Thief

The Intel Core Ultra X7 368H paired with the NVIDIA GeForce RTX 5090 delivers a striking performance curve in Thief, where the frame rate scales from a highly playable 147 FPS at 1080p Ultra down to 59 FPS at 4K Ultra. The data indicates a system that is overwhelmingly GPU-limited at higher resolutions, yet still retains enough headroom to push well past refresh-rate territory at 1080p. This combination ranks in the 87th percentile for CPUs and 92nd percentile for GPUs, placing it in the upper echelon of tested hardware, though its in-game rank of 470 out of 3720 combos shows that the specific pairing has distinct characteristics.

Resolution Scaling

The measured frames per second reveal a clear and predictable pattern of degradation as resolution increases, but the rate of that drop is telling. At 1080p with Low settings, the combo produces 204 FPS, which falls to 180 FPS at 1440p and then plummets to 108 FPS at 4K. This represents a 47% reduction in performance from 1080p to 4K at Low settings, a steep decline that points directly at the GPU as the primary bottleneck once pixel count rises.

The scaling at Ultra settings is even more instructive. The system achieves 147 FPS at 1080p, 101 FPS at 1440p, and only 59 FPS at 4K. The gap between 1080p and 1440p is 31%, while the gap between 1440p and 4K is 42%. This accelerating drop-off is classic behavior for a GPU that is being saturated by the 4K resolution, especially given the RTX 5090’s massive 32 GB frame buffer and 1.79 TB/s memory bandwidth, which should theoretically handle the data throughput but is still constrained by raw shading power at 4K.

The 1080p to 4K transition at Medium settings shows a similar story: 185 FPS drops to 139 FPS at 1440p and then to 82 FPS at 4K. The 1080p numbers are so high that they suggest the CPU is nearing its limit, with the Core Ultra X7 368H’s 16 cores and 16 threads potentially becoming a factor at lower resolutions where the GPU is not fully taxed. The data shows that at 4K, the average FPS across all settings stays within a narrow band of 59 to 108 FPS, indicating that resolution, not settings, is the dominant variable at that pixel count.

What this scaling tells us is that the limiting component at 1080p and 1440p is a mix of CPU and GPU, but at 4K, the RTX 5090 is unequivocally the choke point. The 1080p Low result of 204 FPS versus the 4K Low result of 108 FPS shows that the GPU’s fill rate and compute are being pushed to their absolute limits by the 3840x2160 resolution, and no amount of CPU headroom can compensate for that geometric increase in pixel workload.

GPU Role

The NVIDIA GeForce RTX 5090 is a monster of a graphics card on paper, and the benchmark results in Thief reflect its dominance, though not without nuance. With a boost clock of 2407 MHz and a base clock of 2017 MHz, the card’s raw frequency is modest compared to its massive 104.8 TFLOPS of FP32 compute, which is the real driver behind its performance. The 32 GB of GDDR7 memory on a 512-bit bus delivers 1.79 TB/s of bandwidth, and this is where the data shows its impact: even at 4K Ultra, the card maintains a 59 FPS average, suggesting that memory bandwidth is not the limiting factor, but rather the 21760 shading units and 176 ROPs are being saturated.

The GPU’s benchmark scores contextualize its in-game behavior. It posts a Passmark G3D score of 39650 and a 3DMark Steel Nomad DX12 score of 18355, with a Passmark GPU Compute score of 26756. These numbers place it in the 92nd percentile of all GPUs, with nearest rivals including the AMD Radeon RX 6850M XT (1.1% slower) and the AMD Radeon Pro Vega 64X (1.4% faster). The RTX 5090’s average benchmark score of 79842 is 1.1% higher than the RX 6850M XT, yet in Thief, the 4K High result of 62 FPS is only marginally above the 4K Medium result of 82 FPS, indicating that the GPU is not scaling efficiently with settings at this resolution.

The VRAM is a non-issue here; 32 GB is far more than Thief requires, given its 2014 release date. The clocks are more relevant: the boost clock of 2407 MHz is what sustains the 4K performance, and the data shows that dropping from High to Medium at 4K only yields a 24% FPS increase (62 to 82 FPS), while dropping from Medium to Low yields a 24% increase as well (82 to 108 FPS). This linear scaling suggests the GPU is compute-bound, not bandwidth-bound, and that the 176 ROPs are the limiting factor for pixel output at 4K.

CPU Role

The Intel Core Ultra X7 368H is a mobile processor with 16 cores and 16 threads, a configuration that favors multi-threaded workloads but lacks simultaneous multithreading. Its base clock of 2.00 GHz and boost clock of 5.00 GHz provide a wide dynamic range, and the 18 MB of shared L3 cache helps feed the cores. In Thief, this CPU shows its strength at lower resolutions: the 1080p Low result of 204 FPS is likely near the engine’s single-thread limit, as the game from 2014 does not scale perfectly across 16 cores.

The CPU’s benchmark scores reveal its position in the market. It achieves a Cinebench R23 multi-core score of 28221 and a single-core score of 3984, with a Passmark multithread score of 32956 and a single-thread score of 4005. These figures place it in the 87th percentile of all CPUs, and its nearest rival, the Intel Core 7 253PE, is only 0.1% faster on average benchmark score. The data shows that the CPU is not the bottleneck at 4K, where the GPU is clearly saturated, but at 1080p, the 147 FPS Ultra result versus 204 FPS Low result indicates that the CPU can keep up with the GPU’s output, but the gap between Low and Ultra at 1080p (28%) is smaller than at 4K (45%), suggesting the CPU is starting to limit the frame rate at the lower resolution.

