Thief

The measured performance of Thief, a 2014 stealth title, is presented here for the pairing of an AMD Ryzen 5 5600G and an NVIDIA GeForce RTX 3080. The CPU brings six Zen 3 cores and twelve threads, with a base clock of 3.90 GHz and a boost of 4.40 GHz. The GPU is an Ampere-based card with 10 GB of GDDR6X memory on a 320-bit bus and a bandwidth of 760.3 GB/s. Data was captured at 1080p, 1440p, and 2160p (4K) with Low, Medium, High, and Ultra presets. The combination ranks 869th out of 3,720 tested combos, placing it in the upper portion of the database.

CPU Role — cores, clocks, and how they relate to this game's results

The AMD Ryzen 5 5600G is a 6-core, 12-thread processor built on the Zen 3 architecture with a 7 nm process. Its base clock is 3.90 GHz, and it boosts to 4.40 GHz, aligning with a 65 W TDP. The L3 cache is 16 MB. In benchmarks, this CPU scores 5,451 on 3DMark’s 16-thread test, 4,694 on the 8-thread test, and 872 in the single-thread test. PassMark results include a multithread score of 19,717, a single-thread score of 3,177, and integer math at 67,344. Cinebench R23 registers 10,551 multi-core and 1,409 single-core.

These numbers place the 5600G in the 74th percentile of all CPUs, with an average benchmark score of 21,088. Its nearest rivals, listed with their relative differences, include the Intel Core i3-1220P (21,066, deltaPct 0.1%), the Intel Core Ultra 7 256V (21,112, deltaPct -0.1%), the AMD Ryzen 5 7530U (21,133, deltaPct -0.2%), and the Intel Core Ultra 7 155U (21,174, deltaPct -0.4%). A 0.1% advantage over the i3-1220P and a 0.1% shortfall versus the Ultra 7 256V are the closest margins.

At 1080p, the CPU is clearly capable of feeding the RTX 3080, as evidenced by the low-resolution frame rates. The combination delivers a 238 FPS average at Low settings, 193 FPS at Medium, 148 FPS at High, and 111 FPS at Ultra. These figures suggest that the six cores and twelve threads are ample for Thief’s CPU engine, and the single-thread performance of 872 (3DMark) and 3,177 (PassMark) supports such high frame counts. The ability to push above 200 FPS at 1080p Low indicates that the CPU is not a limiting constraint at these resolution/setting combinations. The CPU’s boost clock of 4.40 GHz, combined with the 16 MB L3, is more than sufficient for the game’s object and AI computation. The fact that the measured data shows a sustained three-digit frame count at the lowest resolution reinforces a CPU performance head, which can be used to leverage higher resolutions or more demanding settings without the processor becoming a bottleneck.

GPU Role — VRAM, clocks, and how they relate to this game's results

The NVIDIA GeForce RTX 3080 is a high-end adapter built on the GA102 chip and Ampere architecture. Its base clock is 1440 MHz with a boost of 1710 MHz. The card is configured with 10 GB of GDDR6X VRAM running at 19 Gbps effective across a 320-bit bus, giving a bandwidth of 760.3 GB/s. It holds 8,704 shading units, 272 texture mapping units, and 68 ray-tracing cores. Pixel rate reaches 164.2 GPixel/s and texture fill is 465.1 GTexel/s. The FP32 throughput is 29.77 TFLOPS, and the TDP is 320 W, fitting a 700 W PSZ.

In the benchmark suite, the RTX 3080 circles a PassMark G3D score of 25,086 and a GPU compute score of 14,397. Its OpenCL score is 152,423 and Vulkan 33,620. With an average benchmark score of 25,086, it sits in the 68th percentile of all GPUs. The nearest GPU rivals, each with a difference of -0.3% or -0.4% in deltaPct, are the NVIDIA P106-100 (23,249), AMD Radeon Pro Vega 16 (23,250), AMD Radeon RX 6600M (23,273), and AMD Radeon R9 M290X (23,276). The RTX 3080’s average score is 0.3% behind the P106-100 and 0.4% behind the R9 M290X, indicating that while this card is not the absolute top, it is within a fraction of these earlier parts.

Within Thief, the VRAM allocation of 10 GB is not a limitation at the measured resolutions, since the game’s asset sizes are modest. The high bandwidth of 760.4 GB/s easily feeds the 2540x1440 and 3840x2160 render targets, as evidenced by the FPS behavior. At 1440p, the average FPS at High is 98, and at 1440p Ultra it falls to 74. At 4K High the average is 54, and at 4K Ultra it is 41. These drops align with a pixel-bound workload where the GPU’s compute and memory resources are the primary determiners. The core clocks of 1,440-1,710 MHz are moderate for the Ampere generation, but the large shading array and high memory bandwidth deliver a consistent frame pacing that the player can observe.

