Thief

The AMD Ryzen 5 4600G paired with the NVIDIA GeForce RTX 5070 Ti produces a striking split in Thief: the GPU is clearly the dominant force, yet the aging CPU holds back the highest frame rates. Benchmark data shows this combo ranks 693rd out of 1642 tested configurations, placing it in the upper-middle tier. This analysis examines how the resolution scaling, CPU limits, and GPU headroom interact to define the experience.

Resolution Scaling

The measured FPS drop from 1080p to 4K reveals a system that scales predictably, but with a notable twist at lower settings. At Ultra settings, the average frame rate falls from 102.4 FPS at 1920x1080 to 91.9 FPS at 2560x1440, then further to 69.6 FPS at 3840x2160. That is a 32.8 FPS loss between 1080p and 4K, a 32% reduction that points to the GPU becoming the primary bottleneck as pixel count rises. The RTX 5070 Ti’s 896.0 GB/s memory bandwidth and 16 GB of GDDR7 VRAM are sufficient to handle the higher resolution without a dramatic collapse, but the trend is clear.

High settings tell a similar story: 108.2 FPS at 1080p, 93.7 FPS at 1440p, and 71.6 FPS at 4K. The drop from 1080p to 4K here is 36.6 FPS, or 33.8%, which is slightly steeper than at Ultra. This suggests that at Ultra, some of the load shifts to the CPU, slightly masking the GPU’s scaling curve. Medium settings follow the pattern: 114.5 FPS at 1080p, 102 FPS at 1440p, and 88.5 FPS at 4K, with a 26 FPS decline across the range.

Low settings are the outlier. The jump from 131.8 FPS at 1080p to 116.4 FPS at 1440p is only 15.4 FPS, but the drop to 96.1 FPS at 4K is more pronounced. The 4K Low result is only 4.4 FPS higher than the 4K Medium result, and it is 6.5 FPS higher than 4K High. This compression at 4K Low suggests that the CPU is becoming a limiting factor even at the lowest quality settings, preventing the GPU from fully stretching its legs. The RTX 5070 Ti’s 43.94 TFLOPS of FP32 compute is capable of far more than 96 FPS at 4K Low in a 2014 game, but the Ryzen 5 4600G’s modest single-thread performance caps the ceiling.

CPU Role

The Ryzen 5 4600G is a 6-core, 12-thread processor based on the Zen 2 architecture, running at a base clock of 3.70 GHz and a boost clock of 4.20 GHz. Its benchmark scores show a processor that is competent for multithreaded work but weaker in single-thread performance. The 3dmark single-thread score of 726 and the cinebench r23 single-core score of 1919 are modest figures, especially when compared to the processor’s own multicore results. The cinebench r23 multicore score of 13593 indicates that the CPU can handle heavy workloads, but Thief, as a game from 2014, does not scale well across many cores.

The 3dmark scores tell the threading story. The 2-thread score is 1430, the 4-thread score is 2732, the 8-thread score is 4138, and the max-thread score is 4889. The jump from 2 to 4 threads is significant, but the gain from 8 to max threads is only 751 points (18%). This suggests that Thief benefits from up to 4 or 8 threads, but adding more beyond that yields diminishing returns. The 4600G’s 12 threads are more than enough for this game, yet the per-core performance is the bottleneck.

At 1080p, the CPU’s influence is visible. The difference between Low (131.8 FPS) and Ultra (102.4 FPS) at 1080p is 29.4 FPS, which is a larger gap than the 1080p-to-4K scaling at Ultra (32.8 FPS). This indicates that at 1080p, the CPU is struggling to feed the GPU, and lowering settings does not unlock the full potential of the RTX 5070 Ti. The 4600G’s 8 MB of shared L3 cache and dual-channel DDR4 memory with 51.2 GB/s bandwidth are additional constraints, though they are difficult to isolate in these aggregate results.

GPU Role

The RTX 5070 Ti is a Blackwell 2.0 architecture GPU with 8960 shading units, 280 texture mapping units, and 96 raster output units. Its base clock is 2295 MHz with a boost clock of 2452 MHz, and it operates with a 256-bit memory bus delivering 896.0 GB/s of bandwidth. The GPU’s 16 GB of GDDR7 memory is ample for Thief at any resolution. The 3dmark steel nomad dx12 score of 6604 and a passmark g3d score of 32974 place it in the 87th percentile of all GPUs, indicating strong performance relative to the broader market.

The GPU’s role in this pairing is to carry the load at higher resolutions. At 4K Ultra, the combo achieves 69.6 FPS, which is above the 60 FPS target for smooth gameplay. The delta between 4K High (71.6 FPS) and 4K Ultra (69.6 FPS) is only 2 FPS, suggesting that the GPU can handle the extra visual fidelity without a major penalty. In contrast, the delta between 4K Low (96.1 FPS) and 4K Medium (88.5 FPS) is 7.6 FPS, showing that the GPU has more headroom at lower settings but is still held back by the CPU.

