Thief

Thief, a 2014 stealth title, presents a unique scaling scenario for the AMD Ryzen 5 7500X3D and NVIDIA GeForce RTX 5090 combination. The data indicates that this pairing is exceptionally powerful, delivering high frame rates across all resolutions, but the performance scaling reveals specific bottlenecks. The system demonstrates that while the RTX 5090 is the dominant force at 4K, the Ryzen 5 7500X3D's capabilities become increasingly relevant as resolution drops, showcasing a balanced, high-end configuration.

Resolution Scaling

The measured FPS data shows a clear and consistent pattern of performance degradation as resolution increases from 1080p to 4K. At 1080p with Low settings, the system achieves an average of 172 FPS. Moving to 1440p Low settings, the average drops to 151 FPS, a reduction of 21 FPS. At 4K Low settings, the average falls further to 127 FPS. This scaling pattern across the Low preset indicates that the GPU is becoming more heavily utilized at higher resolutions, which is typical behavior.

However, the scaling becomes more nuanced when examining the High and Ultra presets. At the High preset, the frame rates are 141 FPS at 1080p, 124 FPS at 1440p, and 86 FPS at 4K. The transition from 1080p to 1440p results in a 17 FPS drop, while the transition from 1440p to 4K results in a larger 38 FPS drop. This suggests that the game's graphical workload at High settings scales superlinearly with pixel count, placing a significant burden on the GPU at 4K.

The Ultra preset exhibits the most pronounced scaling impact. At 1080p, the system manages 138 FPS, which drops to 122 FPS at 1440p and then plummets to 82 FPS at 4K. The 40 FPS difference between 1440p and 4K Ultra is the largest single-resolution gap in the entire dataset. This indicates that Ultra settings introduce graphical effects that are particularly demanding on the GPU's compute and memory resources, making 4K Ultra the most challenging scenario for this hardware combination.

The data suggests that the limiting component shifts depending on the resolution. At 1080p and 1440p, the frame rates are high enough that the CPU, with its 6 cores and 12 threads, is likely providing sufficient instruction throughput to keep the GPU fed, but the GPU is still the primary determinant of performance. At 4K, the RTX 5090's role becomes absolute, as the sheer pixel count and shading workload overwhelm the CPU's ability to be the limiting factor. The 4K results, particularly the 82 FPS average at Ultra, show that this is a GPU-bound scenario, with the RTX 5090's 32 GB of GDDR7 memory and 1.79 TB/s bandwidth being fully exercised.

GPU Role

The NVIDIA GeForce RTX 5090 is the undisputed performance leader in this configuration, and its specifications dictate the results at higher resolutions. The GPU features a boost clock of 2407 MHz, a base clock of 2017 MHz, and a massive 32 GB of GDDR7 memory on a 512-bit bus, providing 1.79 TB/s of bandwidth. This memory configuration is crucial for 4K gaming, as high-resolution textures and effects require substantial memory capacity and bandwidth to avoid stuttering or frame drops.

Benchmark results show the RTX 5090's raw compute power. It achieves a PassMark G3D score of 39650, placing it in the 92nd percentile among all GPUs. Its average benchmark score is 79842, which puts it in close competition with the NVIDIA Tesla P100 PCIe 16 GB, which scores 79605 (a deltaPct of 0.3%), and the NVIDIA Tesla P100 PCIe 12 GB, which scores 79396 (a deltaPct of 0.6%). The RTX 5090 is also 1.1% ahead of the AMD Radeon RX 6850M XT and 1.4% behind the AMD Radeon Pro Vega 64X. These rival scores, while from different market segments, illustrate that the RTX 5090's overall compute capability is at the top tier of the benchmark database.

In Thief specifically, the GPU's role is most evident in the 4K results. The difference between 4K Low (127 FPS) and 4K Ultra (82 FPS) is 45 FPS, a 35% performance penalty for enabling the highest settings. This is a direct consequence of the increased shading, texture, and post-processing demands. The GPU's 104.8 TFLOPS of FP32 compute is the engine driving these frames, and the data shows it is fully capable of handling this 2014 title at modern resolutions, even with the most demanding settings. The GPU's 176 ROPs and 680 TMUs are also critical for pixel throughput and texture filtering, respectively, which are heavily utilized in Thief's detailed environments.

CPU Role

The AMD Ryzen 5 7500X3D provides the processing foundation for this system, and its performance characteristics become more visible at lower resolutions. This CPU features 6 cores and 12 threads, with a base clock of 4.00 GHz and a boost clock of 4.50 GHz. It has a substantial 96 MB of shared L3 cache, which is a key attribute for gaming performance. Its average benchmark score is 44573, placing it in the 89th percentile among all CPUs.

The CPU's benchmark scores show its strengths and weaknesses. Its PassMark Single Thread score is 3494, which is a solid result for gaming. Its PassMark Multi Thread score is 25047, indicating strong multi-core performance. The nearest rivals in the benchmark database include the Intel Core Ultra X9 388H, which scores 44466 (a deltaPct of 0.2% slower), the Intel Core i9-13950HX, which scores 44342 (a deltaPct of 0.5% slower), and the AMD Ryzen AI Max 385, which scores 44309 (a deltaPct of 0.6% slower). The Intel Core i5-14600 is 0.7% faster with a score of 44889. These results show the 7500X3D is competitive with high-end laptop and desktop chips, but its advantage lies in its cache architecture rather than raw clock speed.