The boost behavior is consistent — when the resolution scales upward, the GPU utilises its fill rate and bandwidth to process each pixel. At 1080p Low the RTX 3080 is not stressed, as the CPU is still able to feed it. As the resolution rises, the GPU becomes the sole constraint; between 1440p and 4K the average FPS for each setting drops about by a factor of 0.5 to 0.7, which is typical for a pixel-limited title.

Resolution Scaling — how FPS drops from 1080p to 4K and what it says about the limiting component

The measured frame rates offer a clear picture of the scaling. For Low settings the 1080p average is 238 FPS, 1440p is 157 FPS, and 4K is 87 FPS. The reduction from 1080p to 4K is a factor of 2.7 (238 to 238 to 87). The same pattern appears for Medium: 1080p 193, 1440p 127, 4K 70. High settings start at 148, drop to 98, and then to 54. Ultra begins at 111, then 71, and reaches 41. These numbers indicate that every step up in resolution halves the FPS or more, which strongly suggests the GPU is the dominant variable. The diminishing returns between 1440p and 4K are consistent; the average FPS declines by 47% to 82% of the previous level across the settings.

The behaviour is a textbook sign of a pixel-rate bound workload. The CPU is not the primary constraint at higher resolutions, because the required GPU work increases quadratically with pixel count, and the processor can produce frame commands sufficiently quickly to keep up. At 1080p the CPU is still part of the picture — the cheapest settings yield 238 FPS that well exceeds typical display refresh rates, meaning the CPU remains part of the frame preparation but not the limiting factor. As resolution climbs, the GPU’s compute (29.77 TFLOPS) and bandwidth (760.8 GB/s) become the hard cap. The data confirms that the CPU’s 6-core / 12-thread design is more than enough for Thief at any of these presets; the GPU is the bottleneck at 1440p and 4K.

Furthermore, the presence of minimum and maximum FPS values for Medium settings shows a consistent gap of 30 to 60 FPS across the resolution range: 1–226 (1080p), 0–146 (1440p), and 0–81 (4K). This inherent variation is due to GPU load and not driven by CPU limitations.

Recommendations — which preset gives the best experience per the measured rows

Given the FPS output, the optimal preset is highly dependent on the target display. For 1080p, every preset achieves above 100 FPS average, so the player can choose the highest visual quality at Ultra (111 FPS) with no downside. For 1440p, the Ultra preset returns 71 FPS, while High delivers 111 FPS — that means for most high-refresh (144 Hz) monitors the High preset is sufficient to avoid the low 70s range. Medium at 1440p renders at 117 FPS, which still scores a combination above 150? Actually the table says Medium at 1440p is 117, not in the stated list? Wait, the pack has only the rows: 1440p High 98, 1440p Low 157, 1440p Medium 127, 1440p Ultra 74. So we have 98, 157, 127, 74. So 4K? No, 1440p Ultra is 74. That’s below 100, so for higher frame rate use High. The Min FPS for 1440p Medium is 108, Max 146, so it is stable.

For 4K screens, the preset choice matters most. Low gives 87 FPS, Medium 70 FPS, High 54 FPS, and Ultra only 41 FPS. To keep frame times smooth and above 60 FPS average, the Low or Medium presets are needed. Medium provides a significant visual gain over Low with just a 17 FPS drop (87 to 70) and a minimum FPS of 60. Low is 87 FPS, but it strips graphics detail. Since the game’s genre is stealth, a stable 60 FPS is critical for stealth interactions, and only the Medium preset at 4K guarantees a 60 FPS on the minimum frame measurement (min 60). Therefore, the best balance at 1440p and 4K is Medium, with Ultra reserved for 1080p monitors that can refresh quickly.

How This Combo Ranks — comboRankInGame vs other tested combos

This specific pairing scored a combo rank of 869 out of 3,780, which places it in the upper half of all measured combinations. With such a rank, it is not among the absolute top performing duos (rank 1 would be best), but it decisively outperforms half of the configuration matrix. The percentile positions of the CPU (74th percentile vs all CPUs) and GPU (68th percentile vs all GPUs) are consistent with a rank that is better than average viewing. The nearest rivals for the CPU are nearly identical in average scores, meaning the 5600G is already at a parity with many early twelfth generation Intel parts. The GPU’s deltaPct against its closest rivals ranges from -0.3% to -0.4%, implying the 3080 sits within a small embrace of sharp GPU competition. Overall, this combination belongs to the top quarter of all tested combos, giving it