Thief

The Intel Core i5-14600KF paired with the NVIDIA GeForce RTX 3080 delivers a strong performance profile in Thief, a game released on 2014-02-24. The measured data shows that this combination is capable of high frame rates across all tested settings and resolutions, with the primary bottlenecks appearing at 4K Ultra settings. The recorded scores indicate that the system holds up well against the demands of the stealth title, though the balance between CPU and GPU shifts noticeably as resolution increases.

Settings Recommendations

The measured FPS data indicates that the Medium preset offers the most balanced experience for this hardware combination. At 1920x1080, Medium produces an average of 185 FPS with a minimum of 157 FPS and a maximum of 213 FPS. This provides a substantial frame rate cushion above the High preset’s 141 FPS average, while maintaining visual quality improvements over the Low preset’s 209 FPS average. The Medium preset’s minimum FPS of 157 ensures smooth gameplay even during demanding scenes, which is a critical factor for a stealth game where consistent frame pacing matters for precise movement and observation.

At 2560x1440, the Medium preset again proves optimal, delivering 127 FPS average with a minimum of 108 FPS and a maximum of 146 FPS. This is nearly double the 4K High result and keeps the experience fluid. The High preset at this resolution drops to 97 FPS, which is still playable, but the Medium preset’s higher minimum frame rate suggests better stability. The Ultra preset at 1440p yields 92 FPS, which falls below the 100 FPS threshold that many users consider a target for high-refresh-rate displays, making Medium the safer choice for consistent performance.

At 3840x2160, the Medium preset produces 75 FPS average with a minimum of 63 FPS and a maximum of 86 FPS. While the High preset at 4K achieves 57 FPS and Ultra drops to 54 FPS, the Medium preset remains above the 60 FPS mark on average. The recorded minimum of 63 FPS indicates that even the lowest frames stay above the 60 FPS line, which is notable for 4K gaming. The Low preset at 4K reaches 98 FPS, but the visual compromise is significant for a game that relies on environmental detail for stealth mechanics.

The data suggests that Ultra settings provide diminishing returns in this title. At 1080p, Ultra averages 134 FPS versus Medium’s 185 FPS, a 51 FPS difference. At 1440p, Ultra averages 92 FPS versus Medium’s 127 FPS, a 35 FPS difference. At 4K, Ultra averages 54 FPS versus Medium’s 75 FPS, a 21 FPS difference. The visual uplift from Medium to Ultra is not quantified in the data, but the frame rate cost is substantial, especially at higher resolutions where the GPU is already under load.

GPU Role

The NVIDIA GeForce RTX 3080 uses the GA102 chip with a 10 GB GDDR6X memory buffer on a 320-bit bus. The memory operates at 1188 MHz with 19 Gbps effective speed, producing a bandwidth of 760.3 GB/s. The GPU has a base clock of 1440 MHz and a boost clock of 1710 MHz. These specifications align with the measured frame rates, particularly at 4K where memory bandwidth and capacity become more critical.

The RTX 3080’s 10 GB VRAM is sufficient for Thief at the tested settings. The game’s release date of 2014-02-24 predates the GPU by several years, meaning the title’s texture and geometry demands are well within the card’s capacity. The measured data does not show any signs of VRAM-related stutters or performance cliffs, as the minimum FPS values at Medium settings across all resolutions remain proportional to their averages. The 4K Medium result with a 63 FPS minimum versus a 75 FPS average suggests stable memory management.

The GPU’s boost clock of 1710 MHz, combined with its 8704 shading units, delivers the raw throughput needed for the higher resolution results. The 29.77 TFLOPS FP32 performance is evident in the 4K Low result of 98 FPS, which shows that the card can push frame rates well above 60 FPS even at 3840x2160 when settings are reduced. The pixel rate of 164.2 GPixel/s and texture rate of 465.1 GTexel/s support the measured fill-rate demands of the game’s environments.

