Thief

FAQ

Q: How does the AMD Ryzen 5 9600X compare to its closest rivals in synthetic benchmarks?

A: The Ryzen 5 9600X posts an average benchmark score of 28246, placing it 0.3% ahead of the Intel Core i5-13490F and AMD Ryzen 5 230, while sitting 0.4% behind the AMD Ryzen 7 8845HS and 0.5% behind the Intel Core i5-14500T. This places the CPU in the 85th percentile of all tested processors.

Q: What is the GPU's performance percentile and how close are its nearest rivals?

A: The NVIDIA GeForce RTX 4070 sits in the 81st percentile of all GPUs with an average benchmark score of 37283. Its closest competitor, the AMD Radeon RX Vega 56, scores 37507, which is 0.6% higher. The NVIDIA GeForce MX570 A trails by 1.9%, and the RTX 4080 Mobile is 2.2% behind, while the AMD Radeon RX 5300M is 2.5% lower.

Q: What is the combo's overall ranking among tested system configurations?

A: This specific pairing of the AMD Ryzen 5 9600X with the NVIDIA GeForce RTX 4070 ranks 214th out of 1642 total tested combinations in Thief. That places it in the top 13% of all recorded system configurations for this title.

Q: How does the CPU's multi-threaded performance compare to its single-threaded results?

A: The CPU achieves a Cinebench R23 multicore score of 25525 versus a single-core score of 3603, showing strong scaling across its 6 cores and 12 threads. The 3DMark results reinforce this pattern, with the max-thread score of 7590 being roughly three times the single-thread score of 1253.

Q: What memory bandwidth does the platform provide for the GPU?

A: The CPU supports DDR5 memory with dual-channel configuration, offering a memory bandwidth of 89.6 GB/s. The GPU, meanwhile, has its own dedicated 12 GB of GDDR6X memory on a 192-bit bus, delivering 504.2 GB/s of bandwidth.

Q: Is the measured FPS data based on actual testing?

A: Yes, the dataIsMeasured field is marked true, confirming that all FPS figures come from direct benchmarking rather than estimation or simulation.

Settings Recommendations

The data reveals a clear tiered performance structure across the four settings presets. At 1920x1080, Low settings produce 223.9 FPS, Medium delivers 191.6 FPS, High reaches 145.4 FPS, and Ultra bottoms out at 140.8 FPS. The gap between High and Ultra is remarkably small—only 4.6 FPS—while the step from Medium to High costs 46.2 FPS. This suggests that Ultra provides nearly identical performance to High while presumably offering additional visual fidelity.

At 2560x1440, the pattern shifts. Low yields 174.1 FPS, Medium drops to 134.9 FPS, High manages 102.5 FPS, and Ultra falls to 95.5 FPS. Here the High-to-Ultra delta is 7 FPS, slightly larger than at 1080p but still modest. The Medium-to-High transition costs 32.4 FPS, which is proportionally similar to the 1080p result.

The 3840x2160 results tell a different story entirely. Low produces 104.4 FPS, Medium 78.4 FPS, High 61.9 FPS, and Ultra just 57.9 FPS. The High-to-Ultra gap shrinks to 4 FPS, while the Medium-to-High drop is 16.5 FPS. Given that Ultra never exceeds 57.9 FPS at 4K, and High maintains 61.9 FPS, the best experience for 4K users would be High settings—it stays above 60 FPS while the Ultra preset dips below that threshold. For 1440p and 1080p, Ultra is viable since even the lowest result of 95.5 FPS remains well above playable thresholds.

GPU Role

The RTX 4070's specifications directly shape these results. Its 5888 shading units, 184 texture mapping units, and 64 ROPs drive the raw rasterization throughput, while the boost clock of 2475 MHz and base clock of 1920 MHz determine how quickly those units operate. The GPU's FP32 performance of 29.15 TFLOPS provides the computational headroom needed for Thief's lighting and shadow calculations.

Memory configuration plays a critical role at higher resolutions. The 12 GB of GDDR6X memory with 504.2 GB/s bandwidth and 21 Gbps effective speed ensures that 4K textures and geometry don't cause memory swapping. The 192-bit bus width, while narrower than some higher-tier cards, still delivers sufficient bandwidth for this 2014 title, as evidenced by the 4K results remaining playable even at Ultra settings.

The GPU's 46 ray tracing cores and 184 tensor cores are present but largely irrelevant for Thief, which predates widespread RT implementation. The DirectX 12 Ultimate support and Vulkan 1.4 compatibility ensure modern driver paths are available, though the game's engine likely uses older API paths. The PassMark DirectX 11 score of 244 versus the DirectX 12 score of 103 hints that this GPU excels on legacy APIs, which aligns with Thief's era.

Measured FPS Breakdown

At 1920x1080, the performance envelope spans from 140.8 FPS at Ultra to 223.9 FPS at Low. The progression is: Low 223.9, Medium 191.6, High 145.4, Ultra 140.8. The drop from Low to Medium is 32.3 FPS, from Medium to High is 46.2 FPS, and High to Ultra is just 4.6 FPS. This clustering of High and Ultra suggests the Ultra preset adds minimal GPU load beyond High, possibly only minor shadow or reflection quality increases.

