Thief

# Thief Performance Analysis: Intel Core i5-12400F + NVIDIA GeForce RTX 5090 D

The combination of an Intel Core i5-12400F and NVIDIA GeForce RTX 5090 D produces a compelling but notably GPU-bound experience in Thief, a 2014 stealth title that scales predictably across resolution and settings presets. Measured frame rates show this pairing delivers triple-digit averages at every tested configuration, with the RTX 5090 D's immense compute resources keeping the aging game engine comfortably above 85 FPS even at 4K Ultra. The data reveals a system where the GPU dominates the performance envelope, while the six-core Alder Lake processor provides sufficient headroom to avoid becoming the primary constraint. The following sections break down the measured results, interpret what they mean for the limiting component, and offer guidance on optimizing the experience.

Resolution Scaling

The FPS progression from 1080p to 4K clearly identifies the GPU as the performance ceiling in this configuration. At High settings, the average frame rate drops from 140 FPS at 1920x1080 to 124 FPS at 2560x1440, then further to 92 FPS at 3840x2160. This represents a 34.3% reduction from 1080p to 4K, which is substantial but not catastrophic — the RTX 5090 D maintains playable framerates throughout. The scaling pattern suggests that pixel throughput is the dominant factor, with each resolution step adding significant rasterization work. From 1080p to 1440p, the drop is 16 FPS (11.4%), while the 1440p to 4K transition costs 32 FPS (25.8%), showing that the pixel-count increase becomes proportionally more expensive at higher resolutions.

Low settings show a similar trajectory: 171 FPS at 1080p, 151 FPS at 1440p, and 127 FPS at 4K. The absolute drop from 1080p to 4K is 44 FPS (25.7%), which is actually smaller in percentage terms than the High preset's decline. This indicates that when the GPU is less burdened by shading work, the resolution scaling becomes more linear and less punishing. Medium settings follow the same pattern with 155 FPS at 1080p, 137 FPS at 1440p, and 114 FPS at 4K. Ultra settings deliver 138 FPS at 1080p, 122 FPS at 1440p, and 87 FPS at 4K — the largest relative drop of 37% from 1080p to 4K, confirming that the highest preset amplifies the GPU's load disproportionately.

The min/max frame data for Medium settings provides additional insight. At 4K Medium, the range spans 97 to 132 FPS, a 35 FPS spread that indicates some scene-dependent variability. At 1440p Medium, the range narrows to 116-157 FPS (41 FPS spread), while 1080p Medium shows 132-178 FPS (46 FPS spread). Interestingly, the spread widens as resolution decreases, suggesting that at lower resolutions the CPU or other system components introduce more frame-time variance, while at 4K the GPU becomes so saturated that frame pacing stabilizes somewhat. This behavior reinforces the interpretation that the RTX 5090 D is the limiting factor across all tested resolutions — the component that determines overall throughput, with the CPU stepping in only at the lowest resolution to add variability.

GPU Role

The NVIDIA GeForce RTX 5090 D is the workhorse of this configuration, and its specifications align with the measured results. The GPU features 32 GB of GDDR7 memory on a 512-bit bus, delivering 1.79 TB/s of bandwidth. This massive memory subsystem is far beyond what Thief requires — a 2014 title with modest texture budgets — but it ensures that memory capacity or bandwidth never becomes a bottleneck. The 92,200 million transistor chip on a 5 nm TSMC process has a base clock of 2017 MHz and a boost clock of 2407 MHz, with 21,760 shading units driving the rasterization work. The GPU's 104.8 TFLOPS of FP32 compute is more than sufficient to saturate the game's rendering pipeline, which is why the framerates remain high even at 4K Ultra where the GPU is clearly the constraining component.

The RTX 5090 D's 92nd percentile ranking among all GPUs, with an average benchmark score of 77,712, places it in the upper echelon of graphics hardware. Its nearest rivals in the benchmark database — the AMD Radeon RX 6650M XT (76,904, 1.1% higher), AMD Radeon RX 6850M XT (78,940, 1.6% lower), and NVIDIA Tesla P100 variants (79,396 and 79,605, 2.1% and 2.4% lower respectively) — all sit within a narrow performance band, but the RTX 5090 D's gaming-focused architecture and driver optimizations likely contribute to its strong showing in Thief's actual frame rates. The GPU's 680 texture mapping units and 176 render output units provide ample texturing and pixel-pushing capability, which directly translates to the high fill rates observed in the benchmark results.

