Thief

The data set assesses the Intel Core i5-10400F and NVIDIA GeForce RTX 4070 SUPER pairing in the 2014 stealth title Thief. The measured frame rates span three resolutions—1920x1080, 2560x1440, and 3840x2160—and four graphics presets: Low, Medium, High, and Ultra. The results show this combination never falls below 63 frames per second (fps) in any tested configuration, and peaks at 125 fps at 1080p Low, confirming that the setup comfortably manages the game's aged engine while leaving headroom for demanding settings or high refresh rate displays.

Resolution Scaling

The most direct way to understand the behavior of this combo is to observe how frame rate changes when the resolution is raised from 1080p to 4K, while presets are kept constant. The measured data reveals a consistent, yet moderate, scaling pattern. At Low settings, the average FPS goes from 125 at 1080p to 110 at 1440p (a 12.0% drop) and then to 92 at 4K (a 26.4% drop from 1080p). At **Medium, the figures are 113, 99, and 83 respectively, a drop of 12.4% to 1440p and 26.5% to 4K. When the preset rises to **High, the averages become 102, 90, and 66—a 11.8% loss at 1440p and a 35.3% loss at 4K. Finally, on *Ultra the pattern is 100, 89, and 63, with a 11.0% decrease at 1440p and a 37.0% decrease at 4K.

Interesting insight: the percentage loss from 1080p to 4K is substantially larger for the higher presets (Ultra vs Low). In Low, the 4K frame rate is still 73.6% of the 1080p value, but for Ultra the ratio drops to 63.0%. This is a classic sign that the GPU becomes the limiting component as the workload multiplies. At Low, the CPU and other system components are much less pressured, so the scaling from 4K to 1080p is less pronounced; there simply isn't enough graphical work to keep the GPU busy. At Ultra, the GPU is forced to process more shaders, texture fetches, and pixels, and a the lower resolution gives the GPU more room to stretch, thus the scaling curve is stronger. Because the overall drops are in a moderate window (26–37%), the configuration does not exhibit a hard GPU wall until 4K Ultra. The recorded 1440p values stay extremely close to their 1080p counterparts (within 9–10 fps across all presets), which points to the CPU keeping pace at those resolutions; the i5-10400F can feed the graphics card sufficiently at 1440p without creating a serious frame bottleneck.

FAQ

Q: What is the average FPS at 2560×1440 with the High preset?

A: The average frame rate at 1440p High is 90 fps, with the minimum not recorded (presumably the v-sync floor), and a peak not recorded. This is a smooth experience for 60 Hz displays, and for 120 Hz–144 Hz monitors with adaptive sync the 90 fps average provides a comfortable baseline.

Q: Is the NVIDIA RTX 4070 SUPER capable of 4K gaming in Thief at Ultra settings?

A: Yes. At 3840×2160, the Ultra preset produces an average 63 fps. That is borderline for full responsive 60 Hz, but it is playable. The frame rate does dip, however, the data does not show a minimum value for Ultra, so we cannot assess the low frame meaning. The Ultra setting is the most demanding—since the 4K Ultra is 37% slower than 1080p Ultra—but the average remains above 60.

Q: What is the difference in average FPS between Low and Ultra at 1080p?

A: The Low preset yields an average of 125 fps, while Ultra drops to 100 fps, a 25% reduction. This indicates that Thief's Ultra settings require significantly more GPU work, but even at Ultra the 100 fps average is beyond the refresh of many common monitors.

Q: Does the combination for Thief support ray tracing or DLSS?

The game launched in 2014 and does not require the RTX 4070 Super’s ray tracing capabilities. The measured data only covers the non-ray traced graphics pipeline. The GPU's ray tracing units (56 cores) will not be exercised in this particular title.

Q: How does this CPU (Intel Core i5-10400F) compare to its closest rivals in overall benchmark score?

The CPU has an average benchmark score of 14,185. The nearest rival, the Intel Xeon 6756E, scores 14,163—0.2% lower. The Core 7 160UL is 0.3% higher, AMD EPYC 7552 is 0.5% higher, and AMD Ryzen 3 7320C is 0.6% lower. In other words, the i5-10400F is precisely at the median of these similar parts.

Q: What is the VRAM capacity on the RTX and does it matter for Thief?

