Thief

This page presents measured performance data for Thief running on a system pairing an AMD Ryzen 5 3600 processor with an NVIDIA GeForce RTX 4070 Ti SUPER graphics card. The dataset includes average frame rates at three different resolutions and four graphics quality settings, along with synthetic benchmark scores for both components and the overall combination’s rank against other tested systems.

FAQ

Q: What is the rank of this CPU+GPU combination in Thief’s benchmark results?

A: The combo ranks 1701st out of 3720 tested combinations. That places it in the upper half of all tested systems.

Q: How does the CPU stack up against its nearest rivals in average benchmark score?

A: The Ryzen 5 3600 has an average score of 17035. Its closest rival, the Intel Core i7-1260P, scores 17007, making the AMD part 0.2% higher. Conversely, the Intel Core i5-11400 scores 17067, which is -0.2% relative to the Ryzen chip.

Q: What is the GPU’s memory architecture and bandwidth?

A: The RTX 4070 Ti SUPER is equipped with 16 GB of GDDR6X memory on a 256-bit bus, delivering a memory bandwidth of 672.3 GB/s.

Q: Which measured configuration produces the highest average frame rate in this dataset?

A: The fastest average FPS is achieved at 1920x1080 with Low settings, reaching 126 frames per second. The same resolution at Ultra still averages 101 FPS.

Q: How does the CPU’s single‑thread performance compare to its multi‑thread results?

A: In Cinebench R23, the CPU scores 2124 in the single‑core test and 15045 in the multi‑core test. The single‑core number is about 14% of the multi‑core result. In Geekbench, it records 1536 single‑core and 7372 multi‑core.

Q: What does the FPS spread look like across quality levels at 1440p?

A: At 2560x1440, the average goes from 111 FPS on Low, to 100 on Medium, to 91 on High, and down to 89 on Ultra. The difference between Low and Ultra is 22 FPS.

GPU Role

The GPU is a GeForce 40-series part based on the AD103 chip with Ada Lovelace architecture. It runs at a 2340 MHz base clock and a 2610 MHz boost clock. The card has 16 GB of GDDR6X memory with a 256-bit interface, delivering 672.3 GB/s of bandwidth. Those numbers show that the GPU has a large frame buffer and high-speed memory, both of which matter for texturing and shader workloads.

In Thief, the measured frame rates reflect the GPU’s capability. The game’s highest measured average FPS at 3840x2160 occurs on Low at 93 FPS, and even at Ultra the average stays at 67 FPS. At 1080p Ultra the GPU pushes 101 FPS average, which is only 10 frames below the Low preset’s 111 FPS? Wait, at 1080p Low is 126, Ultra is 101. Actually the frame rate never drops below 60 FPS at any preset in the dataset, indicating that the GPU is not the limiting factor in this game. The 16 GB VRAM is ample for Thief’s texture demands, even at 4K. The card’s memory bandwidth of 672.3 GB/s allows high fill rates, though the 4K Ultralows to 67 FPS suggests that compute and ROP throughput may be more important at that resolution than raw bandwidth.

GPU synthetic scores like passmark_g3d at 31829 and passmark_gpu_compute at 18372 show strong raw performance, but they do not directly correlate with in‑game frame rates. Still, the GPU’s 8448 shading units and 96 ROPs help explain why it can maintain playable rates at 4K.

How This Combo Ranks

The combination of ryzen 5 3600 and RTX 4070 Ti SUPER occupies rank 1701 of 3720 tested combos for Thief. With that rank, it outperforms nearly 54% of all systems tested in the game. This rank places it in the second quartile of the overall distribution. The exact position indicates that this configuration delivers solid, though not top‑tier, performance for this title.

The components individually have different standing: the CPU has a 71st percentile against all CPUs, while the GPU has a 76th percentile against all GPUs. The CPU sits slightly above the 71% mark, and the GPU is slightly above the 76%, so the system is roughly in line with what those percentile placement would suggest. The combination’s rank being 1701 out of 3720 aligns with the fact that the CPU can maintain high frame rates when paired with a stronger GPU. The CPU’s average benchmark score of 17035 falls within 0.2% of its nearest rivals, which highlights that minor performance differences among comparable CPUs are unlikely to produce large shifts in game FPS when the GPU is the dominant component.

Resolution Scaling

The data shows a clear relationship between resolution and frame rate. At High settings, the average FPS goes from 103 at 1080p to 91 at 1440p and then to 70 at 4K. That is a 25% frame‑rate drop between 1080p and 4K, and the drop is more pronounced at lower settings. At the Low preset, the same sequence is 126, 111, and 93 FPS. Thus, decreasing resolution boosts FPS significantly at all quality levels, but the absolute increase is much larger at Low than at Ultra.

