Thief

Thief is an older stealth title, but this pairing of a Ryzen 5 4600G with an RTX 3070 produces a wide spread of frame rates depending on how you configure it. The data shows that the game scales significantly with both resolution and quality settings, and finding the right balance is key to a smooth experience.

Settings Recommendations

The benchmark results for this combo in Thief are unusual: there is a massive performance cliff between Low and Ultra settings, but the differences within the middle tiers are more nuanced. At 1080p, the measured FPS ranges from 130 on Low down to 105 on Ultra. That 25 FPS drop is noticeable but not catastrophic, and it suggests that even the highest settings are playable on this hardware at that resolution. However, looking closer at the data, the Medium preset at 1080p delivers 118 FPS average, with a minimum of 100 and maximum of 135. The High preset scores 106 FPS, and Ultra scores 105. This means that for the cost of roughly 12 FPS compared to Medium, you get almost nothing in visual fidelity moving from High to Ultra (a 1 FPS difference), while dropping from Medium to High only costs 12 FPS and provides a substantial quality jump. The sweet spot at 1080p is clearly High, as it nearly matches Ultra in performance while being visually superior to Medium.

At 1440p, the pattern shifts. Medium delivers 104 FPS average (with min/max of 88/119), High drops to 83 FPS, and Ultra falls to 79 FPS. The gap between Medium and High here is 21 FPS, which is more significant than at 1080p. Low settings at 1440p hit 115 FPS, which is only 11 FPS above Medium. Given that the game is a stealth title where framerate consistency matters for smooth camera movement, the Medium preset at 1440p is the recommended choice. It keeps the average above 100 FPS and the minimum above 88 FPS, which should feel fluid. High is playable but drops into the 80s, and Ultra is barely above that at 79 FPS, meaning you are leaving performance on the table for minimal visual gain.

At 4K, the situation becomes more constrained. Medium is the only preset with full min/max data, showing 64 FPS average with a minimum of 54 and maximum of 74. High drops to 49 FPS, and Ultra falls to 46 FPS. Low settings hit 84 FPS. Here, the recommendation is straightforward: use Medium. It is the only preset that maintains a minimum above 50 FPS, which is important for avoiding dips into stutter territory. High and Ultra are technically above 45 FPS average, but their lack of minimum data suggests they are likely to dip lower, and the 15-18 FPS gain from High to Medium is too large to ignore. Low is faster at 84 FPS but sacrifices too much visual quality for a stealth game that relies on environmental detail.

Across all resolutions, the data indicates that Ultra is a trap for this combo. The performance difference between High and Ultra is consistently negligible (1 FPS at 1080p, 4 FPS at 1440p, 3 FPS at 4K), while the difference between Medium and High is substantial. This suggests that Ultra likely enables effects that are poorly optimized for this hardware, or that the CPU becomes a bottleneck at the highest settings. For the best experience, use High at 1080p, Medium at 1440p, and Medium at 4K.

GPU Role

The RTX 3070 is the dominant component in this system for Thief, and the data shows its memory and clock characteristics directly influence the results. The GPU has 8 GB of GDDR6 memory on a 256-bit bus, providing 448.0 GB/s of bandwidth. This is a critical factor because Thief is a 2014 title that was designed for much less capable hardware, but the high-resolution textures and effects in the Ultra preset likely push against this 8 GB limit. The fact that Ultra settings perform so close to High across all resolutions (105 vs 106 FPS at 1080p, 79 vs 83 FPS at 1440p, 46 vs 49 FPS at 4K) suggests that the game is not running out of memory at High, but the Ultra preset may be hitting a wall where additional memory allocation provides no benefit.

The GPU's clock speeds are also relevant. With a base clock of 1500 MHz and a boost clock of 1725 MHz, the RTX 3070 runs at a moderate frequency for its generation. The benchmark results show that this combo achieves 130 FPS at 1080p Low, which indicates the GPU is not fully saturated at that resolution and settings. The game's engine appears to be CPU-limited at lower resolutions and lower settings, which is why the FPS numbers plateau. The GPU's shading units (5888) and texture mapping units (184) are more than sufficient for a game this old, so the limiting factor is likely the game's draw call overhead or the CPU's ability to feed the GPU.