The RTX 3080’s percentile rank of 68 among all GPUs in the database, with an average benchmark score of 23172, places it as a mid-to-high tier card. Its nearest rival, the NVIDIA P106-100, scores 23249 with a delta of -0.3 percent, indicating that the RTX 3080 is essentially at parity with that card in synthetic benchmarks. The AMD Radeon Pro Vega 16 and AMD Radeon RX 6600M also fall within 0.4 percent of the RTX 3080’s score, showing that the card occupies a specific performance band that is competitive with those alternatives.

Resolution Scaling

The frame rate scaling from 1920x1080 to 3840x2160 follows the expected pattern of increasing GPU load. At Medium settings, the average FPS drops from 185 at 1080p to 127 at 1440p, a 31 percent reduction, and then to 75 at 4K, a further 41 percent reduction from the 1440p result. This steep decline indicates that the RTX 3080 becomes the limiting component as resolution increases, with the CPU’s role diminishing.

At Low settings, the scaling is more pronounced: 209 FPS at 1080p, 167 FPS at 1440p, and 98 FPS at 4K. The 4K Low result is less than half of the 1080p Low result, which shows that even with minimal GPU work, the resolution increase places significant demands on the rendering pipeline. The High settings show a similar pattern: 141 FPS at 1080p, 97 FPS at 1440p, and 57 FPS at 4K. The 4K High result falls below the 60 FPS threshold, confirming that this resolution and settings combination is the first to break the playability barrier.

The Ultra settings produce the lowest scaling curve: 134 FPS at 1080p, 92 FPS at 1440p, and 54 FPS at 4K. The 4K Ultra result is the lowest recorded measurement in the dataset, indicating that the combination of maximum settings and maximum resolution pushes the GPU to its limits. The difference between 1080p and 4K at Ultra is 80 FPS, which represents a 60 percent drop, while the same comparison at Medium is a 110 FPS drop, or 59 percent. This proportional scaling suggests that the GPU is consistently the bottleneck across all settings levels, with no sudden CPU-related ceiling appearing at lower resolutions.

Measured FPS Breakdown

At 1920x1080, the Low preset delivers the highest average at 209 FPS. The Medium preset follows at 185 FPS with a minimum of 157 FPS and a maximum of 213 FPS. The High preset averages 141 FPS, and the Ultra preset averages 134 FPS. The spread from Low to Ultra is 75 FPS, which is the largest absolute gap among the three resolutions, indicating that the CPU and GPU combination has headroom to spare at 1080p.

At 2560x1440, the Low preset averages 167 FPS. The Medium preset averages 127 FPS with a minimum of 108 FPS and a maximum of 146 FPS. The High preset averages 97 FPS, and the Ultra preset averages 92 FPS. The spread from Low to Ultra is 75 FPS, matching the 1080p spread exactly. This suggests that the performance scaling between settings is consistent regardless of resolution, reinforcing the GPU-bound nature of the workload.

At 3840x2160, the Low preset averages 98 FPS. The Medium preset averages 75 FPS with a minimum of 63 FPS and a maximum of 86 FPS. The High preset averages 57 FPS, and the Ultra preset averages 54 FPS. The spread from Low to Ultra is 44 FPS, which is narrower than at lower resolutions. This indicates that at 4K, the GPU is so heavily loaded that even reducing settings from Medium to Low only recovers 23 FPS, whereas at 1080p the same settings change recovers 24 FPS.

The Medium preset’s recorded minimum and maximum values provide insight into frame pacing. At 1080p, the minimum of 157 FPS versus an average of 185 FPS represents a 15 percent variance, while the maximum of 213 FPS is 15 percent above the average. At 1440p, the minimum of 108 FPS is 15 percent below the 127 FPS average, and the maximum of 146 FPS is 15 percent above. At 4K, the minimum of 63 FPS is 16 percent below the 75 FPS average, and the maximum of 86 FPS is 15 percent above. This consistent variance across resolutions indicates stable performance without unexpected spikes or drops.

