Thief

# Thief — Intel Core Ultra 7 155H + NVIDIA GeForce RTX 5050 Mobile

The measured frame rates for this combination in Thief show a system that is firmly GPU-limited at higher resolutions, with the RTX 5050 Mobile's 8 GB GDDR7 memory and 384.0 GB/s bandwidth acting as the primary constraint. Across the three tested resolutions (1920x1080, 2560x1440, 3840x2160) and four quality presets (Low, Medium, High, Ultra), the average FPS ranges from 30 to 135, with the largest performance cliff occurring between 1440p and 4K. The data indicates a 2014 stealth title that scales predictably with resolution but shows diminishing returns at Ultra settings, where the GPU's 2560 shading units and 32 ROPs become the bottleneck.

Resolution Scaling

The FPS drop from 1080p to 4K is substantial and reveals the limiting component clearly. At High settings, the average frame rate falls from 78 FPS at 1920x1080 to 54 FPS at 2560x1440 (a 30.8% reduction), then further to 32 FPS at 3840x2160 (a 40.7% reduction from 1440p). This compounding decline indicates that the RTX 5050 Mobile's pixel throughput — rated at 48.00 GPixel/s — is the dominant factor at higher resolutions.

Low settings show a similar pattern but with steeper absolute drops. The average FPS at 1080p Low is 135, which falls to 93 at 1440p (31.1% lower) and then to 55 at 4K (40.9% lower than 1440p). The consistency of these percentage drops across presets suggests that resolution scaling is largely independent of quality settings, pointing to a fixed GPU bottleneck rather than a memory or CPU limitation.

Medium settings, which include minimum and maximum FPS data, provide additional insight. At 1080p Medium, the average is 103 FPS with a min of 87 and max of 118. Moving to 1440p Medium drops the average to 71 FPS (31.1% lower), with min/max of 60/81. At 4K Medium, the average collapses to 41 FPS (42.3% lower than 1440p), with min/max of 35/48. The widening gap between min and max FPS at higher resolutions — from a 31 FPS spread at 1080p to a 13 FPS spread at 4K — suggests that frame pacing becomes more consistent but at a much lower performance level.

Ultra settings are the most demanding, with average FPS of 75 at 1080p, 51 at 1440p, and 30 at 4K. The 1080p-to-1440p drop is 32.0%, while the 1440p-to-4K drop is 41.2%. These percentages align almost exactly with the other presets, confirming that resolution scaling is the primary variable affecting performance, with no evidence of a CPU limitation at any resolution.

The data shows that the RTX 5050 Mobile, with its 1500 MHz boost clock and 7.680 TFLOPS FP32 throughput, is designed for 1080p and 1440p gaming. At 4K, even Low settings only achieve 55 FPS average, which falls below the 60 FPS threshold commonly desired for smooth gameplay. The 3840x2160 Ultra result of 30 FPS average is essentially a slideshow for a stealth game that requires precise timing.

Settings Recommendations

Based on the measured rows, the optimal balance of visual quality and performance is achieved at Medium settings for most resolutions. At 1920x1080, Medium delivers 103 FPS average with a minimum of 87 FPS, which is comfortably above the playable threshold while providing better visual fidelity than Low. The jump from Low to Medium costs 32 FPS (from 135 to 103), but the visual improvement is generally worth the trade-off in a stealth game where lighting and shadow detail matter for gameplay.

At 2560x1440, Medium is again the sweet spot, with 71 FPS average and a minimum of 60 FPS. This is the lowest resolution at which the minimum FPS stays at or above 60, making it the highest quality preset that guarantees smooth gameplay. High settings at 1440p average 54 FPS, which is playable but may experience noticeable stutter in busy scenes, while Ultra at 51 FPS average pushes the minimum below comfortable thresholds.

For 3840x2160, no preset achieves a 60 FPS average. Low settings provide 55 FPS average, which is the closest to playable, but Medium drops to 41 FPS and High/Ultra fall to 32/30 FPS respectively. The data suggests that 4K is not a viable resolution for this game with this GPU combination unless the user accepts significant compromises in visual quality. Low settings at 4K are the only option that approaches playability, and even then, the minimum FPS is likely to dip during action sequences.

The Ultra preset offers negligible benefits over High in terms of measured frames — at 1080p, Ultra averages 75 FPS versus 78 FPS for High, a mere 3.8% difference. At 1440p, the gap is 51 FPS versus 54 FPS (5.6%), and at 4K it is 30 FPS versus 32 FPS (6.3%). Given that Ultra typically adds resource-intensive effects like higher draw distances and post-processing, the marginal FPS gain from dropping to High suggests that High is the more efficient choice for users who want the best visual quality without paying a significant performance penalty.

GPU Role

The NVIDIA GeForce RTX 5050 Mobile is built on the Blackwell 2.0 architecture with a GB207 chip fabricated on a 5 nm process at TSMC. The GPU contains 16,900 million transistors on a 149 mm² die, with 2560 shading units, 80 TMUs, and 32 ROPs. These specifications are modest for a 50-series mobile part, but the memory subsystem is notably strong for the class: 8 GB of GDDR7 on a 128-bit bus delivers 384.0 GB/s of bandwidth at 1500 MHz (24 Gbps effective).