Memory clocks of 1750 MHz (28 Gbps effective) and the 1.79 TB/s bandwidth are critical for maintaining consistent frame delivery at 4K, where large texture fetches and high-resolution frame buffers demand substantial memory throughput. Even at Ultra settings, the 32 GB VRAM capacity is never taxed, which means the GPU can focus its full compute power on rendering rather than managing memory pressure. The data shows the GPU is the clear performance driver: at every resolution, the difference between Low and Ultra settings is 43-44 FPS (171 to 138 at 1080p, 151 to 122 at 1440p, and 127 to 87 at 4K), a consistent ~25-31% performance cost for the higher preset that reflects the GPU's workload scaling rather than any CPU limitation.

Settings Recommendations

The measured data points to Medium settings as the optimal balance for this configuration, particularly at higher resolutions. At 4K Medium, the average of 114 FPS with a minimum of 97 FPS provides a smooth experience that stays well above the 90 FPS threshold, while High and Ultra drop to 92 FPS and 87 FPS respectively — the latter dipping into territory where frame pacing could become noticeable on high-refresh displays. The Medium preset's min FPS of 97 at 4K is the strongest minimum among all 4K configurations, suggesting better frame-time consistency than the higher presets which lack min/max data but average lower.

At 1440p, Medium delivers 137 FPS average with a 116 FPS minimum, which is only 7 FPS below the High preset's average of 124 FPS? Wait — that's inverted. Medium is actually 13 FPS higher than High at 1440p, which is an anomaly worth noting. The data shows Medium at 137 FPS average versus High at 124 FPS, and Ultra at 122 FPS. This suggests that the High preset may include some inefficiency or that Medium's settings are better optimized for the RTX 5090 D's architecture. Similarly at 1080p, Medium achieves 155 FPS versus High's 140 FPS and Ultra's 138 FPS. The Medium preset consistently outperforms High by a margin of 11-15 FPS across all resolutions, making it the clear choice for maximum performance without sacrificing visual quality significantly.

For users prioritizing visual fidelity, Ultra at 1440p delivers 122 FPS, which is still highly playable and only 15 FPS below Medium's output. At 1080p, Ultra's 138 FPS is indistinguishable from Medium in practical terms. However, at 4K, Ultra's 87 FPS average is the only measured configuration that falls below 90 FPS, and users targeting 4K would be better served by Medium or High. The recommendation is to use Medium at 4K for the best combination of frame rate and stability, or High at 1440p and below if visual quality is paramount. Low settings provide diminishing returns — the 4K Low average of 127 FPS is only 13 FPS higher than Medium, and the visual compromise is significant for a relatively small gain.

FAQ

Q: Is this configuration capable of 4K gaming in Thief?

A: Yes, the data shows playable 4K performance across all settings. The lowest average is 87 FPS at 4K Ultra, while Medium delivers 114 FPS and Low reaches 127 FPS. All configurations exceed 60 FPS at 4K.

Q: Why does Medium outperform High in every resolution?

A: The measured averages show Medium at 155 FPS (1080p), 137 FPS (1440p), and 114 FPS (4K), compared to High's 140 FPS, 124 FPS, and 92 FPS. This 11-22 FPS advantage suggests the High preset may enable less efficient rendering paths or features that disproportionately burden the GPU.

Q: What is the minimum frame rate I can expect at 1440p Medium?

A: The measured minimum is 116 FPS, with a maximum of 157 FPS and an average of 137 FPS. This indicates a smooth experience with limited frame-time variance.

Q: How does the RTX 5090 D compare to its nearest benchmark rivals?

A: The GPU scores 77,712 on average, placing it 1.1% below the AMD Radeon RX 6650M XT (76,904 is actually higher? Wait — the delta shows RX 6650M XT at 76,904 with +1.1% deltaPct, meaning the RTX 5090 D is 1.1% lower). The RX 6850M XT scores 78,940 (1.6% higher), while the Tesla P100 variants score 79,396 and 79,605 (2.1% and 2.4% higher).

Q: Is the i5-12400F bottlenecking the RTX 5090 D?

A: The data does not explicitly measure CPU utilization, but the resolution scaling pattern — where FPS drops consistently as resolution increases — indicates the GPU is the primary constraint. At 1080p Low, the 171 FPS average suggests the CPU can feed the GPU adequately even at high frame rates.

Q: What is the best settings preset for 4K play?

A: Medium is recommended, delivering 114 FPS average with a 97 FPS minimum. High averages 92 FPS, and Ultra drops to 87 FPS, making Medium the only preset that maintains an average above 100 FPS at 4K.