The GPU's memory bandwidth of 448.0 GB/s is substantial, but the 8 GB capacity is the more likely constraint as resolution increases. At 4K, the FPS drops to 49 on High and 46 on Ultra, which is a 56% reduction from 1080p High (106 FPS). This scaling is consistent with a GPU that is memory-bandwidth limited at high resolutions, especially given that the game was released before high-resolution texture packs were common. The GPU's passmark G3D score of 22214 puts it in the 61st percentile of all GPUs, and its nearest rival is the AMD Radeon RX 7600 XT with a score of 17083 (0.7% higher), indicating this is a solid mid-range performer for its generation.

Resolution Scaling

The FPS drop from 1080p to 4K on this combo is significant and reveals where the bottleneck lies. At 1080p High, the combo averages 106 FPS. At 1440p High, it drops to 83 FPS (a 22% reduction). At 4K High, it falls to 49 FPS (a 54% reduction from 1080p). This scaling pattern indicates that the GPU becomes progressively more stressed as resolution increases, but the CPU is still able to keep up until 4K. The fact that 1080p High (106 FPS) and 1440p Low (115 FPS) are relatively close suggests that the CPU is the limiting factor at lower resolutions, as lowering the settings at 1440p only yields 9 more FPS than High at 1080p.

Looking at the Medium preset, the scaling is more telling. At 1080p Medium, the combo hits 118 FPS. At 1440p Medium, it drops to 104 FPS (a 12% reduction). At 4K Medium, it falls to 64 FPS (a 46% reduction from 1080p). The 4K drop is much steeper than the 1440p drop, which is typical for a GPU that has enough power for 1440p but struggles with the pixel count at 4K. The minimum FPS data supports this: 100 FPS minimum at 1080p Medium, 88 FPS at 1440p Medium, and 54 FPS at 4K Medium. The 4K minimum of 54 FPS is only just above the 50 FPS threshold for smooth gameplay, so any demanding scene could cause noticeable hitches.

The resolution scaling also shows that the game is more CPU-bound at lower resolutions. At 1080p, the difference between Low (130 FPS) and Ultra (105 FPS) is 25 FPS, which is a 19% reduction. At 4K, the difference between Low (84 FPS) and Ultra (46 FPS) is 38 FPS, a 45% reduction. This widening gap as resolution increases confirms that the GPU is the limiting factor at 4K, while the CPU holds back the GPU at 1080p. The RTX 3070's PCIe 4.0 x16 interface helps mitigate bandwidth issues, but the Ryzen 5 4600G only supports PCIe Gen 3 with 20 lanes, which could slightly reduce data transfer rates in high-detail scenes. However, the fact that the combo still achieves 130 FPS at 1080p Low suggests this is not a major bottleneck.

FAQ

Q: Can this combo run Thief at 4K with playable framerates?

A: Yes, but only on Medium settings. The data shows 64 FPS average at 4K Medium, with a minimum of 54 FPS. High (49 FPS) and Ultra (46 FPS) are below the 50 FPS threshold for smooth play, while Low (84 FPS) is playable but sacrifices visual quality.

Q: What is the best resolution for this CPU and GPU pairing in Thief?

A: 1440p is the sweet spot. At Medium settings, it delivers 104 FPS average with an 88 FPS minimum. This provides a significant visual upgrade over 1080p while maintaining high framerates. The 1080p results are higher (118 FPS on Medium) but the difference in fidelity is more noticeable than the 14 FPS gap.

Q: Is the Ultra preset worth using in this game?

A: No. The performance difference between High and Ultra is negligible across all resolutions: 1 FPS at 1080p, 4 FPS at 1440p, and 3 FPS at 4K. Given that Ultra typically includes additional effects and higher texture quality, the lack of FPS impact suggests the settings are not scaling effectively with this hardware.

Q: Does the CPU bottleneck the GPU in Thief?

A: Yes, at lower resolutions. At 1080p, the framerate only varies from 105 to 130 FPS across all settings, which indicates the CPU (6 cores, 12 threads, 3.70 GHz base, 4.20 GHz boost) is limiting the GPU's output. At 4K, the framerate varies from 46 to 84 FPS, showing the GPU becomes the limiting factor.

Q: How does this combo's performance compare to other tested configurations?

A: This combo ranks 1968 out of 3720 tested combinations in Thief, placing it in the 47th percentile. This is a mid-table result, which is consistent with the CPU's 71st percentile ranking among all CPUs and the GPU's 61st percentile ranking among all GPUs. The combo performs adequately but is not top-tier for this game.

Q: What settings should I use for a 144Hz monitor?

