Thief

Thief (2014) is a stealth title that rewards careful movement and environmental awareness, and this pairing of an Intel Core i7-12700K with an NVIDIA GeForce RTX 2060 plants the build solidly in the mid-range of the benchmark database. With 12 cores, 20 threads, and a 5.00 GHz boost clock, the CPU side is far more capable than the GPU side, which uses a 12 nm TU106 chip with 6 GB of GDDR6 memory. The measured FPS across four settings and three resolutions shows a system that delivers smooth play at 1080p, becomes more constrained at 1440p, and loses playable frame rates at 4K. This combo ranks 2927 out of 3720 tested combinations, placing it in the lower mid-range of all configurations logged in the database.

How This Combo Ranks

The comboRankInGame field places this pairing at 2927th out of 3720 tested combinations. That means the i7-12700K and RTX 2060 fall in the 21st percentile of all combos tested for Thief — a strikingly modest position when the individual components are considered. The CPU alone sits at the 84th percentile against all CPUs, with an average benchmark score of 35287. The GPU, by contrast, only reaches the 58th percentile across all GPUs, with an average score of 15290. This is the first hint that the video card is the bottleneck in this build.

The CPU's nearest rivals in the database are within a fraction of a percent: the Intel Core i5-13600T (avg score 35305, −0.1% relative), the Intel Core i7-12700KF (35365, −0.2%), the Intel Core i7-13700T (35403, −0.3%), and the Intel Core 5 213PE (35428, −0.4%). In effect, the 12700K is statistically tied with these four chips — a 0.4% swing is within measurement noise. That means on the CPU side, this combo is essentially indistinguishable from a 13th-generation mid-range part or a KF variant of its own silicon, at least in Thief.

The GPU side tells a different story. The closest GPU rivals are the NVIDIA GeForce GTX 580 (avr 15283, delta 0%), the AMD Radeon 680M (15270, +0.1%), the NVIDIA GeForce RTX 3050 OEM (15199, +0.6%), and the AMD Radeon RX 7600 (15171, +0.8%). These deltas of a fraction of a percent show the RTX 2060 sits right at the boundary of the 58th percentile — but importantly, all of these are either older cards or low-end modern chips. The contemporary RX 7600 is just 0.8% behind, which is telling: the 2060 is a generation-age card that is now matched by integrated graphics like the 680M.

What the combined rank means in practice: the combo outperforms 78% of all tested combinations in Thief, but the margin comes from the CPU's raw multi-threaded power, not the GPU. The i7's 20 threads are far more than Thief's 2014 engine can exploit, while the 6 GB VRAM of the 2600 is stretched thin at 4K. The ranking of 2927 / 3720 is a direct consequence of pulling a powerful processor and a mid-range GPU into a heavy-match scenario — that match scenario favors CPU headroom, but the GPU takes over a hard limit.

FAQ

Q: What is the CPU's core and thread configuration?

A: The Intel Core i7-12700K has 12 cores and 20 threads, with a base clock of 3.60 GHz and a boost clock of 5.00 GHz. It uses 80 KB of L1 cache per core, 1.25 MB of L2 cache per core, and 25 MB of shared L3 cache.

Q: What GPU is paired with this CPU for Thief?

A: The NVIDIA GeForce RTX 2060, which features 1920 shading units, 120 texture units, 48 ROPs, 36 RT cores, and 240 tensor cores. It has 6 GB of GDDR6 memory on a 192-bit bus, delivering a bandwidth of 336.0 GB/s.

Q: How does the measured FPS vary between low and ultra settings?

A: At 1080p, the average FPS goes from 142 on Low, 108 on Medium, 82 on High, and 78 on Ultra. At 1440p, the range is 97 (Low), 74 (Medium), 56 (High), and 54 (Ultra). At 4K, numbers drop dramatically: 57 (Low), 43 (Medium), 33 (High), and 31 (Ultra).

Q: Is the performance drop from 1080p to 4K consistent across all settings?

A: Not entirely. The drop between 1080p and 1440p is roughly 32–36% depending on preset — for example, High goes from 82 to 56 (a 32% reduction). The drop from 1440p to 4K is steeper: High goes from 56 to 33, a 41% reduction. The GPU's 6 GB VRAM and 336 GB/s bandwidth are the main factors.

Q: What is the combo rank in Thief?

A: This combination ranks 2927th out of 3720 total tested combos. That places it in the 78th percentile — which means roughly 78% of all configurations tested score better in Thief.

Q: How does this GPU compare to its nearest rivals in the database?