How This Combo Ranks

The Intel Core i5-14600KF and NVIDIA GeForce RTX 3080 combination ranks 756 out of 3720 tested combos in the database for Thief. This places the combo in the top 20 percent of all tested systems, as 756 divided by 3720 equals 20.3 percent. The rank reflects the strong absolute performance of the CPU and GPU, though it does not account for the specific settings or resolutions used in the ranking.

The CPU’s average benchmark score of 49394 places it in the 90th percentile among all CPUs. Its nearest rival, the AMD Ryzen 9 7900, scores 49228 with a delta of 0.3 percent, meaning the Intel chip is 0.3 percent faster. The AMD Ryzen 7 PRO 5755G scores 49196 with a delta of 0.4 percent, and the Intel Core Ultra 5 245 scores 48995 with a delta of 0.8 percent. The AMD Ryzen AI Max+ 388 scores 49796 with a delta of -0.8 percent, indicating it is 0.8 percent faster than the i5-14600KF. These close deltas show that the CPU sits in a tightly contested performance band.

The GPU’s average benchmark score of 23172 places it in the 68th percentile among all GPUs. Its nearest rivals, including the NVIDIA P106-100, AMD Radeon Pro Vega 16, and AMD Radeon RX 6600M, all score within 0.4 percent of the RTX 3080. This proximity suggests that the GPU’s synthetic performance is comparable to those cards, though the RTX 3080’s real-world gaming results, as measured in Thief, may differ due to driver optimizations and game-specific features.

The combo rank of 756 out of 3720 is a composite measure that reflects both the CPU and GPU contributions. The CPU’s 90th percentile ranking is higher than the GPU’s 68th percentile, which suggests that the processor has more relative headroom than the graphics card. This asymmetry explains why the frame rates at 4K Ultra drop to 54 FPS, as the GPU becomes the limiting factor while the CPU still has capacity to spare.

CPU Role

The Intel Core i5-14600KF features 14 cores and 20 threads, with a base clock of 3.50 GHz and a boost clock of 5.30 GHz. The processor uses the Raptor Lake architecture on a 10 nm process node from Intel. It has 80 KB of L1 cache per core, 2 MB of L2 cache per core, and 24 MB of shared L3 cache. The CPU supports DDR4 and DDR5 memory in a dual-channel configuration.

The CPU’s multi-core performance is substantial, as shown by its Cinebench R23 multi-core score of 32544 and single-core score of 4594. The Geekbench multi-core score of 17085 and single-core score of 2445 further confirm the processor’s capabilities. The PassMark multi-thread score of 38697 and single-thread score of 4273 indicate strong performance in both threaded and single-threaded workloads.

In Thief, the CPU’s role is most prominent at lower resolutions. At 1080p Low, the combo achieves 209 FPS, which is a high frame rate that requires strong single-thread performance to maintain. The boost clock of 5.30 GHz provides the necessary frequency for the game’s simulation and AI logic. At 1080p Ultra, the frame rate drops to 134 FPS, but this is still well above the 60 FPS threshold, indicating that the CPU is not a limiting factor even at maximum settings.

At 1440p and 4K, the CPU’s influence diminishes as the GPU takes over the rendering load. The frame rate differences between 1440p and 4K at the same settings are larger than the differences between 1080p and 1440p, which confirms that the GPU is the primary bottleneck at higher resolutions. The CPU’s 14 cores and 20 threads are more than sufficient for Thief’s threading model, which was designed for the hardware available at the game’s 2014 release.

The CPU’s TDP of 125 W and its socket compatibility with Intel Socket 1700 indicate that it requires a capable cooling solution, though the data does not specify thermal performance in this title. The processor’s 90th percentile ranking among all CPUs, with an average benchmark score of 49394, places it ahead of 90 percent of tested processors. This high ranking suggests that the i5-14600KF has ample computational headroom for the game, and the measured frame rates are consistent with a system that is not CPU-limited in most scenarios.