Thief

Measurements from the Steam benchmark show that the Intel Core i5-14600K and NVIDIA GeForce RTX 4070 Ti pair produce very high frame rates in Thief—the game runs at 210 FPS average at 1080p on Low, 172 FPS on High, and still stays above 60 FPS at 3840x2160 on Ultra, averaging 66 FPS. The component that binds depends on the resolution and preset: at 1080p, the i5-14600K's single-core boost of 5.30 GHz capably feeds the GPU, while at 4K the RTX 4070 Ti becomes the clear constraint, with High preset average a 70 FPS. This analysis uses the measured data only, which confirms the pair lands in the top 7.4% of all tested configurations.

Resolution Scaling

Frame rate data across Thief's measured presets reveals how much load shifts from the Central coprocessor to the graphics card. At 1920x1080 High, the average is 172 FPS, and at 2560x1440 High it drops to 118 FPS, a reduction of 31.9%. Climbing one more step to 3840x2160 High reduces that to 70 FPS, a 59.3% fall from the 1080p High value. The scaling is roughly in line with the pixel count, but the exact percentages change when the preset changes, which is a signature of a component bound to the GPU.

Looking at Low settings, 1080p returns 210 FPS, 1440p returns 185 FPS, and 4K falls to 120 FPS. The transition from 1080p to 1440p only loses 11.9% of the frame rate, much less than the 31.4% measured range of High peace. At Low settings the GPU's shading load is lower, so the CPU's ability to deliver frames grows more noticeable; the i5-14600K can supply 210 FPS at that resolution, and the drop to 185 at 1440p is still small. Between 1440p and 4K, Low loses 48% of frames, while High loses 40.7% over the same step. This shows that as the pixel output rises, the RTX 4070 Ti's volumetric workload (its 504.2 GB/s memory bandwidth and 40.09 TFLOPS FP32 performance) becomes the primary limit for all presets.

The plot gets clearer when comparing the scaling across presets at the same step. From 1080p to 1440p, the average rate falls about 31% for High (172→118), but only 12% for Low (210→1845). For the same step, the Medium drop is 17.9% (189→155), and Ultra drops 31.1% (164→113). At 1080p, all presets produce >160 FPS, so the i5-14600K with 14 cores supporting up to 20 threads is not saturated; the GPU still has headroom. Conversely, as you jump to 4K, the distribution tightens: Low gives 120 FPS, Medium 91, High 70, Ultra 66, meaning the difference between lowest and highest preset at that resolution is just 54 frames rather than the 46 frames seen at 1080p. The narrower spacing at 4K is the classic sign that the frame-rate is pinned by the GPU, not by the CPU cores.

The practical takeaway is that the r-bound condition manages Thief nicely. At 1080p, all settings are GPU-limited, because even at High the frame rate exceeds 170 fps, which might be more effective on a high-refresh display than X. But if you want to hit 144 Hz at 1080p, you don't need to lower settings—every preset from Medium to Ultra already does so. At 1440p, High is 118 FPS, Medium 155, Low 185, so Low/Medium provide margins versus High. At 4K, system is just capable of staying above 60 FPS (Ultra 66, High 70, Medium 91, Low 120); the raw numbers indicate that the clamp of GPU makes Ultra a marginal choice.

FAQ

Q: What average frame rates does this combo yield at 1080p?

At 1920x1080, the measured average FPS are 210 on Low, 189 on Medium, 172 on High, and 164 on Ultra. The Medium run had a minimum of 161 FPS and a maximum of 218 FPS, so the frame pacing is very consistent.

Q: How much faster is 1080p than 1440p?

Using the High preset, the drop is from 172 FPS at 1080p to 118 FPS at 1440p, a reduction of 54 FPS (about 31%). On the Medium preset it goes from 189 to 155 FPS (34 FPS lost), and on the Low preset from 210 to 185 FPS (25 FPS lost). These reductions are proportionally smaller the lighter the rendering load.

Q: Can it maintain 4K Ultra 60?

Median stats say 4K Ultra shows an average of 66 FPS. That is above 60, but with a bar that is thin—the data does not include a min or max for that row. The 4K Medium row holds a minimum of 78 FPS and maximum of 105 FPS (average 91 FPS), so Medium is much more safe.

Q: What are the most extreme FPS points in the dataset?

For the Medium settings, the 1080p run reached a maximum of 218 FPS and a minimal of 161. The 4K Medium run clocked a minimum of 78 FPS and a maximum of 105 FPS. The single highest average is the 1080p Low run at 210 FPS.

Q: Does the CPU ever become the bottleneck?