In Thief, a game released in 2014, the GPU's raw compute capabilities are more than sufficient for the game's rendering demands. The RTX 5050 Mobile's 7.680 TFLOPS FP32 throughput and 120.0 GTexel/s texture rate are far beyond what a 2014-era flagship GPU offered. However, the game's performance is constrained by the GPU's pixel fill rate (48.00 GPixel/s) and memory bandwidth at higher resolutions, as evidenced by the consistent FPS drops when moving from 1440p to 4K.

The 8 GB memory capacity is not a limiting factor in Thief, as the game's texture requirements are modest by modern standards. However, the 384.0 GB/s bandwidth is likely the key constraint at 4K, where the ROPs must write more pixels per frame. The measured FPS at 4K (30-55 across presets) aligns with what a 32-ROP GPU with this bandwidth would typically achieve in a 2014 title.

The GPU's percentile ranking of 83 among all GPUs puts it near the top quartile, but its nearest rivals include the NVIDIA Quadro M6000 24 GB (deltaPct 0), Quadro M6000 (deltaPct -0.1), and RTX 4070 SUPER (deltaPct 0.1). The RTX 5050 Mobile's average benchmark score of 43268 is nearly identical to these desktop-class parts, yet its mobile power envelope of 50 W TDP means it delivers this performance in a much more thermally constrained environment. In Thief specifically, the GPU's 1500 MHz boost clock and 1020 MHz base clock are sufficient to maintain high frame rates at 1080p, but the 4K results suggest the clock speeds cannot compensate for the memory bandwidth and ROP limitations.

FAQ

Q: What is the best resolution to play Thief with this GPU/CPU combo?

A: Based on the measured data, 1920x1080 is the only resolution where all four presets maintain at least 75 FPS average. At 2560x1440, only Low and Medium presets exceed 60 FPS average. At 3840x2160, no preset reaches 60 FPS average, with even Low settings capped at 55 FPS.

Q: How much FPS does the Ultra preset cost compared to High?

A: At 1920x1080, Ultra averages 75 FPS versus 78 FPS for High, a 3.8% penalty. At 2560x1440, the gap is 51 FPS versus 54 FPS (5.6%). At 3840x2160, Ultra averages 30 FPS versus 32 FPS for High, a 6.3% penalty.

Q: Does the 8 GB VRAM limit performance in Thief?

A: The measured data does not indicate VRAM capacity as a bottleneck. The performance drops are consistent with GPU compute and bandwidth limitations rather than memory exhaustion, as Thief's 2014-era textures fit comfortably within 8 GB GDDR7.

Q: What minimum FPS can be expected at 1440p Medium?

A: The measured data shows a minimum of 60 FPS at 2560x1440 Medium, with an average of 71 FPS and a maximum of 81 FPS. This is the only 1440p preset with a documented minimum at or above 60 FPS.

Q: How does this combo rank among all tested combinations in Thief?

A: This combination ranks 2982 out of 3720 tested combos, placing it in the 19.8th percentile of all systems tested for this game. This low ranking reflects the RTX 5050 Mobile's position as an entry-level 50-series part rather than any CPU deficiency.

Q: Is the CPU or GPU the limiting factor at 1080p?

A: The data shows that even at 1080p Ultra (75 FPS), the GPU is the limiting component. The Intel Core Ultra 7 155H, with 16 cores and 22 threads, has a PassMark multithread score of 24873, which is more than adequate for a 2014 title. The FPS scaling from Low to Ultra at 1080p (135 to 75) is consistent with GPU-bound behavior.

How This Combo Ranks

The Intel Core Ultra 7 155H + NVIDIA GeForce RTX 5050 Mobile combination ranks 2982 out of 3720 tested combos in Thief, placing it in the bottom 19.8% of all systems. This low ranking is surprising given the individual component percentiles — the CPU ranks at the 83rd percentile and the GPU also ranks at the 83rd percentile. The discrepancy suggests that Thief's performance profile favors GPUs with higher memory bandwidth and ROP counts rather than modern architecture features.

The CPU's nearest rivals provide context for its performance class. The Intel Core i5-14600T has an average benchmark score of 32707 (deltaPct 0), while the AMD Ryzen AI 7 PRO 360 scores 32662 (deltaPct 0.1), the AMD Ryzen 5 7400F scores 32750 (deltaPct -0.2), and the AMD Ryzen 7 PRO 6850H scores 32812 (deltaPct -0.4). The Core Ultra 7 155H's average score of 32697 places it in a tight cluster, indicating that CPU choice among these rivals would have minimal impact on Thief's frame rates.

