Thief

Thief, released in 2014, remains a demanding title for modern hardware, and the pairing of the AMD Ryzen 5 4600G with the NVIDIA GeForce RTX 5090 D presents a stark contrast in generational capability. The benchmark data for this combination, measured across three resolutions and four settings presets, reveals a system where the GPU is rarely the limiting factor, and the CPU’s older architecture constrains overall performance. The results indicate a bottleneck scenario where the Ryzen 5 4600G’s Zen 2 cores cannot fully feed the RTX 5090 D’s immense rendering throughput, particularly at lower resolutions and higher frame rates.

Resolution Scaling

The measured FPS data shows a clear, though non-linear, relationship between resolution and performance. At 1080p with Low settings, the system produces an average of 130 FPS. Moving to 1440p at the same settings drops the average to 115 FPS, a reduction of 15 frames. At 4K, the average falls further to 96 FPS. This scaling pattern—from 130 to 115 to 96—represents a total 26% drop from 1080p to 4K, which is relatively modest for a game of this era.

The pattern changes when examining higher settings. With High settings, the progression from 1080p (106 FPS) to 1440p (94 FPS) to 4K (79 FPS) shows a steeper decline, with the 4K result being 25% lower than the 1080p figure. The Ultra preset follows a similar trajectory, with 105 FPS at 1080p, 92 FPS at 1440p, and 77 FPS at 4K. The fact that the gap between 1080p and 1440p is consistently smaller than the gap between 1440p and 4K suggests that as pixel count increases, the GPU begins to take on more of the workload, though it never becomes fully saturated.

The critical observation is that the FPS deltas between resolutions are not as large as one would expect from a GPU as powerful as the RTX 5090 D. In a purely GPU-bound scenario, 4K performance would typically be less than half of 1080p performance. Here, the 4K Low result (96 FPS) is 74% of the 1080p Low result (130 FPS). This compression of the scaling curve is a classic symptom of a CPU bottleneck, where the processor cannot issue draw calls fast enough to keep the GPU busy at lower resolutions, thereby capping the maximum achievable frame rate.

GPU Role

The NVIDIA GeForce RTX 5090 D is an exceptional piece of hardware on paper, featuring 32 GB of GDDR7 memory on a 512-bit bus, delivering a bandwidth of 1.79 TB/s. Its boost clock of 2407 MHz and 104.8 TFLOPS of FP32 compute suggest it should obliterate a 2014 title like Thief. The GPU’s benchmark percentile of 92 and an average benchmark score of 77712 place it in the upper echelon of all tested graphics cards, with its nearest rivals including the AMD Radeon RX 6850M XT and the NVIDIA Tesla P100 PCIe variants.

However, in this specific game and with this specific CPU, the GPU’s role is diminished. The measured FPS data shows that even at 4K Ultra, the system manages 77 FPS, which is a playable frame rate but hardly a showcase for a card of this caliber. The GPU’s massive memory pool and bandwidth are largely irrelevant here, as Thief’s textures and geometry from 2014 are unlikely to stress even a fraction of the 32 GB VRAM capacity. The card is effectively operating at a fraction of its potential, held back by the data feed from the CPU.

The GPU’s performance data from passmark (G3D score of 44065) and 3DMark (Steel Nomad DX12 score of 14326) confirm its raw power. Yet, the in-game results show that the RTX 5090 D is not the deciding factor in this pairing. The consistent FPS figures across settings—for instance, the difference between High (106 FPS) and Ultra (105 FPS) at 1080p is just one frame—indicate that the GPU is processing the increased graphical load without any significant impact on the final output. The GPU is so far ahead of the game’s requirements that even the jump from Low to Ultra settings does not create a meaningful performance penalty, reinforcing that the processor is the primary constraint.

CPU Role

The AMD Ryzen 5 4600G is a 6-core, 12-thread processor based on the Zen 2 architecture, with a base clock of 3.70 GHz and a boost clock of 4.20 GHz. Its benchmark scores are modest by modern standards, with a Cinebench R23 multi-core score of 13593 and a single-core score of 1919. The processor’s percentileVsAllCpus is 71, meaning it outperforms 71% of all CPUs in the database, but its nearest rivals—such as the AMD Ryzen 3 PRO 5355GE and the Intel Core i3-14100T—are all lower-end or lower-power parts, indicating it sits in the mid-range tier.

The CPU’s role in this pairing is to act as the limiting factor. The evidence is in the FPS data: the scaling from 1080p to 4K is far too gentle for a GPU-bound scenario. At 1080p Low, the system hits 130 FPS, but at 4K Ultra, it still manages 77 FPS. If the CPU were capable of feeding the GPU adequately, we would expect the 1080p Low result to be significantly higher, potentially exceeding 200 FPS, while the 4K Ultra result would be much lower, perhaps in the 40-50 FPS range. Instead, the system’s top speed appears capped around 130 FPS, regardless of settings, which points to the CPU’s single-thread performance and memory bandwidth limitations.

The Ryzen 5 4600G’s 8 MB of shared L3 cache and dual-channel DDR4 memory support (with a theoretical bandwidth of 51.2 GB/s) are also factors. Thief, being a game from 2014, was designed for CPUs with far fewer cores and lower clock speeds. The 6-core, 12-thread configuration should be sufficient, but the per-core performance of Zen 2, combined with the relatively low boost clock of 4.20 GHz, means the CPU struggles to maintain high frame rates. The data shows that the CPU is the bottleneck because increasing the resolution from 1080p to 4K only reduces the frame rate by about 25-30%, rather than the 50-60% reduction that would occur if the GPU were the primary constraint.

