Thief

Thief, a 2014 stealth title, places the Intel Core i5-13400F with an AMD Radeon RX 570 in a configuration that ranks 3551st out of 3,720 tested combinations, putting it in the lower percentile of systems on this benchmark database. The CPU is a 10-core, 16-thread part from the Core 13th Gen line with a base clock of 2.50 GHz and a boost clock of 4.60 GHz, carrying a launch MSRP of $196. The GPU is the AMD Radeon RX 570, a 14nm Polaris 20 chip with 4GB GDDR5 memory, a base clock of 1168 MHz, and a boost clock of 1244 MHz, its launch MSRP $169 USD. Measured across three resolutions and four settings presets, the combination shows moderate performance at 1080p but falls off sharply at higher resolutions, a pattern that reflects the GPU's limits rather than the CPU's capabilities.

Resolution Scaling

The measured FPS data shows a clear slide from 1080p through to 4K. In the Low preset, the average frame rate descends from 75 FPS at 1920x1080 to 51 FPS at 2560x1440 and further to 30 FPS at 3840x2160. This is a steep decline: the 4K figure is exactly the same as the Low preset's 1440p result, indicating a halving of frames when stepping from 1440p to 4K at the same settings. Medium follows a similar pattern: 57 FPS at 1080p, 39 at 1440p, and 23 at 4K. High goes from 43 to 30 to 17, and Ultra from 41 to 28 to 17.

The character of these drops reveals that the system's limiting component is firmly on the graphics side. At 1080p, the frame rate remains relatively high—the GPU can still render the scene with the CPU feeding it frames efficiently. But as the pixel count grows, the frame rate collapses: the 1080p to 1440p step drops by roughly a third for each preset, and the step to 4K halves performance again. The RX 570, with its 34 GB/sec of bandwidth and 4 GB of VRAM, is clearly the component being stressed; the CPU's 3DMark 2-thread and single-thread performance (1,880 and 960) are far above what Thief demands in the measured rows. The data points to a scenario where the GPU's fill rate and memory bandwidth are exhausted well before the CPU's 10 cores become a factor, as the identical 17 FPS average at 4K for both High and Ultra suggests the GPU is saturated at that resolution regardless of preset.

The single-thread and multi-thread workloads are not bottlenecked at any resolution because the CPU carries a 20MB L3 cache and a high boost clock that keeps it ahead of the GPU's feed—the 1080p Low result of 75 FPS indicates the CPU can comfortably sustain 75 FPS, while the GPU's 4K ceiling of 30 FPS is the limiting cliff.

GPU Role

The AMD Radeon RX 570 is designed for 1080p and Gen 4 eras, but Thief’s engine perches on it unevenly. The GPU’s 1168 MHz base and 1244 MHz boost clocks, combined with 32 ROPS and a 2,048 shading unit count, are enough to handle the 2014-era geometry at lower settings. Its VRAM pool of 4GB GDDR5 over a 256-bit bus gives 224 GB/s of bandwidth, but that is not enough to keep up at 4K Ultra, where the measured average droops to 17 FPS—the same as High, and both exactly half of the Low preset’s 30 FPS at that resolution. The pixel fill rate and texture rate are 39.81 GPixels/s and 159.2 GTexels/s respectively, but the primary pressure point is the memory subsystem: 4 GB is quickly saturated when textures are side by side.

At 1080p, the GPU’s capabilities show a clearer picture. With Low settings, the system produces 75 FPS, which is an acceptable frame rate, and even Medium and High stay above 50 FPS. The 4 GB VRAM is sufficient for 1080p and many settings, but the drop when moving to 1440p (where Medium yields 41) shows the chip reaching its limits. The RX 570’s clock speed differences (base vs boost) don’t change the picture; the measured results are consistent with a card that can hold 1080p but struggles beyond that. The nearest rivals in the database—AMD Radeon RX 680M, R9 M295X, RX 470, and Quadro RTX 8000—have average scores within 1% of each other, indicating that the RX 570 sits in a performance class where small differences matter.

Settings Recommendations

For the best experience per the measured rows, the data suggests using Medium at 1920x1080, where the average frame rate hits 57 FPS, and the min and max range spans 54 to 59 FPS (min 41, max 65) depending on scene. This gives a solid blend of visual quality and fluid game. If a player prefers D a higher frame rate, the Low preset delivers 75 FPS while keeping 1080p, but the visual sacrifices are substantial. On the other hand, High and Ultra at 1080p both average in the 40s (High 43, Ultra 41), which is still playable but does not hit the modern target of vsync 60.

At 2560x1440, the situation tightens. Low is the only preset that stays above 50 FPS—51 average—and provides a good option to the user. Medium averages 39 FPS (min 33, max 45), which may be acceptable for a slower-paced stealth game, but it’s borderline. High and Ultra fall to 30 and 28 FPS respectively, meaning that even at lower preset the frame rate is below the common threshold for smooth action. At 3840x2160, every preset stops below 30: Low 30 (marginal), Medium 23, High 17, Ultra 17. For a stealth title, a 23 FPS experience is far from ideal; only the Low preset at 4K passes the playability bar, and even that is below the usual 60 FPS target.