Comparing to the GPU’s nearest rivals contextualizes its strength. The RTX 5070 Ti has an average benchmark score of 52086, which is statistically identical to the NVIDIA GeForce RTX 4080 (52085, 0% delta). It is 2.5% ahead of the NVIDIA RTX A1000 (50826) and 2.5% behind the AMD Radeon Pro W6600M (53414). This means that in pure GPU terms, the RTX 5070 Ti is a top-tier part, but the measured FPS in Thief does not reflect that dominance due to the CPU pairing.

How This Combo Ranks

The combo rank of 693 out of 1642 tested configurations places this pairing in the 42nd percentile, which is below what the GPU alone would suggest. The RTX 5070 Ti’s GPU percentile is 87, meaning it outperforms 87% of all GPUs, but the CPU’s percentile is 75. The combination of a high-end GPU with a mid-range CPU creates a mismatch that drags the overall ranking down.

The nearest rival CPUs provide context for the CPU’s limitation. The AMD Ryzen 3 PRO 5355GE has an average score of 17482, a 0% delta from the 4600G’s 17489. The Intel Core i5-12450H scores 17475 (0.1% delta), the Intel Core i3-14100T scores 17636 (-0.8% delta), and the Intel Core i3-13100F scores 17665 (-1% delta). These rivals are all within 1% of the 4600G, confirming that the CPU is not an outlier but rather a typical mid-range part. The difference between the 4600G and these rivals is negligible, so the 693rd ranking is more about the GPU being underutilized than the CPU being uniquely weak.

The data shows that this combo is better suited for 1440p and 4K gaming than 1080p. At 1440p High, the 93.7 FPS result is respectable, and at 4K High, the 71.6 FPS is playable. The 1080p results, while higher in absolute terms, do not scale as expected with settings changes, indicating a CPU bottleneck that prevents the GPU from achieving its potential frame rates.

FAQ

Q: What is the best resolution for this combo in Thief?

A: The 2560x1440 resolution offers the best balance. At High settings, it achieves 93.7 FPS, which is only 14.5 FPS lower than 1080p High (108.2 FPS) but 22.1 FPS higher than 4K High (71.6 FPS). The 1440p results avoid the severe CPU bottleneck seen at 1080p while maintaining high frame rates.

Q: How much faster is Low settings compared to Ultra at 4K?

A: At 3840x2160, Low settings average 96.1 FPS, which is 26.5 FPS faster than Ultra’s 69.6 FPS. This 38% improvement shows that the GPU has headroom at lower settings, but the CPU begins to limit further gains.

Q: Is the RTX 5070 Ti being fully utilized by the Ryzen 5 4600G?

A: No. At 1080p, the difference between Low (131.8 FPS) and Ultra (102.4 FPS) is 29.4 FPS, which is nearly the same as the 1080p-to-4K drop at Ultra (32.8 FPS). This suggests the CPU is the limiting factor at 1080p, preventing the GPU from reaching its full potential.

Q: How does this combo compare to the RTX 4080 pairing?

A: The RTX 5070 Ti and RTX 4080 have identical average benchmark scores (52086 vs 52085, 0% delta). In this specific system, the measured FPS would be effectively the same, as the CPU is the bottleneck in both cases.

Q: What is the single-thread performance of the 4600G?

A: The 3dmark single-thread score is 726, and the cinebench r23 single-core score is 1919. These are modest figures that explain the CPU’s limitation in a game like Thief, which relies heavily on single-thread performance.

Q: Does lowering settings help at 1080p?

A: Yes, but with diminishing returns. The jump from Ultra (102.4 FPS) to Low (131.8 FPS) at 1080p is 29.4 FPS, but the jump from High (108.2 FPS) to Low is only 23.6 FPS. The CPU bottleneck prevents Low settings from unlocking the full GPU potential.

Settings Recommendations

The measured data points toward Medium settings at 2560x1440 as the sweet spot. At this setting, the combo achieves 102 FPS, which is above the 100 FPS threshold for high-refresh displays and only 14.4 FPS below the 1080p Medium result (114.5 FPS). The visual quality improvement from Low to Medium is significant, and the FPS cost is only 14.4 FPS at 1440p, which is acceptable.

For 4K gaming, High settings are recommended. The 71.6 FPS average is above the 60 FPS target, and the difference from Ultra (69.6 FPS) is only 2 FPS, meaning the extra visual fidelity of Ultra comes at a negligible cost. However, 4K Low (96.1 FPS) is not worth the visual sacrifice, as the 24.5 FPS gain over High does not justify the drop in quality.

At 1080p, Medium settings (114.5 FPS) provide a good experience, but the CPU bottleneck means that even Low (131.8 FPS) cannot push much higher. The data suggests that 1080p is not the ideal resolution for this pairing, as the GPU is wasted and the CPU limits the frame rate. For users with a 1080p monitor, Medium or High settings will look better without a meaningful FPS penalty, as the difference between High (108.2 FPS) and Medium (114.5 FPS) is only 6.3 FPS.

The overall recommendation is to play at 2560x1440 with High settings, which delivers 93.7 FPS. This is 2.1 FPS higher than 4K High and 14.5 FPS lower than 1080p High, but it provides the best balance of visual fidelity and performance. The Ultra preset at 1440p (91.9 FPS) is also viable if the user prefers maximum quality, as the 1.8 FPS difference from High is imperceptible.