How This Combo Ranks

The combination of the AMD Ryzen 5 4600G and the NVIDIA GeForce RTX 5090 D ranks 1094 out of 2851 tested combos for this game. This places it in the 62nd percentile of all combinations, meaning it outperforms roughly 61.6% of the systems tested. While this is a solid result, it is far below what the GPU alone would suggest. The RTX 5090 D’s percentileVsAllGpus is 92, indicating it is better than 92% of GPUs, but the combo’s rank is dragged down by the CPU.

The rank of 1094 out of 2851 indicates that this is not an optimal pairing. The data suggests that the CPU is holding back the GPU to such an extent that the overall system performance is closer to what one would expect from a mid-range GPU paired with this CPU, rather than a flagship GPU. For context, if the GPU were the sole determinant of the combo’s rank, we would expect it to be in the top 10% of all combos. Instead, it sits in the middle of the pack, demonstrating that the CPU is the dominant factor in determining this system’s in-game performance.

The nearest rival GPUs to the RTX 5090 D in terms of average benchmark score include the AMD Radeon RX 6650M XT (deltaPct 1.1) and the AMD Radeon RX 6850M XT (deltaPct -1.6). These are mobile-class or lower-tier desktop GPUs, and the fact that the combo rank is not significantly higher despite the RTX 5090 D’s massive advantage highlights the severity of the CPU bottleneck. The system is effectively leaving a large portion of the GPU’s performance untapped in this title.

FAQ

Q: Does the RTX 5090 D provide a significant performance boost over its rivals in this game?

A: No. The combo rank of 1094 out of 2851 suggests that the GPU’s power is not being fully utilized. The GPU’s nearest rivals have average benchmark scores within 2.4% of the RTX 5090 D, but the in-game FPS is capped by the CPU, so the GPU’s theoretical advantage does not translate to higher frame rates.

Q: Is the Ryzen 5 4600G a good match for the RTX 5090 D?

A: The data indicates it is not an ideal match. The CPU’s benchmark scores (e.g., Cinebench R23 multi-core of 13593) are adequate for general use, but the FPS scaling across resolutions suggests the CPU limits the system to around 130 FPS maximum, preventing the GPU from reaching higher frame rates.

Q: Why is the FPS difference between Low and Ultra settings so small at 1080p?

A: At 1080p, the CPU is the bottleneck. The average FPS for Low is 130, and for Ultra it is 105, a difference of only 25 FPS. This small gap shows that the GPU can handle the increased graphical workload without much penalty, but the CPU cannot push frame rates higher.

Q: Which resolution provides the most balanced experience?

A: Based on the data, 1440p offers a good balance. At High settings, it delivers 94 FPS, which is 12 FPS lower than 1080p but 15 FPS higher than 4K. The 1440p results show a more reasonable workload distribution, though the CPU still limits performance.

Q: Is the system capable of 4K gaming in Thief?

A: Yes, it is playable. At 4K Ultra, the average FPS is 77, and at 4K High, it is 79. These are above the 60 FPS threshold for smooth gameplay, but the system is not achieving the high refresh rates possible at lower resolutions.

Q: How does the CPU’s memory bandwidth affect performance?

A: The Ryzen 5 4600G has a dual-channel memory bus with a theoretical bandwidth of 51.2 GB/s. This is low compared to the GPU’s 1.79 TB/s, but for a CPU-bound game like Thief, the memory bandwidth can limit how quickly the CPU can access game data, contributing to the frame rate cap observed in the results.

Settings Recommendations

Given the CPU bottleneck, the optimal settings are those that maximize visual quality without sacrificing frame rate, since the CPU limits the top end regardless. The data shows that at 1080p Ultra, the system averages 105 FPS, which is only 1 FPS less than High (106 FPS) and 25 FPS less than Low (130 FPS). The jump from High to Ultra costs almost nothing in performance, so Ultra is the recommended preset for 1080p.

At 1440p, the story is similar. Ultra averages 92 FPS, while High averages 94 FPS. The 2 FPS difference is negligible, making Ultra the better choice for visual fidelity. At 4K, Ultra averages 77 FPS, and High averages 79 FPS. Again, the difference is minimal, so Ultra is recommended.

The Medium preset, which has min/max data (e.g., 100-135 FPS at 1080p), shows slightly higher average FPS than Ultra, but the difference is not significant enough to warrant lowering settings. The data suggests that the system can handle Ultra at all resolutions without a meaningful performance penalty, so the recommendation is to use the Ultra preset for the best visual experience, as the CPU bottleneck ensures that lower settings will not unlock substantially higher frame rates.

Measured FPS Breakdown

1920x1080 (1080p):

• Low: 130 FPS average
• Medium: 118 FPS average (min 100, max 135)
• High: 106 FPS average
• Ultra: 105 FPS average

2560x1440 (1440p):

• Low: 115 FPS average
• Medium: 104 FPS average (min 88, max 119)
• High: 94 FPS average
• Ultra: 92 FPS average

3840x2160 (4K):

• Low: 96 FPS average
• Medium: 87 FPS average (min 74, max 100)
• High: 79 FPS average
• Ultra: 77 FPS average

The data shows a consistent pattern: each resolution step down reduces performance by roughly 10-15% at the same settings, and each settings increase from Low to Ultra reduces performance by about 20-25% at the same resolution. The min FPS values for Medium settings (e.g., 74 FPS at 4K) indicate that the system maintains playable frame rates even in demanding scenes. The max FPS values (e.g., 135 FPS at 1080p Medium) show that the system can reach higher frame rates, but the average is capped by the CPU’s limitations. Overall, the system delivers a smooth experience across all tested configurations, but the potential of the RTX 5090 D is clearly not fully realized with the Ryzen 5 4600G.