A: At 1080p, use High settings to get 106 FPS (close to but below 144Hz). At 1440p, use Medium to get 104 FPS. Neither resolution can consistently hit 144 FPS, as even 1080p Low only reaches 130 FPS. For a 144Hz display, you would need to drop to 1080p Low or accept a lower refresh rate.

How This Combo Ranks

The Ryzen 5 4600G + RTX 3070 combination holds the 1968th position out of 3720 tested combos in Thief, which places it in the 47th percentile of all configurations tested. This is a below-average ranking for a modern setup, but the context matters. The CPU itself ranks in the 71st percentile of all CPUs with an average benchmark score of 17507, and its nearest rivals are all similarly performing chips: the AMD Ryzen 3 PRO 5355GE (0.1% higher score), the Intel Core 7 150U (0.6% higher), and the AMD Ryzen 3 8300GE (0.6% lower). The GPU ranks in the 61st percentile with an average score of 17208, and its nearest rival is the AMD Radeon RX 7600 XT (0.7% higher).

The discrepancy between the individual component rankings and the combo rank suggests that Thief does not fully utilize the potential of this hardware. The combo's 47th percentile ranking is likely due to the game's age and its single-threaded nature, which does not benefit from the CPU's 12 threads. The CPU's single-thread performance is moderate (726 in 3dmark single-thread, 1919 in Cinebench R23 single-core), and Thief's engine appears to be heavily dependent on single-core performance. The fact that this combo performs worse than 53% of tested configurations, despite having a GPU that is in the 61st percentile, indicates that the CPU is holding back the system in this specific title.

Compared to other combos, this result means that many systems with weaker GPUs but faster single-core CPUs would outperform this setup in Thief. The GPU's 8 GB memory and 448.0 GB/s bandwidth are more than adequate for this game, so the bottleneck is purely computational. The combo's ranking also reflects the game's age; newer titles that leverage multi-core CPUs would likely see this combo rank higher, as the Ryzen 5 4600G's 6 cores and 12 threads would be better utilized. For Thief specifically, the data suggests that a CPU with higher single-thread performance would be a better pairing with this GPU, even if it has fewer cores.

CPU Role

The AMD Ryzen 5 4600G is a 6-core, 12-thread processor based on the Zen 2 architecture (codenamed Renoir), running on the AM4 socket. Its base clock is 3.70 GHz with a boost clock of 4.20 GHz, and it has a 65W TDP. The CPU's cache configuration includes 64 KB L1 per core, 512 KB L2 per core, and 8 MB of shared L3 cache. In Thief, the CPU's role is primarily to feed the GPU with draw calls and game logic, and the benchmark data shows that it is a limiting factor at lower resolutions.

At 1080p, the framerate is nearly flat across all settings: 130 FPS on Low, 118 on Medium, 106 on High, and 105 on Ultra. This plateau indicates that the CPU is maxed out at around 130 FPS, regardless of what the GPU is doing. The CPU's multi-threaded performance is decent (13593 in Cinebench R23 multicore), but its single-thread score of 1919 in Cinebench R23 is where the limitation lies. Thief's engine, released in 2014, likely uses one or two primary threads, so the CPU's 3.70 GHz base clock and 4.20 GHz boost clock are the key factors, not the 12 threads.

The CPU's nearest rivals in benchmark scores are the AMD Ryzen 3 PRO 5355GE (0.1% higher avg score), the Intel Core 7 150U (0.6% higher), and the AMD Ryzen 3 8300GE (0.6% lower). All of these chips have similar single-thread performance, which explains why the 4600G does not stand out in Thief. The CPU's 12 threads provide no advantage in this game, as evidenced by the fact that the 3dmark 16-thread score (4863) is only slightly higher than the 8-thread score (4138) and much higher than the single-thread score (726). This scaling pattern shows that Thief is not using more than a few threads.

At 4K, the CPU becomes less of a bottleneck because the GPU is working harder. The framerate drops to 84 FPS on Low and 49 FPS on High, which shows that the GPU is now the limiting factor. However, the CPU's 8 MB of L3 cache and dual-channel DDR4 memory support (51.2 GB/s bandwidth) are sufficient for Thief's memory access patterns. The CPU's integrated Radeon Vega 7 graphics are not used in this configuration, but they do not impact performance when a discrete GPU is installed. For this game, the Ryzen 5 4600G is a capable but not exceptional partner for the RTX 3070; its single-thread performance holds back the GPU at lower resolutions, while at higher resolutions the GPU's limitations take over.