The GPU's nearest rivals are more diverse. The NVIDIA Quadro M6000 24 GB scores 43262 (deltaPct 0), the Quadro M6000 scores 43301 (deltaPct -0.1), the RTX 4070 SUPER scores 43223 (deltaPct 0.1), and the RTX 4090 Mobile scores 43667 (deltaPct -0.9). Despite the RTX 5050 Mobile's similar average benchmark score, its 8 GB VRAM and 128-bit bus place it at a disadvantage in memory-heavy scenarios, which may explain its low ranking in Thief specifically.

The rank of 2982 out of 3720 indicates that most tested combos achieve higher frame rates in Thief. This is likely because Thief is an older title that does not utilize modern GPU features like ray tracing or DLSS, so the RTX 5050 Mobile's Blackwell 2.0 architecture provides no advantage over older, higher-bandwidth GPUs. The 4K results, where even Low settings only average 55 FPS, are particularly weak compared to combos with larger memory buses or higher ROP counts.

CPU Role

The Intel Core Ultra 7 155H is a 16-core, 22-thread mobile processor based on the Meteor Lake architecture, manufactured on Intel's 7 nm process. It features a base clock of 3.80 GHz and a boost clock of 4.80 GHz, with a 24 MB shared L3 cache and 112 KB L1 cache per core. The CPU's 28 W TDP is typical for a mobile part, and its PassMark multithread score of 24873 places it at the 83rd percentile of all CPUs.

In Thief, the CPU's role is minimal due to the game's age and the GPU-bound nature of the workloads. The measured FPS data shows no evidence of CPU throttling at any resolution or preset. At 1080p Low, where the GPU has the least work per frame, the average FPS of 135 is still below what the CPU could theoretically support, given its 22 threads and 4.80 GHz boost capability.

The CPU's single-thread performance, with a PassMark single-thread score of 3468 and a Cinebench R23 single-core score of 1743, is more than sufficient for a 2014 game. Thief's AI, physics, and game logic are all single-threaded in practice, and the Core Ultra 7 155H's high boost clock ensures that these tasks complete quickly. The Cinebench R15 single-core score of 255 and R20 single-core score of 1235 confirm that the CPU's per-core performance is competitive with desktop parts from the same era.

The CPU's memory bandwidth of 89.6 GB/s and dual-channel DDR5 support are also not limiting factors in Thief. The game's data streaming requirements are modest, and the 24 MB L3 cache helps reduce memory latency. The CPU's integrated Arc Xe-LPG 128EU graphics are irrelevant for this test since the discrete RTX 5050 Mobile handles all rendering.

The nearest CPU rivals all have similar average benchmark scores (within 0.4% of the Core Ultra 7 155H), which means that swapping in a different CPU from this performance tier would have no measurable impact on Thief's frame rates. The CPU's 83rd percentile ranking is misleading in this context because the game is overwhelmingly GPU-limited, as evidenced by the consistent FPS scaling with resolution changes.

Measured FPS Breakdown

At 1920x1080, the full preset range was measured. Low settings achieve an average of 135 FPS, which is the highest frame rate measured in this test. Medium settings average 103 FPS with a minimum of 87 FPS and a maximum of 118 FPS. High settings average 78 FPS, and Ultra settings average 75 FPS. The spread from Low to Ultra is 60 FPS, indicating that the GPU can handle 1080p with significant headroom even at maximum quality.

At 2560x1440, the performance drops are consistent across presets. Low settings average 93 FPS, Medium averages 71 FPS with a minimum of 60 FPS and a maximum of 81 FPS, High averages 54 FPS, and Ultra averages 51 FPS. The spread from Low to Ultra is 42 FPS, showing that the GPU is beginning to become the bottleneck at this resolution, particularly at higher quality settings.

At 3840x2160, the performance drops are severe. Low settings average 55 FPS, Medium averages 41 FPS with a minimum of 35 FPS and a maximum of 48 FPS, High averages 32 FPS, and Ultra averages 30 FPS. The spread from Low to Ultra is only 25 FPS, indicating that the GPU's pixel throughput and memory bandwidth are saturated even at Low settings. The 4K Medium minimum of 35 FPS and maximum of 48 FPS show significant frame time variance, which would be noticeable in gameplay.

Comparing presets across resolutions, the percentage differences are remarkably stable. Low settings drop 31.1% from 1080p to 1440p and 40.9% from 1440p to 4K. Medium settings drop 31.1% and 42.3% respectively. High settings drop 30.8% and 40.7%. Ultra settings drop 32.0% and 41.2%. This consistency suggests that Thief's rendering pipeline scales linearly with pixel count, with no sudden threshold effects from memory or cache limitations.

The only preset with complete min/max data is Medium, which shows that frame time stability decreases at higher resolutions. At 1080p Medium, the min/max spread is 31 FPS (87 to 118). At 1440p Medium, the spread narrows to 21 FPS (60 to 81). At 4K Medium, the spread narrows further to 13 FPS (35 to 48). This narrowing indicates that as the GPU becomes fully saturated, frame times become more consistent but at a much lower performance level, which is typical of GPU-bound scenarios.