Thief

The measured pairing of Intel Core i5-10400F and AMD Radeon RX 6600 XT in Thief produces 1080p performance that ranges from 85 to 125 FPS depending on quality preset, with 1440p still holding above 59 FPS in every tested setting, but 4K collapses to a 36–63 FPS band that clearly exposes the AMD card as the limiting component. The configuration ranks 2812 out of 3720 tested combos in this game, placing it in the bottom quarter of the measured set — a surprising result given the CPU and GPU sit at the 68th and 70th percentile against all tested parts respectively. Bearing in mind the game’s 2014 release, these data points indicate a system that is more than adequate for standard resolution play but will need lower settings or a GPU upgrade for 4K consistency.

Resolution Scaling

Resolution scaling follows a textbook image-quality versus frame-rate curve, but the amplitude varies starkly by preset. At 1080p Low the combination produces 125 FPS; the RX 6600 XT drops to 117 FPS at 1440p (93.7% of 1080p) and then to 63 FPS at 4K, which is only 50.4% of the frame count seen at 1080p. For the Higher visual settings, the gap is even more dramatic: High at 1080p shows 90 FPS, 1440p shows 62 FPS (68.9% of 1080p), and 4K High rests at 36 FPS, just 40% of the 1080p value. A similar 1080p High to 1440p High step of 28.1% suggests that both the ROP and memory bandwidth are being taxed, while the 1440p to 4K step of another 26 FPS confirms that the GPU’s pixel fill rate is the limiting factor at the top resolution.

The fact that Low at 4K still yields 63 FPS shows the memory interface is not a total barrier—63 FPS is generally playable. Yet at Med settings 4K delivers 48 FPS with a min of 40 and a max of 55, which shows the frame-time spread is wider than at lower resolutions. The 4K average results for High and Ultra are equal (36 vs 36), which argues that after a certain geometric and texture volume is reached, the frame rate ceases to degrade further because all resources are saturated. On the other end, the 1080p Ultra reading of 85 FPS as opposed to Medium’s 113 FPS and Low’s 125 FPS exhibits a surprisingly linear rating-to-quality cost relationship, suggesting the game IoT is not the limiting but rather the pixel workload.

GPU Role

The AMD Radeon RX 6600 XT is built on the RDNA 2.0 architecture with 2,048 shading units, 128 texture mapping units, 64 ROPs, and a 8 GB GDDR6 memory on a 128-bit interface, providing a bandwidth of 256.0 GB/s. The memory clock sits at 2,000 MHz (16 Gbps effective), while the base clock is 1,358 MHz and the boost clock reaches 2,589 MHz. A card of this chip sits at the 70th percentile of all GPUs, with an average benchmark score of around 24,442. It is paired with an i5-10400F that itself rests in the 68th percentile for CPUs.

Given these stats, the GPU is well able to supply. Yet Thief runs on an older engine: at 1080p the data shows no obvious CPU constraint, so the system’s frame rate is largely a GPU function. The 8 GB of VRAM proves sufficient even for 4K settings, as neither the 4K High nor 4K Ultra results degrade excessively beyond frame-rate drops—the data shows no such VRAM-related stutter. The 128-bit bus delivers 63 FPS on 4K Low, proving that at low quality bandwidth doesn’t fully bottleneck. However, the upward move from medium to High cost 15 FPS in 4K, and Ultra costs another 12 frames from Medium, a rate step that the 256 GB/s bus alone cannot explain fully.

The RX 6600 XT ceases at 10.60 TFLOPS FP32, which appears moderately sufficient for this 2014 title at 1440p but becomes the main constraint at 4K. The GPU’s launch MSRP of 379 USD puts a class context, and within this benchmark the combination of 2,048 shading units and 64 ROPs is precisely the balanced approach—At 1080p the GPU keeps the CPU fed with over 90 FPS in most presets, but at 4K the ROP/“shader” equilibrium shifts and the card trails beyond 60 FPS at all quality expectations only on Low.

Combined Performance

The combo is ranked 2812 out of 3720 configurations in Thief, a position that places it in the bottom percentile of measured frames. The 2,812th rank signifies that 75.6% of the test pool delivered higher frame rates, or that this system is slower than more than three-quarters of all in-game, core-based setups. This is a stark contrast to the 70th percentile report of the GPU alone, suggesting that when combined with the 6-core Core i5-10400F, the platform is still relatively slow.

Specifically, the CPU has the capability, as its GPUs percentile is 68, and its Cinebench R23 multi-core score is 10,283 with a single-core of 1,451. With passmark single thread at 2,541 and multicore at 12,115, the CPU cannot be considered a weak link for a DirectX 11-era game. Rather, the low combo rank likely stems from the large number of more capable flagship slots (including combinations that may include higher-clocked processors and faster gpus). This system, when paired with an RX 6600 XT, produces 85 to 125 FPS at 1080p; therefore the rank is a relative indicator of the whole number of configs rather than an absolute stress check.

FAQ

Q: Can this configuration run Thief at 4K comfortably from the measured data?

A: All four 4K presets are below 60 FPS: Low 65, Medium 48 (with min 40), High 36, and Ultra 36. Only the Low setting provides a 63 FPS average, meaning 4K requires a significant visual quality compromise or a GPU upgrade.

Q: What is the best 1440p choice based on the FPS numbers?