The RTX 4070 SUPER carries 12 GB of GDDR6X memory across a 192-bit bus. For a game of Thief’s generation (2014), 12 GB is more than enough to avoid any VRAM limitations at all tested resolutions.

How This Combo Ranks

In the benchmark database, the pairing of the i5-10400F with the RTX 4070 SUPER in Thief occupies the 1769th spot out of a total of 3720 tested combinations. That translates to a position roughly at the 47.5th percentile—just below the median of all measured configurations. This is not a top-tier combo for the game; many systems with newer or higher-end GPUs (and the same or more capable CPUs) will outperform it. But the rank is far from the bottom: there are 3720 − 1769 = 1951 configurations that fall below it in the ordering, meaning this setup finishes in the second half of field.

The rank is sensitive to other components; for example, a CPU with a higher single-thread score or a GPU with more shading units could move the pairing significantly. The combination Rank reflects the combined measured performance, not each component separately. The CPU’s own benchmark percentile against all CPUs is 68%, while the GPU’s is 83%. Both are above average, but they are not the highest. The observed combo rank is lower than both indices would suggest—a sign that Thief may bottleneck on one part, perhaps a CPU core at lower resolutions. The rank distribution suggests that many combos using a more modern CPU achieve higher frame rates in this specific gameplay arise because the game core nearly always ties to the CPU’s responsiveness. Yet the 1769 figure still shows that the pairing is competitive, and the framerate data proves it can deliver 100+ fps at most settings.

Measured FPS Breakdown

The following table provides the complete set of measured frame rates for every resolution and preset tested.

| Resolution | Setting | Average FPS | Min FPS | Max FPS |

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

| 1920×1080 | Low | 125 | — | — |

| 1920×1080 | Medium | 113 | 96 | 130 |

| 1920×1080 | High | 102 | — | — |

| 1920×1080 | Ultra | 100 | — | — |

| 2560×1440 | Low | 110 | — | — |

| 2560×1440 | Medium | 99 | 85 | 114 |

| 2560×1440 | High | 90 | — | — |

| 2560×1440 | Ultra | 89 | — | — |

| 3840×2160 | Low | 92 | — | — |

| 3840×2160 | Medium | 83 | 71 | 96 |

| 3840×2160 | High | 66 | — | — |

| 3840×2160 | Ultra | 63 | — | — |

The only preset for which 90 has min/max is Medium, and all configurations — minimums and maximums are only available for the Medium preset. The Medium test runs show that the 1080p frame provides a window between 96 and 130, 1440p between 85 and 114, and 4K between 71 and 96. These ranges show substantial fluctuation, but the 90th percentile remains stable even in heavy parts of the frame. The step between High and Ultra is the smallest: at 1440p it only drops from 90 to 89 (1.1%); at 4K, the change is from 66 to 63 (4.5%). That suggests The Ultra preset is not hugely more costlier than High, but still brings some effects that added load.

GPU Role

The GPU in this configuration is an NVIDIA GeForce RTX 4070 SUPER, based on the Ada Lovelace architecture and fabricated on a 5 nm TSMC process. That chip, specific AD104, carries 7,168 shading units, 224 texture mapping units, and 80 ROPs across a 192-bit interface. The card sits on a dual‑slot cooler and requires one 16‑pin PCIe power connector, drawing a rated TDP of 220 W. These specifications are plenty for a title published in 2013, and the measured fps reeks up to 125 at a low preset. The GPU’s memory subsystem uses 12 GB of GDDR6X running at 1313 MHz (21 Gbps effective), delivering a bandwidth of 504.2 GB/s over a 192‑bit bus. Memory ‹ in this game is not likely to limit the system, as the highest texture cache is less than 12 GB VRAM.

The proportions of the GPU also include 56 physical ray tracing cores, 224 tensor cores, a pixel rate of 198.0 GPixel/s, and a texture rate of 554.4 GTexel/s. The 294 mm²‑large die demonstrates a 121.8M transistors per mm² density. The real decisive factor for Thief is the raw mathematics of the GPU: single‑precision compute of 35.48 TFLOPS and the direct access from the PCIe 3.0 x16 bus with a 16‑lane connection to the CPU. This results in no practical bottleneck, as the game is modest. From the benchmark dataset, the GPU’s average score is 43,223 and sits in the 83rd percentile, illustrating its headroom.