Moving to 2560x1440, the results are: Low 174.1, Medium 134.9, High 102.5, Ultra 95.5. The resolution increase from 1080p to 1440p reduces frame rates by 49.8 FPS at Low, 56.7 FPS at Medium, 42.9 FPS at High, and 45.3 FPS at Ultra. The larger absolute drops at lower settings indicate that fill-rate limitations become more pronounced as resolution increases, but the GPU still maintains triple-digit FPS through Medium settings.

At 3840x2160, the numbers compress significantly: Low 104.4, Medium 78.4, High 61.9, Ultra 57.9. The 4K resolution exacts a heavy toll—Low at 4K barely beats Ultra at 1440p (104.4 vs 95.5). The gap between Low and Ultra at 4K is 46.5 FPS, representing a 44.5% performance reduction from the lowest to highest quality preset. Notably, the High-to-Ultra delta of 4 FPS is the smallest across all resolutions, suggesting diminishing returns from the Ultra preset at 4K.

How This Combo Ranks

The 214th position out of 1642 tested combos places this configuration in the 87th percentile—meaning 87% of tested systems perform worse. This ranking reflects the balance between the Ryzen 5 9600X and RTX 4070, neither of which is a flagship component but both sit comfortably above mid-range territory.

The CPU's 85th percentile ranking and GPU's 81st percentile ranking align closely with the combo's overall position, indicating that neither component dramatically bottlenecks the other. The slight discrepancy—combo rank at 87th percentile versus CPU at 85th and GPU at 81st—suggests the pairing extracts slightly more performance than the individual component percentiles would predict, likely due to the game's moderate CPU demands allowing the GPU to stretch its legs.

Given that only 13% of tested combos outperform this setup, the system is well-positioned for high-refresh-rate gaming at 1080p and 1440p, while remaining capable of 4K at reduced settings. The data shows no obvious weakness in this pairing for Thief specifically.

CPU Role

The Ryzen 5 9600X's 6 cores and 12 threads, based on the Zen 5 Granite Ridge architecture, provide the processing backbone for this configuration. With a base clock of 3.90 GHz and boost clock of 5.40 GHz, the CPU can rapidly adjust to single-threaded demands—critical for a stealth game where AI calculations and physics simulations often run on a few heavily loaded threads.

The CPU's 32 MB of shared L3 cache, along with 1 MB L2 per core and 80 KB L1 per core, keeps frequently accessed game data close to the cores. The 3DMark results show strong scaling: 2456 at 2 threads, 4649 at 4 threads, 6726 at 8 threads, and 7590 at max threads. The relatively flat scaling from 8 to 12 threads (7590 vs 6726, a 12.8% increase) suggests diminishing returns beyond 8 threads, which is typical for games from the 2014 era.

The CPU's PassMark single-thread score of 4570 and multithread score of 30027 indicate robust per-core performance. For Thief, which likely relies heavily on 2-4 threads for game logic, the 3DMark 4-thread score of 4649 is particularly relevant—it shows the CPU can deliver strong performance in lightly threaded scenarios, complementing the GPU's output.

The 65W TDP and TSMC 4nm process node mean the CPU generates modest heat, allowing sustained boost clocks without thermal throttling during extended gaming sessions. The PCIe Gen 5 support with 24 lanes provides ample bandwidth for the GPU, though the RTX 4070 uses PCIe 4.0 x16, which is fully backward compatible.

Resolution Scaling

The FPS drops from 1080p to 1440p to 4K reveal the limiting component at each stage. From 1080p to 1440p, the average FPS reduction across all settings is approximately 30-35%, which aligns closely with the 1.78x increase in pixel count (1920x1080 = 2.07M pixels vs 2560x1440 = 3.69M pixels). This proportional scaling indicates the GPU is the primary constraint at these resolutions—the CPU has sufficient headroom to feed frames as fast as the GPU can render them.

However, from 1440p to 4K, the scaling becomes more aggressive. The pixel count increase from 3.69M to 8.29M (2.25x) should theoretically reduce FPS by roughly 55% if GPU-bound. The actual drops are: Low from 174.1 to 104.4 (40.1% reduction), Medium from 134.9 to 78.4 (41.9% reduction), High from 102.5 to 61.9 (39.6% reduction), and Ultra from 95.5 to 57.9 (39.4% reduction). These reductions are shallower than the pixel-count increase would predict, suggesting that at 4K, the CPU begins to contribute as a limiting factor, or that memory bandwidth constraints ease the scaling curve.

The 1080p to 4K cumulative drops are: Low 53.4%, Medium 59.1%, High 57.4%, and Ultra 58.9%. The consistency across settings—all falling within a 5.7 percentage point band—indicates that resolution scaling is largely independent of quality preset. This uniformity suggests the primary limiter at 4K is the GPU's fill rate and memory bandwidth rather than any specific rendering feature. The RTX 4070's 504.2 GB/s bandwidth, while substantial, becomes the bottleneck when pushing 8.29M pixels with complex shading, explaining why even Low settings at 4K (104.4 FPS) cannot match Medium at 1440p (134.9 FPS).