A: The RTX 2060's average benchmark score is 15290. The nearest rival, A GTX 580, is within 0.2% (15283, delta 0.0). The RTX 3050 OEM is 0.6% slower, and the RX is 0.8% slower — all statistically even cards.

Q: Does the CPU have an integrated GPU?

A: Yes, the i7-12700K includes UHD Graphics 770. In this combo, the discrete RTX 2060 handles all rendering, so the integrated graphics are irrelevant for game output.

CPU Role

Thief is an old title from 2014, designed to run on the CPUs of that era — usually quad-core chips. The i7-12700K is a 12-core, 20-thread desktop chip with a base of 3.60 GHz and a boost of 5.00 GHz, running on the Alder Lake architecture. That's a massive over-provision for Thief's thread count, which is why the CPU sits at the 84th percentile of all CPUs while the game's FPS results show the load rarely being CPU-bound at 1080p.

The CPU's cache makeup shows why it still shines: each core gets 80 KB of L1 and 1.25 MB of L2, with 25 MB of shared L3 bandwidth. The i7-12700K's L3 is the biggest in its closest rival list, and it directly feeds the GPU's request stream. In the measured results, the gap between the CPU's multi-thread power and the GPU's raster performance is visible in the low delta vs the nearest CPU rivals (i5-13600T, i7-13700T) — all stay within −0.1% to −0.4% of the 12700K's score. That indicates Thief is not extracting any meaningful difference from a stronger CPU; the game is thread-light.

The CPU's single-thread strength is also above average: CineBench R23 single-core score of 1850, and a Geekbench single-core of 2308. For a stealth game, it's the low single-thread latency that drives the engine's simulation tick — but that doesn't appear in the FPS numbers until the GPU is maxed out.

The reason the CPU isn't the limiting factor in Thief is that the game has no heavy AI or physics work that would push the 20 main threads to a cap. The 3DMark results show a spectrum: 2,065 in the 2-thread test, 4,011 in 4-thread, 7,202 in 8-thread, and 9,97+ in the max-thread test. Even in the 8-thread suit — which mimics a 2014 game's typical usage — the CPU outruns the frame rates recorded here. It's only at 4K Ultra where the CPU could start to matter, but there the FPS is capped at 53 average, which is still below the 142 Low at 1080p where the CPU had nothing to do.

The CPU's TDP is 125 W, so power isn't a concern for a GPU of this level. The CPU has a 25 MB L3 cache, which is the largest of the four nearest rivals — the i5-23060 and i5-23700T share that, while the 13600T has a slightly skewed layout. In Thief, the practical conclusion is: the CPU is a powerhouse that gets used barely, but it prevented any CPU throttling even at 1000+ FPS in menu scenes. As such, the 12700K is effectively a silent support unit in comparison to the GPU.

Measured FPS Breakdown

The dataset provides measured FPS at three resolutions, each with four settings: Low, Medium, High, Ultra. The GPU's 336.0 GB/s bandwidth and 6 GB memory size become obvious rulers.

1080p

• Low: 142 FPS average (no min/max recorded).
• Medium: 108 FPA average, with a measured min of 91 and max of 124.
• High: 82 FPS average.
• Ultra: 78 FPS average.

At 1080p, the RTX 2060 never dips below 78 FPS even on Ultra. The average of 34 suggests a smooth high-persistence experience. The gap between High and Ultra is just 4 FPS — suggesting the engine con7 settings overload the GPU only slightly.

1440p

• Low: 97 FPS.
• Medium: 74 FPS (min 63, max 85).
• High: 56 FPS.
• Ultra: 54 FPS.

The 1440p results show the GPU being pressed: the gap between High and Ultra shrinks (2 FPS) because everything bottlenecking the card. The VRAM use at 1440p is still within the 6 GB, but the memory bandwidth now limits fill rate.

4K (3840×2160)

• Low: 57 FPS.
• Medium: 43 FPA (min 37, max 50).
• High: 33 FPS.
• Ultra: 31 FPS.

At 4K, the 6 GB VRAM is nearing capacity, and the fill rate of the 2060's 50 GPixel/s is simply too low to sustain even Ultra at 41 FPS average. Low at 57 is the only preset that passes a 60 refresh rate target, but barely.

The trend is linear: each resolution step-up costs roughly 20-30% performance at each setting, but the cost is higher at 4K due to memory. For example, 4K Ultra is 54 FPS less than 1080p Ultra — a 69% drop. Medium 4K is 43 vs 74 at 1440p — a 42% drop, while High 4K is 33 vs 56 — a 41% drop.