A: At 1440p, Low yields 107 FPS, Medium 81 FPS, High 62 FPS, and Ultra 59 FPS. For a high-refresh monitor the Medium preset is safe with a 81 FPS average and a 69 min, while Ultra sits barely below 60 average, so users must choose between frame rate stability and one extra quality drop.

Q: How does the CPU influence this particular game?

A: The Intel Core i5-10400F contains 6 cores and 12 threads, with a base of 2.90, Boost 4.30 GHz, 12 MB of shared L3, and memory bandwidth of 42.7 GB/s. Since Thief appears sensitive to pixels rather than physics, the CPU is rarely the bottleneck; even at 1080 Keys (125 FPS Low) the six threads are above or above the demand, so the limiting role is the mind.

Q: What are the GPU’s peak resource numbers and why do they matter here?

A: The forger is a 7 nm chip with 8 GB GDDR6, 128-bit bus 256.0 GB/s bandwidth, and a boost clock of 2,589 MHz. These provide the unexpected 4K frame growth, because while 8 GB is enough to hold Trail Level textures, the 128-bit facia is not high for 4K—tally scaling from 1080 to 4K causes a huge 2.5x drop on high.

Q: How does this combo compare to the 1080p results in other testing?

A: In the benchmark ratios, the GPU itself sits at the 70th percentile versus all GPUs, and the CPU at the 68th percentile. In the actual game, the combo ranks 2nd out of 3,720, which means 75% of measured combos exceed the performance, so the specific combination is not near the top, understandably because the 10400F is by many measures an entry-level 14nm chip.

Measured FPS Breakdown

The following table lists The exact measured average FPS (and min and max when recorded) for all tested resolution-preset combinations.

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

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

| 1920x1080 | High | 90 | — | — |

| 1920x1080 | Low | 125 | — | — |

| 1920x1080 | Medium | 113 | 96 | 130 |

| 1920x1080 | Ultra | 85 | — | — |

| 2560x1440 | High | 62 | — | — |

| 2560x1440 | Low | 107 | — | — |

| 2560x1440 | Medium | 81 | 69 | 93 |

| 2560x1440 | Ultra | 59 | — | — |

| 3840x2160 | High | 36 | — | — |

| 3840x2160 | Low | 63 | — | — |

| 3840x2160 | Medium | 48 | 40 | 55 |

| 3840x2160 | Ultra | 36 | — | — |

Observe that at 1080p, all presets stay above 85 and the minimum frametime for Medium is 96 FPS, showing a largely stutter-free frame pacing. At 1440p, only High (62 FPS) and Ultra (59 FPS) fall just below the common refresh threshold, but the median (Medium) still manages an average of 81 with the min at 69—that is perfectly playable on 60Hz displays. 4K is break-even: both Low 63 FFS and Medium with its 40 FPS min show signs of the GPU, while the High and Ultra both lock to 36 FPS, meaning that V-sync or frame rate variables inside the management limit certain drops.

At each resolution, the trend is clearly GPU-bound. A 125 na fps low at 1080p is dominated by the fill rate; when resolution doubles, frame output halves to threequarters for a high enough pixel-count reduction. Meanwhile the CPU’s activity is uniform across resolution, as visible from the 1080p filters massively exceeding 60 and from no decline in minimal at Lower the setting where the CPU would be more likely bottlenecked—the rules still reach 107.

CPU Role

Despite the 600 series from 2020 Intel campaign, the i5-10400F is a Comet Lake offspring: 6 cores, 12 threads, running with base cluster 2.90 GHz and 4.30 GHz boost on a 65 W TDP. Its cache topology is modest—12 MB total shared L3 with 256 KB L2 per core—but its measured capabilities are far from fragmented. In the baseline database it scores 4748 in 3DMark 16-thread portion, 4,318 in Cinebench R20 multi, 6,257 in Geekbench multicore, and a Passmark multithread 12,109. These values correspond to a 68th-percentile status and an average benchmark a just 14,185.

Given Thief’s age, the game includes few CPUs that can scale beyond 8 threads. i5’s 6 cores / 12 threads therefore provide sufficient margin for CPU side. At 1080p low is 125 fps and medium 99–113, crash shows no sign of CPU limitation—the frame rate operating or not CPU core is either padding at red. The role of the CPU is more subtle at higher resolution, where the GPU becomes bound; therefore the same CPU yields nearly half the frame rate with unchanged geometry load. The CPU’s memory bandwidth (42.7 GB/s) and PCIe Gen 3 lane count also do not hinder the RX 6600 XT given PCIe Gen 3 x16 bus with the card running directly in x8 mode.

The absence of sustained internal measurement or lower minimums across resolutions confirm the shoulder: the CPU is stable, and the bottleneck is the GPU ROP/ memory interface. A 6-core processor is sufficient to feed this GPU at 1080 ST and the FPS spread of 40–55 in 4K Medium shows that the CPU does not clip the frame pacing at any actual meter.

When framing the system as the score, i play, the net (14.72) and the cpu (combined 16.1) combination represents an average target for a 4K challenging setting. The 2812 rank of 2 out of 3,720 is therefore not a reflection of a weak CPU — that metric is computer-driven and made, but as a benchmark database should show that the i5-10400F + RX 6600 XT must be placed mostly for span 1440p and dynamic gaming, not as a 4 thousand-heavy system.