In Thief, the CPU's role is evident in the 1080p results. At 1080p Low, the system achieves 172 FPS, which is the highest frame rate in the dataset. This suggests that the CPU is capable of feeding the GPU with enough data to reach this high frame rate. However, the scaling from 1080p to 1440p at Low settings (172 to 151 FPS) shows that the GPU's workload is increasing, but the CPU is not the primary bottleneck. At 1080p Ultra (138 FPS) versus 1080p Low (172 FPS), the 34 FPS difference is due to the increased GPU workload. The CPU's 4.50 GHz boost clock and 96 MB L3 cache help maintain high minimum frame rates, ensuring a smooth experience even when the GPU is under heavy load. The data indicates that the 7500X3D is sufficient to prevent CPU-related stutters in this title, allowing the RTX 5090 to perform at its full potential.

FAQ

Q: What is the maximum average FPS achievable at 1080p with this hardware?

A: The highest average FPS recorded is 172 FPS, achieved at 1080p with Low settings. This indicates the system is capable of extremely high frame rates at the lowest resolution tested.

Q: How does the RTX 5090's overall benchmark score compare to its nearest rivals?

A: The RTX 5090's average benchmark score is 79842. This makes it 0.3% faster than the NVIDIA Tesla P100 PCIe 16 GB (79605), 0.6% faster than the NVIDIA Tesla P100 PCIe 12 GB (79396), and 1.1% faster than the AMD Radeon RX 6850M XT. It is 1.4% slower than the AMD Radeon Pro Vega 64X, which scores 80959.

Q: What is the FPS difference between Low and Ultra settings at 4K?

A: At 4K, the average FPS drops from 127 FPS on Low settings to 82 FPS on Ultra settings. This represents a 45 FPS decrease, showing the significant performance cost of enabling the highest visual quality.

Q: What are the key specifications of the AMD Ryzen 5 7500X3D's cache?

A: The CPU has an L1 cache of 64 KB per core, an L2 cache of 1 MB per core, and a large L3 cache of 96 MB shared across all cores. This large L3 cache is a defining feature of the X3D series.

Q: Which resolution and settings combination provides the highest minimum FPS?

A: The data shows that the 1080p Medium preset has a minimum FPS of 132. The 1440p Medium preset has a minimum FPS of 116, and the 4K Medium preset has a minimum FPS of 96. The 1080p Medium setting provides the highest recorded minimum FPS.

Q: Does the data suggest this system is more CPU-bound or GPU-bound in Thief?

A: The data suggests the system is primarily GPU-bound, especially at 4K where the frame rates drop significantly with higher settings. At 1080p, the high frame rates (up to 172 FPS) indicate the CPU is keeping up, but the performance scaling is still dictated by the GPU's workload.

Measured FPS Breakdown

The measured FPS data provides a comprehensive look at performance across three resolutions and four settings presets. All values are average FPS unless otherwise noted.

At 1920x1080, the system delivers its highest performance. The Low preset achieves an average of 172 FPS. The Medium preset is next, with an average of 155 FPS and a range of 132 to 179 FPS. The High preset achieves an average of 141 FPS, and the Ultra preset achieves an average of 138 FPS. The difference between Low and Ultra at 1080p is 34 FPS, which is the smallest gap between the two settings across all resolutions. This indicates that at 1080p, the CPU is capable of maintaining high frame rates even with the most demanding settings, and the GPU's workload is the limiting factor.

At 2560x1440, performance remains high but shows a clear step down from 1080p. The Low preset averages 151 FPS. The Medium preset averages 137 FPS, with a range of 116 to 157 FPS. The High preset averages 124 FPS, and the Ultra preset averages 122 FPS. The gap between Low and Ultra at 1440p is 29 FPS. Notably, the difference between High (124 FPS) and Ultra (122 FPS) is minimal, suggesting that the Ultra preset's additional effects have a marginal impact at this resolution compared to the jump from Medium to High.

At 3840x2160, the GPU is pushed to its limits. The Low preset averages 127 FPS. The Medium preset averages 113 FPS, with a range of 96 to 130 FPS. The High preset averages 86 FPS, and the Ultra preset averages 82 FPS. The gap between Low and Ultra at 4K is 45 FPS, the largest in the dataset. The 4K results show a steep drop-off from Medium to High (27 FPS), indicating that the High preset introduces a significant graphical load at this resolution. The Ultra preset's 82 FPS average is still very playable, but it is the only setting where the average drops below 100 FPS.

Settings Recommendations

Based on the measured data, the optimal settings for this hardware depend on the user's priority for frame rate versus visual fidelity. The data shows that the system can handle any settings combination, but the most balanced experience is achieved with a specific approach.

For users with a 1080p display, the High preset (141 FPS) is the recommended choice. It offers a significant visual improvement over Medium (155 FPS) with only a 14 FPS penalty, and it is only 3 FPS slower than Ultra (138 FPS). The Ultra preset provides negligible visual gains over High for this title, making High the most efficient choice for maintaining a very high refresh rate.

For 1440p displays, the Medium preset (137 FPS) is the sweet spot. It provides a smooth experience with a minimum FPS of 116. The High preset (124 FPS) is also viable, but the 13 FPS drop from Medium to High is more noticeable at this resolution. The Ultra preset (122 FPS) offers almost no advantage over High, so it is not recommended. The Medium preset delivers a high frame rate while maintaining good visual quality.

For 4K displays, the Medium preset (113 FPS) is the recommended choice. It is the only preset at 4K that maintains an average above 100 FPS, and its minimum FPS of 96 ensures a mostly smooth experience. The Low preset (127 FPS) is faster but sacrifices visual detail. The High preset (86 FPS) and Ultra preset (82 FPS) are playable but represent a significant performance drop for incremental visual gains. The data suggests that the Medium preset at 4K is the best compromise, leveraging the RTX 5090's power without pushing the game into less stable frame rate territory.