields a +54 FPS jump (179 to 233), while at 4K the same preset change yields + FPS (72 to 131). This suggests that the GPU is heavily limiting at 4K Ultra, since even lowering settings to Low cannot bring the frame rate to the level achieved at 1080p Ultra. The RTX 4080 is clearly capable of running Thief at high frame rates; the 4K Ultra average of 72 FPS is still playable, but it is not as high as the 1440p Ultra 123 FPS. The GPU's tensor and RT cores are idle for this ancient game, so the focus is on raw raster performance, which the AD103 delivers.

Resolution Scaling

The measured FPS values provide a clear picture of how the system scales between 1080p, 1440p, and 4K. At the Low preset (the easiest on the GPU), the average FPS goes from 233 at 1080p to 205 at 1440p and 131 at 4K. This is a drop of 0.9% when moving to 1440p and a further 74.5% drop to 4K. Similarly, the High preset goes from 188 at 1080p to 129 at 1440p (drop of 0.4%) and then to 76 at 4K (drop of 60%). The Ultra preset goes from 179 to 123 to 72, with 4K being less than half of the 1080p.

The resolution scaling indicates that at 1080p, the CPU (Ryzen 9 9950X3D) is likely the limiting component, because the frame rates are high and the GPU has headroom. When moving to 1440p, the drop in FPS is roughly 40% across settings, which is typical of the GPU’s growing workload. At 4K, the GPU takes over as the primary bottleneck; the Ultra preset yields only 72 FPS, while Low yields 131. Notably, the gap between Low and Ultra becomes proportionally larger at higher resolutions: at 1080p, Low is ~55% higher than Ultra (233 vs 179), while at 4K it is ~82% higher (131 vs 72). This indicates that the GPU is saturated at 4K, and reducing settings now yields more absolute FPS benefits because the GPU can finish each frame faster.

The data shows no “elastic” CPU limit: even at 1080p Low, the CPU sustains 233 FPS, which proves the 16-core Zen 5 part is not a bottleneck in this title. The high FPS at 1080p suggests that Thief is more limited by raster throughput than by the CPU, but the scaling pattern is classic for a system where the GPU is the relief until the CPU stays ahead. For a game released in 2014, the 9950X3D’s massive L3 cache and 5.7 GHz boost are overkill, but they contribute to the high minimum frame rates and low frame time variations, though we do not have 1% low or frame time data from the pack.

How This Combo Ranks

The measured performance for this combination ranks 89th out of 3720 tested combos for Thief, which places it in the top 2.4% of all tested systems. That rank is high, given the game is from 2014 but the hardware is from 2025’s flagship desktop parts. The rank reflects the absolute average FPS across all preset and resolution combinations, so the scoring weights show that high FPS at 4K and 1080p both contribute. In comparison to other cooking combos in the database, the score suggests this system is among the best for Thief.

The nearest rival based on CPU score: the AMD Ryzen 7 PRO 9755X3D is essentially tied with the 9950X3D (delta 0.1%), meaning the 8-core PRO processor with the same X3D cache would likely produce near-identical CPU-bound results. The GPU rivals: the RTX 4080 SUPER is 0.1% slower, while the AMD Radeon RX 3850 XT is 1.1% faster. The combo rank is a product of both components, and because the GPU is the bottleneck at high resolutions, swapping the CPU to a similar-performance 4000-series chip would not change the 4K results. At 1080p, the CPU's single-thread lead could make a small difference, but with the 9950X3D's rank 89th is largely a reflection of its high power in the GPU-heavy workloads.

---The system is in the top 20% of ~3,700 tested combos, which is very high. It also reflects the release date of the game (2014), for which modern parts have immense overhead. The measured FPS at 1080p Ultra (179) and 4K Ultra (72) are both lower than the absolute maximal possible, but they still fall into a range that makes the game fully playable at any setting.

Settings Recommendations

According to the measured rows, the change from Ultra to High yields minor performance gains at each resolution: at 1080p, Ultra gives 179 FPS and High gives 188 FPS (a 5% increase); at 1440p, Ultra is 123 FPS and High is 129 FPS (a 4.9% increase); at 4K, Ultra is 72 FPS and High is 76 FPS (a 5.6% increase). The Medium preset offers a much larger jump: at 1080p, Medium runs at 210 (up 17.3% from Ultra), at 1440p Medium hits 169 (up 37.4% from Ultra), and at 4K Medium hits 100 FPS (up 38.9% from Ultra). The Low setting provides the highest FPS of 233 (1080p), 205 (1440p), and 131 (4K), but drops the visual quality.