How This Combo Ranks

This specific combination of Intel Core i5-12400F and NVIDIA GeForce RTX 5090 D ranks 311th out of 2,851 tested combos in Thief, placing it in the top 10.9% of all configurations. The rank reflects the system's strong overall performance — the RTX 5090 D's high-end GPU capabilities are tempered by the mid-range i5-12400F, but the game's age means the CPU is rarely a limiting factor. The CPU itself ranks in the 73rd percentile among all processors with an average benchmark score of 19,039, while the GPU sits in the 92nd percentile with a score of 77,712. The disparity between CPU and GPU percentiles suggests the combo's ranking is driven primarily by the graphics card's dominance.

Within the context of the 2,851 tested combos, a rank of 311 indicates that this pairing outperforms the vast majority of systems tested in Thief. The game's relatively low system requirements compared to modern titles mean that even mid-range CPUs like the i5-12400F can achieve high frame rates when paired with a flagship GPU. The rank also benefits from the RTX 5090 D's massive memory bandwidth and compute resources, which excel in a game that does not heavily tax modern multi-core processors. The combination's position in the top 11% of all tested configurations underscores its capability for high-refresh-rate gaming at 1440p and solid 4K performance.

CPU Role

The Intel Core i5-12400F provides the processing foundation for this configuration, and while the GPU dominates the performance picture, the CPU's characteristics still influence results. This Alder Lake-S architecture processor features 6 cores and 12 threads, with a base clock of 2.50 GHz and a boost clock of 4.40 GHz. Its 18 MB of shared L3 cache and 1.25 MB L2 per core provide adequate data locality for a 2014 game with modest thread demands. The CPU's 3DMark scores — 5,912 max threads and 909 single thread — indicate balanced multi-core and single-thread performance, which is sufficient for Thief's engine that likely uses only a few threads effectively.

The CPU's benchmark performance shows it sits at 73rd percentile among all processors, with an average score of 19,039. Its nearest rivals include the AMD Ryzen 5 7535HS (19,047, 0% delta), Intel Core 3 201E (19,056, -0.1%), and Intel Core i5-1335U (18,982, 0.3%). These competitors all score within 0.3% of the i5-12400F, indicating that the CPU is positioned in a competitive mid-range segment. The single-thread score of 3,481 in PassMark and 909 in 3DMark single-thread suggest that Thief's primary game thread — which typically runs on one core — has enough headroom to avoid bottlenecking the GPU.

The evidence for CPU non-bottlenecking comes from the resolution scaling. At 1080p Low, where the GPU has the least work, the system achieves 171 FPS. If the CPU were a significant constraint, this frame rate would plateau regardless of resolution. However, the data shows a consistent drop across resolutions — 171 to 151 to 127 FPS from 1080p to 4K on Low settings — which indicates the GPU is the scaling factor. The CPU's 12 threads handle the game's background tasks and physics processing without issue, allowing the RTX 5090 D to stretch its legs. The 65W TDP and 10nm process node keep power and thermals manageable, though these figures do not directly appear in the game performance metrics.

Measured FPS Breakdown

The complete dataset of measured frame rates provides a comprehensive view of this configuration's performance in Thief:

1920x1080 (1080p):

• Low: 171 FPS average
• Medium: 155 FPS average, 132 FPS minimum, 178 FPS maximum
• High: 140 FPS average
• Ultra: 138 FPS average

2560x1440 (1440p):

• Low: 151 FPS average
• Medium: 137 FPS average, 116 FPS minimum, 157 FPS maximum
• High: 124 FPS average
• Ultra: 122 FPS average

3840x2160 (4K):

• Low: 127 FPS average
• Medium: 114 FPS average, 97 FPS minimum, 132 FPS maximum
• High: 92 FPS average
• Ultra: 87 FPS average

The pattern across all resolutions shows Low > Medium > High ≈ Ultra, with Medium consistently outperforming High by a significant margin. At 1080p, the spread from Low to Ultra is 33 FPS (171 to 138), representing a 19.3% performance cost for the highest preset. At 1440p, the spread narrows slightly to 29 FPS (151 to 122, 19.2% cost), while at 4K it widens to 40 FPS (127 to 87, 31.5% cost). This indicates that the Ultra preset's additional rendering work becomes more expensive at higher resolutions, as the GPU is already saturated with pixel processing.

The min/max data for Medium settings shows the frame time consistency worsens as resolution decreases: the 4K range is 97-132 FPS (35 FPS spread), 1440p is 116-157 FPS (41 FPS spread), and 1080p is 132-178 FPS (46 FPS spread). This counterintuitive pattern suggests that at lower resolutions, where the GPU finishes frames faster, the CPU's frame submission cadence introduces more variability. At 4K, the GPU's longer frame times smooth out these fluctuations, resulting in a narrower range relative to the average. The absence of min/max data for other presets limits further analysis, but the Medium preset's comprehensive data provides the most detailed view of this system's behavior across resolutions.