Yes, at 1080p Low the frame rate of 210 FPS is only 21 less than the 1080p Low results, and the Low→Ultra gap at 1080p is just 46 FPS (210 vs 164). These small offsets indicate the i5-14600K converts enough compute for the GPU, but at 1080p Low the average of 210 FPS is close to the CPU-dependent limit of this pairing.

Q: What is the ranking of this combo compared to other tested setups?

The combination ranks 276th out of 3720 tested combinations, meaning it sits in the top 7.4% of all combos. The CPU lands in the 90th percentile of all installed CPUs, and the GPU at the 84th percentile of all cards in the database.

CPU Role

The Intel Core i5-14600K is the engine for draw calls and game logic, and Thief is a 2014 stealth game that leans on a single core rather than many. The CPU's architecture is a Raptor Lake Refresh, built at a process node of 10nm, with a base clock of 3.50 GHz and a boost of 5.30 GHz. To maintain stable performance, the chip has 14 physical cores (6 of them being the large P-cores) and 20 threads (technology's Hyper-Threading enables 2 extra threads each). Caches stack up: 80 KB L1 per core, 2 MB L2 per core, and 24 MB shared L3. These aspects are not directly measured in the game, but they influence the IPC and the boost behavior.

The benchmark results for this CPU show it is a heavy single-core performer: a Geekbench single-core score of 2064 and a Cinebench R23 single-core score of 2064 (the game only uses a handful of threads). Those numbers are moderate, but for an eleven-year-old title the increase from the 3400-speed 512-bit is more than enough. The CPU's overall average benchmark is 48618, which puts it in the 92 percentile vs all CPUs, and its nearest rival is the Intel Xeon Gold 5318H with avgScore 48698 (delta+0.2%) and the AMD EPYC 4345P with avg 48470 (delta+0.3%).

Thief's performance map at the CPU level is revealed by the frame spread: at 1080p High (172 FPS) the game is not likely CPU-capped because the difference to 1440p High (118 FPS) is a 54 FPS loss—ample GPU scaling. But at 1080p Low, the combination achieves 210 FPS, which is an indication that the GPU can go beyond the 210 average and that the CPU is not a hard bottleneck. To maintain the 210-FPS hypothesis, the i5-14600K's single-core boost of 5.30 GHz is important; a lower-clocked chip would likely show lower frame rates in this scenario.

The data for Medium at each resolution gives a sense of the CPU's influence. At 1080p Medium, the 189 FPS average catches within a min/max of 161/218. At 1440p Medium, the average 155 FPS with min 132/max 179. At 4K Medium, the average is 91 FPS but the min is 78. That minimum is the dip points that can be perceived as stutters; even at 4K Medium the i5-14600K keeps the presentation stable enough that the 70-FPS average is not shamed. The CPU's cache hierarchy of 24 MB L3 is large enough that the game's scene data for a mission can also be hoisted in the last-level cache.

Single-thread honesty: the 5.30 GHz boost clock sits against a 14-core allotment. Some people might worry that a 2014 engine on many threads; but Thief's executable is not multi-core optimized. The measured of the combination shows the i5-14600K is not a limiting factor in any resolution except possibly at the HighestFPS at 1080p since frames plateau at 210 FPS, which is the cur. In all other cases the GPU curves the result. Thus the role of the CPU is to ensure a constant upper limit elsewhere the GPU can push into 1080p and 1440p, and to not throttle events at 5.30 MHz.

How This Combo Ranks

The measured combo lands at position 276 out of 3720 total configurations in the database, placing it in the top 7.4% of all combinations. This is a high rank for a mainstream mid- to high-end socket, consistent with the matched results: the RTX 4070 Ti holds a 44795 average benchmark score in the "Sum" and sits at the 84th percentile of all GPUs; the i5-14600K has an average score of 48618 and is at the 90th percentile of CPUs.

Looking at the paired benchmarks of the graphics card vicinity at parity: the RTX 4070 Ti's closest rivals in the GPU table are the NVIDIA GeForce RTX 5090 Mobile (avg 45152, 0.8% faster), the AMD Radeon Pro 5500 XT (avg 45384, 1.3% faster), the NVIDIA RTX A6000 (avg 44075, this card is 1.6% faster than that), and the Intel Arc A730M (avg 45592, 1.7% faster). Its average of 44795 is lower than the RTX 5090 Mobile by 0.8%, and lower than the A6000 by 1.6% the other way—so the GPU is placed in the same tier as those, but the 90th percentile for such a close grouping.

The CPU's rivals in the table: the Intel Xeon Gold 5318H has 48698 avg (delta-0.2%), the AMD EPYC 4345P 48470 (delta+0.3), the Intel Core Ultra 5 245HX 48287 (delta+0.7), and the Core