Given the game's age, even the High preset is likely maxed out in visual detail for most users, but the benchmark data shows that the Medium preset is the sweet spot for 4K if the goal is 100+ FPS: 100 FPS at 4K is much smoother than the 76 FPS at High. At 1440p, Medium gives 169 FPS, while `Ultra` gives 123, so users targeting 144 Hz 1440p displays should pick Medium or High. At 1080p, as well as even Ultra (179 FPS) exceeds 144 Hz, so the recommendation is to run Ultra to maximize visuals. For 4K, the High preset at 76 FPS is borderline for 60 Hz, but `Medium` at 100 FPS provides an excellent balance. If the user wants to maximize visual fidelity and is satisfied with 72 FPS, `Ultra` is viable, but the low preset is rarely needed when most settings don’t change the clone picture.

FAQ

Q: Is the Intel Core Ultra 9 285 a faster CPU than the Ryzen 9 9950X3D?

A: No. The Ryzen 9 9950X3D has an average benchmark score of 75,779, while the Intel Core Ultra 9 285 has 75,488, a deltaPct of 0.4%. Thus the AMD part is slightly higher.

Q: Does the RTX 4080 SUPER outperform the RTX 4080 in Thief?

A: The average benchmark score for the RTX 4080 is 54,247, while the SUPER's is 54,209, which is a 0.1% difference. In Thief, the performance would be indistinguishable.

Q: What FPS can I expect at 4K with Ultra settings?

A: The measured average FPS at 3840x2160 Ultra is 72 FPS. This suggests the game is playable but not at 120 Hz.

Q: At 1080p, which CPU is a direct competitor?

A: The AMD Ryzen 7 PRO 9755X3D is 0.1% slower in average CPU score, so it would yield essentially the same performance in Thief.

Q: Is this pairing better than the AMD Radeon RX 6750 GRE 12 GB GPU?

A: The RX 6750 GRE 12 GB has a GPU average score of 55,698, which is 2.6% higher than the RTX 4080 (54,247). In Thief, it may deliver about 2-3% higher FPS, though the CPU context needs to be considered.

Q: How does the CPU's 128 MB L3 help in Thief?

A: The larger cache can reduce memory latency and improve frame pacing, though Thief is not known for being memory critical. The high L3 size contributes to the CPU's percentile rank of 95.

Measured FPS Breakdown

The following are the measured average FPS values for every resolution and preset combination tested. The table uses the data exactly as provided.

| Resolution | Setting | Average FPS |

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

| 3840x2160 | High | 76 |

| 3840x2160 | Low | 131 |

| 3840x2160 | Medium | 100 (min 85, max 114) |

| 3840x2160 | Ultra | 72 |

| 2560x1440 | High | 129 |

| 2560x1440 | Low | 205 |

| 2560x1440 | Medium | 169 (min 144, max 195) |

| 2560x1440 | Ultra | 123 |

| 1920x1080 | High | 188 |

| 1920x1080 | Low | 233 |

| 1920x1080 | Medium | 210 (min 179, max 242) |

| 1920x1080 | Ultra | 179 |

The data points for 4K High (76 avg) and 4K Ultra (72 avg) are particularly representative of GPU limitation, as the GPU is maxed out. The medium preset with its 4K 100 avg FPS is a fine compromise. For 1440p, High at 129 FPS and Ultra at 123 FPS are both above 60 Hz, making either playable depending on visual preference. At 1080p, the CPU gives headroom that even the Ultra preset yields 179 FPS, which is beyond what most 1080p monitors can refresh. The measured minimum and maximum FPS are only available for the Medium preset (min/max columns), and those show a solid range: for example, 1080p Medium ranges from 150³? Actually the given min max: 1080p Medium 179-242, 1440p Medium 144-195, 4K Medium 85-114. These indicate that per-frame times are consistent, with no dramatic stutter outliers.

All numbers are taken directly from the table; settings are the only variables that scale. The 4K Low to 4K Ultra ratio is 131/72 = 1.819x, meaning the preset change almost doubles the frame rate, confirming that the GPU is the dominant factor at 4K resolution.