The measured data makes it clear that the only settings that deliver a comfortable experience are 1080p with Medium or Low presets. To avoid stutter and preserve your built-in, a resolution of 1440p with Medium can be tolerated, but 4K should be left to better hardware.

How This Combo Ranks

The combo rank of 3551 out of 3,720 tested configurations places it near the bottom of the barrel, in the lower 5% of all combinations. This indicates that, relative to other CPU-GPU pairs in the database, this pairing does not shine. The CPU itself—the Intel Core i5-13400F—has a percentile rank of 77 among all CPUs, and the GPU, the RX 570, sits at percentile 74 among all GPUs. Yet as a pair, they fall behind most other combinations, likely because the GPU’s memory bandwidth and 4 GB VRAM limit the potential of the fast CPU. The CPU’s nearest rivals in the specification database include the AMD Ryzen 9 6900HS, which has an average score of 25,294 (0% delta) and the Intel Core 5 120 in a -0.3% delta, while the GPU’s nearest rivals include the RX 680M ( -0.4% ), R9 M295X ( +0.7% ), and RX 470 ( -0.8% ). These small gaps show that the components individually are mid-tier, but the combination in Thief falls far short of what other pairings achieve, often because of the GPU’s age.

The frame data confirms that the pairing is not a balanced one: frame rates at 1080p are decent, but at 4K they fall to the 17-30 FPS range, which place the system's cumulative in the 3551st position. The large number of better-scoring combinations—3,720 total—indicates that most other tested combos deliver higher frame rates in this specific title. The rank is derived from measured FPS across all resolutions and settings, so this is a robust indicator that the GPU is the main drag inside the performance envelope.

CPU Solidarity

The Intel Core i5-13400F is a 10-core, 16-thread CPU with a base clock of 2.5 GHz and a boost of 4.6 GHz. It contains 20 MB of L3 cache and runs on a 10 nm Raptor Lake-S die; it will support DDR4-3600 and covered. Its cinebench r23 multi-core score of 22,604 and single-core of 3,191 underscore a strong CPU that is far from the GPU's level. In Thief, the CPU has little direct effect on the FPS limits—instead, the measured FPS data shows that the CPU rarely becomes the limiter. At 1080p Low, the system 67 FPS, which is still relatively close to the CPU’s likely during that scene, but at 4K the same CPU yields only 17-30 FPS, the same CPU core that could deliver 75 FPS at 1080p is idle-baiting because the GPU can’t keep up.

The CPU’s 10 cores and 16 threads haven’t fully exploited by Thief’s engine; the game is from 2014 and uses fewer threads, so the heavy single-thread performance (a geekbench single-core score of 1,996) is what matters. With a boost clock of 4.6 GHz across P-cores, the CPU’s response is enough to fuel the GPU at 1080p, but when the GPU’s pixel rate falls off, the CPU has headroom to spare. The benchmark scores show this CPU is roughly equal to the AMD Ryzen 9 6900HS (delta 0%), the Intel Core 5 120 (‑0.3%), and the Ryzen 7 6800H ( +0.4%) in its that CPU benchmark class, meaning the CPU is competitive but the measured game results pay not off the CPU strength.

Actual FPS Numbers and Measured FPS Breakdown

The table below lists the exact measured averages, minimums, and maximums for each settings preset and resolution, taken from the database's "measured" rows.

| Resolution | Setting | Avg FPS | Min FPS | Max FPS |

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

| 3840x2160 (4K) | High | 17 | — | — |

| 3840x2160 (4K) | Low | 30 | — | — |

| 3840x2160 (4K) | Medium | 23 | 19 | 26 |

| 3840x2160 (4K) | Ultra | 17 | — | — |

| 2560x1440 (1440p) | High | 30 | — | — |

| 2560x1440 (1440p) | Low | 51 | — | — |

| 2560x1440 (1440p)| Medium | 39 | 33 | 45 |

| 2560x1440 (1440p)| Ultra | 28 | — | — |

| 1920x1080 (1080p)| High | 43 | — | — |

| 1920x1080 (1080p)| Low | 75 | — | — |

| 1920x1080 (1080p)| Medium| 57 | 48 | 65 |

| 1920x1080 (1080p)| Ultra | 41 | — | — |

The data shows a clear pattern: the average FPS at 1080p is above 40 even for the highest presets, while at 4K the average falls to 17–30 FPS regardless of the setting. The minimum-only frames for Medium at 4K (19) and for Medium at 1440p (33) and 1080p (48) confirm that the system's worst-case drops are also low at high pixel counts. At 1080p Medium, the minimum is still 48 FPS, showing a stable enough frame delivery; at 1440p Medium, the minimum dips to 33; and at 4K Medium, it touches 19, which would result in noticeable hitches. These measured frames are the only hard evidence to base settings selections on, and they show the boundary where playability ends: for this combo, 1080p High is marginally playable (43), 1440p is only playable on Low (51), and 4K is only close to 30 for Low. The overall ranking of the combination at the 3551st place among 3720 combos reflects just how many lower scores exist, but the measured numbers are the primary guide for user expectation.