One percent: 0.1% ahead of the NVIDIA Quadro M6000 24 GB, 0.1% behind the RTX 5050 Mobile, 0.2% behind the Quadro M6000, and 1.0% behind the RTX 4090 Mobile. Those deltas are petite because all those GPUs are closely spaced in the comprehensive profile. The 4070 SUPER’s capability comfortably stuffs the single frame-generation pipeline, but the game’s maximum visual output does not push the card to its boundary at 1080p. In fact, even at 4K the card is reporting an average above 60 without v-sync, indicating it is not fully saturated for the title’s load.

Settings Recommendations

The ideal setting—the one that maximizes visual quality while maintaining a fluid display without excessive sacrifice—is the High preset. At 1080p High delivers 90 fps at an average, which is plenty for any 60 Hz or 120 Hz display; 1440p High gives 90 fps (which is also useful for a 144 Hz monitor when using syncing and allows you to hit the refresh limit), and even 4K High maintains a 66 fps average, which touches sixty if a 60 Hz monitor allows. The one closer difference between High and Ultra is minimal (1–3 fps at 1440p/4K), yet Ultra also employs more shadows and physically‑based shading effects, so the High preset yields a difference in visual reproduction at a fraction of the cost. At Medium the framerate is comparable with high but the quality is slightly lower; however, the 4K Medium (83 fps) is much faster than High (66 fps) at the same resolution. Thus, if the player owns a 144 Hz panel, they may prechoose Medium at 1440p to stay above 99 fps. For a system aiming at 60 Hz, High is the most balanced that saves nearly every margin.

At 1080p, even Ultra is sufficient: 100 fps average is 67% above the needed 60; so any setting is equally viable. The only reason to use Low or Medium would be to manually raise FPS for high refresh blogs or to enable DLSS in other games—but here Thief does not support it. The ‟ best best experience” across all resolutions is High, because it takes advantage of the GPU’s comfort zone without paying the high load of Ultra.

CPU Role

The Intel Core i5-10400F, with the brand name Comet Lake, is built on Intel’s 10 years old process; it is a 6‑core, 12‑thread parts with a base clock of 2.90 GHz and a maximum 4.30 GHz boost. The processor carries 64 KB L1 per core, 256 KB per L2 per core, and 12 MB shared L3. It supports dual‑channel DDR4 memory at a theoretical bandwidth of 42.7 GB/s (Intel’s peak). The chip uses an Intel Socket 1200 and a 65 W TDP, with PCIe Gen 3.0 and 16 lanes from the CPU. It has no integrated graphics, relying on the dedicated card.

In benchmark data, the processor averages a total score of 14,185 spanning various tests; it champions 3DMark single‑threading on 1,688 score? Actually the provided data shows a 3DMark single‑thread of 688, which is the lowest among all tests. However, the benchmark package includes Cinebench R23 multi‑core at 10,283 points and a Geekbench multi‑core of 6,257. The single‑thread score is 1,451 in Cinebench R23, indicating that per‑core performance is modest for modern standards but well beyond what Thief (released 9 years prior) demanded.

Thief’s engine is known to have a notable dependency on CPU in the form of physics, AI, and game logic. The i5-10400F’s 6 cores thrived, and the measured frame rate scaling shows that at 1440p, a 11–12% drop from 1080p is about the same across settings, indicating that the CPU is not a bottleneck until you push into 4K, where GPU load growth exceeds the CPU. The 1440p and 1080p numbers are close, which implies that a 4‑thread workload part of the game may be running on 2 cores. The CPU's max‑threads score of 4,785 from 3DMark shows only a slightly advantage over the 2‑threads score (1,350), indicating that scaling beyond 2 threads is not effective for this game. We can infer Thief core count threshold. With 6* cores / 12 threads, the i5-10400F has extra capacity, but the processor is not the exact limit. The 1080p average of 125 fps (Low) is a sign that the : GPU is far from bottleneck at this configuration, so the maintained ceiling —overhead determined by the game.

The i5-10400F’s memory bandwidth of 42.7 GB/s is not a issue for a 2014 game. The 12 threads are ample, and the chip provides enough single‑core throughput to keep the 4.30GHz boost core active. The union of these details reaffirms that the CPU’s role in the combination is to let the GPU floor out at low resolution. The result is a balanced pairing that sustains very high frame rates for a last‑gen title, proving a strong choice for Thief enthusiasts.