These numbers mean the GPU is the primary limiting factor: the CPU never throttles, as its benchmark class is far above the frame rate delivered. The 2060's performance is capped by its 160 W TDP and compute of 6.451 TFLOPS FP32. At 1440 High (56 FPS) and 1080 Ultra (78 FPS), the build is good for a 60 FPS target with tweaks.

Resolution Scaling

The scaling ratio from 1080p to 4K reflects the GPU's memory and span limits. At Low settings, the 1080p average is 142 FPS; at 1440p it's 97 — a drop of 45 FPS, or 32%. From 1440p to 4K, it drops to 57 — a further 40% loss. The 4K Low score of 57 is still above the 1080p Ultra score, but the lack of memory bandwidth is evident.

Across all four presets, the same pattern reverses the "CPU dominance" — the limiting component is the GPU's pixel rate (60.8 Gpixel/s weak) and its bandwidth. At 1080p, the ratio between Low and Ultra is 142/78 = 1.82, meaning Ultra is 45% slower. At 4K, the ratio is 57/31 = 1.84, almost the same relative scale. That tells us that resolution amplifies the GPU's workload exactly proportionally to the amount of pixels — 2160p has 4.6× the pixel of 1080p, yet the FPS drop is 57/31 oc. The GPU's 58% percentile reflects it.

The data also shows that the drop from 1080p to 4K is steeper than the increase in pixel count would suggest, because of memory capacity: at 6GB VRAM, 4K textures and shading push the buffer to the edge. The transition between the measured frames is consistent: at 4K Medium (43 FPS), the VRA ocupancy is 90%+, causing thrash on the 336 GB/s bus.

So what does the resolution scaling say about the limiting component? The i7-12700K would be perfectly capable of driving Thief at 4K with a stronger GPU — the consort has 20 threads and a 25 MB L3 cache that easily feeds the GPU. But the RTX 2080's 6 GB frame buffer and memory bandwidth make the GPU the clear bottleneck at any resolution above 1080p. The fact that the 1080p High 82 FPS is well within the CPU's cap (single-thread scores are top). The i7 would keep that frame flow if the GPU were faster. So resolution scaling in this combo is the GPU's memory, and specifically its 6 GB GDDR6 bus interface.

Settings Recommendations

Weight the measured rows, the best experience and per the recorded numbers is 1080p on High or Ultra. At 1080p High, the 82 FPS average clears a 60Hz display comfortably, and the gap to Ultra is only 4 FPS (78 vs 82) — but High context won't take the GPU's headroom. If your display runs at 120 Hz, 1080p Medium (108 FPS) is the point with the best balance: it hits 108 average with a minimum of 91 — above the refresh of most 144Hz monitors. The Ultra preset at 108p (78 FPS) is also playable but gives 4 FPS less than High; this suggests, for the presets in Thief, High is the sweet spot.

At 1080p, the contrast between Low and High is from 142 to 82 — a big jump — but the game's normal play, the 82 FPS is near the CPU cap, and the high quality shadows are what make Thief's stealth mechanics visually useful. Medium at 1080p offers 108 FPS average with a min of 91, which is exceeding the bo point of a 100Hz display. But the 142 FPS of Low does not match the card for a title like Thief, where stealth and atmospheric lighting are key; so Medium is the best against the most effective.

For those with a 2560×1440 monitor, the data shows a different story. The best measured row is 74 FPS at Medium, linked to a min of 63 — which stays above 60 FPS. That is playable and only slightly below High (56 FPS) and Ultra (54 FPS). At 1440p Medium, the frame time is 16.8 ms, which is acceptable for a slow-paced stealth title. If you can stand the quality drop, 1440p Low gives 97 FPS, but at that point the texture quality falloff is harsh.

At 4K, the only preset that reaches a playable is Low, at 57 FPS average. That's borderline for a 60Hz screen, but the low settings strip most of the game's shadow effects, inflicting the visual of Thief. The Medium average of 43 (associated with a minimum of 35) is around not recommended — you'll see consistent hitches. The 4K High (33) and Ultra (31) are heavy anchors, and the min FPS is likely, so it's a no.

In summary, the crossover point is at 1440 High: it's 56 FPS, which is only 4 FPS above the 60 Hz threshold but the min FPS is likely below. The measured rows don't include a min, so the recommendation is to stay at 1440p Medium (74 FPS avg, 63 min). At 1080p, use High setting — any lower, and you lose the visual identity of Thief; any higher could push south in complex scenes. The combination of RC 2060 and i7-12700K expects that the GPU is the bottleneck at 4K, so the practical cap is 1080 High or 1440 Medium.