The 16 threads are a double-edged sword: they provide strong multi-core performance for modern titles, but Thief’s engine likely relies on fewer threads, meaning the 5.00 GHz boost clock is more critical than the core count. The data shows that at 1440p, the CPU still has headroom, as the 1440p Ultra result of 101 FPS is not dramatically lower than the 1080p Ultra result of 147 FPS, indicating that the GPU is taking over the workload. The CPU’s 153.6 GB/s memory bandwidth and dual-channel LPDDR5X support are sufficient for this game, and the data does not indicate any memory-related stuttering or frame drops, as the minimum FPS values are only recorded for Medium settings and show a 15-20% drop from average, which is typical.

Settings Recommendations

Based on the measured rows, the optimal settings for this combo depend on the target resolution. At 1080p, the Ultra preset delivers 147 FPS, which is more than enough for a smooth experience on a 144Hz display, and the drop to High only improves this to 154 FPS, a negligible 5% gain. The Medium preset at 1080p yields 185 FPS with a minimum of 157 FPS, making it the best choice for high-refresh-rate monitors, but the visual fidelity loss may not be worth it given that Ultra is already playable.

At 1440p, the High preset offers 106 FPS, which is a sweet spot for most users, while Ultra drops to 101 FPS, a 5% reduction that is unlikely to be noticeable in a stealth game. The Medium preset at 1440p provides 139 FPS with a minimum of 118 FPS, which is the best option for competitive play, but for a single-player stealth title, the visual quality of High is preferable. The Low preset at 1440p produces 180 FPS, but this is overkill and sacrifices too much visual detail.

At 4K, the data suggests that High is the best balance, with 62 FPS average, but Ultra drops to 59 FPS, which is borderline for smooth gameplay. The Medium preset at 4K yields 82 FPS with a minimum of 70 FPS, which is the most consistent experience, offering a 32% improvement over High while retaining acceptable visual quality. The Low preset at 4K produces 108 FPS, but the graphical compromises are severe. For a 4K display, the Medium preset is recommended, as it provides the best balance of frame rate and fidelity per the measured rows, with the minimum FPS of 70 ensuring no major stutters.

How This Combo Ranks

This specific combination of the Intel Core Ultra X7 368H and NVIDIA GeForce RTX 5090 ranks 470th out of 3720 tested combos in Thief, placing it in the top 12.6% of all configurations. This rank is lower than the individual component percentiles (87th for CPU, 92nd for GPU) would suggest, indicating that the pairing is not perfectly optimized for this game. The data shows that the GPU is the dominant factor in the ranking, as the RTX 5090’s nearest rivals in the GPU benchmark averages (RX 6850M XT at 1.1% slower, Pro Vega 64X at 1.4% faster) do not directly translate to in-game performance, where the 4K results are only moderate.

The combo’s rank is likely dragged down by the CPU’s 16-thread configuration, which may not be ideal for Thief’s older engine, and the fact that the RTX 5090’s massive compute power is not fully utilized at the game’s moderate graphical demands. The 1080p results, which are very high, suggest that the CPU is not a major bottleneck, but the 4K results, which are comparatively lower, show that the GPU is not scaling as well as its benchmark scores would imply. Compared to other combos, this pairing is strong but not top-tier, and the data indicates that a CPU with higher single-thread performance or a GPU with better 4K optimization would rank higher.

FAQ

*Q: What is the best resolution to play Thief with this combo?*

A: The data shows that 1440p High provides 106 FPS, which is a strong balance of quality and performance, while 4K High drops to 62 FPS. For a consistent 60+ FPS experience, 1440p is the sweet spot.

*Q: How does the RTX 5090’s 32 GB VRAM affect Thief performance?*

A: The 32 GB frame buffer is far more than Thief requires, and the data shows that the GPU is not memory-limited at any resolution, as the 4K Low result of 108 FPS is not constrained by VRAM capacity.

Q: Is the CPU a bottleneck at 4K?

A: No, the data indicates the opposite. At 4K, the GPU is the limiting factor, as the FPS drops from 108 at Low to 59 at Ultra, while the CPU’s 16 cores and 5.00 GHz boost clock provide ample headroom.

Q: What is the difference between 1080p and 4K performance at Ultra settings?

A: The Ultra preset yields 147 FPS at 1080p and 59 FPS at 4K, a 60% reduction. This steep drop is attributed to the GPU’s pixel throughput limits at 4K.

Q: How does this combo rank compared to other tested systems?

A: It ranks 470th out of 3720 combos, placing it in the top 12.6%. The CPU and GPU individually rank in the 87th and 92nd percentiles, respectively, but the in-game rank is lower due to the game’s specific engine characteristics.

Q: Does the Medium preset offer better frame consistency than High?

A: Yes, at 4K Medium, the average FPS is 82 with a minimum of 70, whereas at 4K High, the average is 62 with no recorded minimum. The Medium preset provides a more stable frame rate.