When the settings are at the lower end, the GPU is dispatching fewer tasks per frame, so the CPU has more room to feed it. At 1080p Low, the average is 126 FPS; the 6 cores and 12 threads of the Ryzen can easily keep the queue full. At 4K Ultra, the average drops to 67, showing that the GPU is heavily loaded and the CPU has capacity left over. The fact that the drop from 1080p to 1440p is smaller for High (103 to 91) than for Low (126 to 111) indicates that the GPU becomes more dominant as quality rises. At the Medium preset, the 1080p vs 1440p difference is 114 to 100, and 1440p to 4K is 100 to 84. Naturally, the scaling is not perfectly linear because the engine of 4K roughly forces the GPU to render more fragments and fills, which is the primary workload in Thief.

The limiting component is therefore the GPU, regardless of the preset. The CPU’s performance already high enough at 1080p—the FPS at 1080p Low (126) is limited by the GPU’s ability to rasterize frames, not by CPU logic. If the CPU were the bottleneck, the 1080p results would be roughly the same across presets; they are not.

Measured FPS Breakdown

| Resolution | Setting | Average FPS | Min FPS | Max FPS |

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

| 3840×2160 | High | 70 | - | - |

| 3840×2160 | Low | 93 | - | - |

| 3840×2160 | Medium | 84 | 71 | 97 |

| 3840×2160 | Ultra | 67 | - | - |

| 2560×1440 | High | 91 | - | - |

| 2560×1440 | Low | 111 | - | - |

| 2560×1440 | Medium | 100 | 85 | 115 |

| 2560×1440 | Ultra | 89 | - | - |

| 1920×1080 | High | 103 | - | - |

| 1920×1080 | Low | 126 | - | - |

| 1920×1080 | Medium | 114 | 97 | 131 |

| 1920×1080 | Ultra | 101 | - | - |

At 1080p, the highest average is 126 FPS on Low, which drops to 101 on Ultra. The variation across, 25 FPS, shows that spacing settings primarily impacts your headline Frame, while the GPU headroom remains large. At 1440p, the range is from 111 (Low) to 89 (Ultra), a 22 FPS gap. At 4K, the spread is 93 (Low) to 67 (Ultra), a 26 FPS gap. The min/max values are only recorded for the Medium preset; at 1080p Medium the min is 97 and max is 131, at 1440p Medium it is 85 to 115, and at 4K Medium the range is 71 to 97. These limits indicate that the frame rate is least stable at 4K Medium, with the 71 to 97 divergence being the largest absolute spread in the dataset.

The 4K Medium result has a min of 71 and average of 84, meaning it has a range of about 13 FPS above average. That inconsistency suggests that the memory bandwidth (672.3 GB/s) or the Raster throughput is occasionally hit by a hit‑and‑miss situation. However, the highest average at any resolution is the 1080p Low at 126, which is above the threshold where frame drop are noticeable.

CPU Role

The AMD Ryzen 5 3600 is a 3000‑series discrete unit with 6 cores and 12 threads. It has a base clock of 3.60 GHz and a boost clock of 4.20 GHz, and it utilizes the AM4 socket with Zen 2 architecture from Matisse. Its 8 core blocks share a total of 32 MB L3 cache, while each core also have 512 KB L2 and 64 KB L1. The CPU has a 51.2 GB/s memory bandwidth to the dual‑channel DDR4 controller and supports PCIe Gen4 lanes.

In synthetic evaluations, the CPU yields a Cinebench R23 multi‑core score of 15045 and a single‑core score of 2124. A Geekbench multi‑core of 7372 and single‑core score of 1536 further attest to its capability. These numbers demonstrate it can comfortably handle the logical threads required for games like Thief, which is released in 2014 and is not heavily multi‑threaded.

The CPU’s nearest rivals in the benchmark pool are the Intel Core i7-1260P, which has an average score of 17007 (0.2% lower), and the Intel Core i5-11400, scoring 17067 (0.2% higher than the Ryzen). These tiny margins suggest that the Ryzen 5 3600’s raw processing power is essentially equivalent to many mainstream CPUs released at around its time. In Thief, the 6‑core/12‑thread design is ample for the game’s AI, physics, and streaming. The turn‑by‑turn analysis shows that the CPU is almost never the limiting factor: at 1080p, which uses lower resolution and thus lower GPU load, the frame rate remains high across all settings, indicating that the CPU can feed the GPU.

The measured FPS at 1080p Low (126) is a strong signal that the CPU can output more than enough frames. Lowering the resolution does not increase FPS proportionally, because the GPU’s raster limit is reached. If the CPU were slow, we would see frame rates bounce at low resolution when the GPU needs more frames per second; the average 126 FPS is far above typical gaming—the 1% low would be a better indicator, but that data is not provided. The CPU’s single‑core performance, 861 on the Geekbench single core, is also above the threshold for being a bottleneck in a game that is almost a decade old.

In synthetic CPU benchmarks, the 3600 shows a PassMark multithread score of 17700, a PassMark single‑thread of 862, and a Cinebench R15 multi‑core 1516 against a single‐core 214. These numbers show the CPU holds its own against mobile and older desktop parts, and they tend to follow the same pattern: the CPU is the stable component in this pairing, letting the GPU's high geometric capability deliver fluctuation in frame rate only driven by resolution and game settings.