Thief

Thief, a 2014 stealth title, places this particular CPU and GPU pairing at rank 2375 out of 2851 tested combos, putting it in the lower 17% of systems evaluated for this game. That rank alone signals a combination that will struggle to deliver a consistently smooth experience at higher settings, and the measured FPS data confirms this is a pairing where the graphics card is the dominant constraint. The data shows a system that can handle the game at 1080p with medium settings comfortably, but one that runs out of headroom quickly as resolution and quality demands increase. This analysis will break down exactly how the Core Ultra 5 245 and GeForce GTX 1070 interact across the four tested presets and three resolutions, using only the measured results and benchmark data provided.

How This Combo Ranks

The combo rank of 2375 out of 2851 tested combinations places this system firmly in the lower quartile of performance for Thief. Benchmark results indicate that most other tested hardware pairings deliver higher frame rates, which is unsurprising given the age and tier of the GeForce GTX 1070 relative to the game’s demands at higher settings. The GPU’s percentile rank of 47 places it just below the median of all tested graphics cards, meaning roughly half of all GPUs in the database outperform it. The CPU, in contrast, sits at the 90th percentile among all processors, a much stronger position. This mismatch is the core story of this pairing: a modern, capable processor paired with a graphics card that is now showing its age in a title that can be demanding at max settings.

Looking at the CPU’s nearest rivals provides context for its strength. The Intel Core Ultra 5 245 posts an average benchmark score of 48995, which is essentially level with the AMD Ryzen 7 PRO 5755G at 49196 (a -0.4% delta) and the AMD Ryzen 9 7900 at 49228 (-0.5%). It also edges out the Intel Xeon Gold 5318H by 0.6% and the Intel Core i5-14600K by 0.8%. These deltas are all within a single percentage point, indicating that the Core Ultra 5 245’s raw compute performance is squarely in the middle of a tight cluster of competitive parts. For Thief, this means the CPU is not the bottleneck; it has ample headroom to feed the GPU in most scenarios.

The GPU’s rival comparisons paint a different picture. The GeForce GTX 1070’s average benchmark score of 9780 is nearly identical to the AMD FirePro W5000 (9803, -0.2%), NVIDIA Quadro M2000M (9832, -0.5%), NVIDIA Tesla M10 (9724, +0.6%), and NVIDIA Tesla C2070 (9716, +0.7%). These are all older or workstation-class parts, and the fact that the GTX 1070 clusters with them rather than with modern gaming GPUs underscores its position in the hierarchy. When combined with the 90th-percentile CPU, the pairing’s low combo rank is entirely attributable to the GPU’s limitations in Thief’s more demanding modes.

GPU Role

The GeForce GTX 1070 is built on the Pascal architecture with a GP104 chip fabricated on a 16 nm process. It carries 8 GB of GDDR5 memory on a 256-bit bus, yielding a bandwidth of 256.3 GB/s. The memory clock runs at 2002 MHz, or 8 Gbps effective, while the base GPU clock is 1506 MHz and the boost clock reaches 1683 MHz. These specifications are directly relevant to Thief’s results because the game’s asset streaming and texture loads at higher resolutions will stress both memory capacity and bandwidth. With 8 GB of VRAM, the card has enough memory for most situations in this title, but the 256 GB/s bandwidth becomes a limiting factor at 4K.

The measured FPS data shows a clear pattern: as resolution increases, the GTX 1070’s performance degrades sharply. At 1080p with Low settings, the card produces 128 FPS average, but at 4K with Low settings, that drops to 51 FPS. This is a 60% reduction in frame rate from the lowest to highest resolution at the same settings, which points to the GPU’s fill rate and memory bandwidth being overwhelmed by the pixel count. The card’s pixel rate of 107.7 GPixel/s and texture rate of 202.0 GTexel/s are modest by modern standards, and Thief’s scenes with dynamic lighting and shadows will push these limits.

The GPU’s FP32 performance is 6.463 TFLOPS, which is adequate for a 2014 game but not exceptional. The card’s DirectX 12 support (12_1) and Vulkan 1.4 API compatibility mean it can run modern rendering paths, but its benchmark scores in DirectX 12 (48 in Passmark) versus DirectX 11 (100) suggest the older API is where it performs relatively better. Thief was released in 2014, and its engine likely favors the DirectX 11 path, which aligns with the card’s stronger score in that category. The data indicates that the GTX 1070’s role here is as the primary limiter; the CPU is not holding it back.

FAQ

Q: What is the best 1080p setting for this combo in Thief?

A: The measured data shows that Low settings yield 128 FPS average, Medium produces 97 FPS (with a minimum of 82 and maximum of 111), High delivers 74 FPS, and Ultra drops to 70 FPS. Medium offers the best balance, staying well above 60 FPS while providing higher quality than Low.

Q: Can this system handle 1440p gaming in Thief?

A: Yes, but with compromises. At 2560x1440, Low settings achieve 88 FPS, Medium hits 66 FPS (min 56, max 76), High averages 51 FPS, and Ultra drops to 48 FPS. Medium is playable, but High and Ultra fall below the 60 FPS threshold.

Q: Is 4K resolution viable for Thief with this hardware?

A: The data shows 4K results are generally poor. Ultra averages 28 FPS, High averages 30 FPS, Medium averages 39 FPS (min 33, max 45), and Low is the only playable option at 51 FPS. Even Low struggles to maintain smooth gameplay, so 4K is not recommended.

Q: How does the CPU compare to its rivals in raw performance?

A: The Intel Core Ultra 5 245 has an average benchmark score of 48995, which is within 1% of the AMD Ryzen 7 PRO 5755G (49196), AMD Ryzen 9 7900 (49228), Intel Xeon Gold 5318H (48698), and Intel Core i5-14600K (48618). It is statistically tied with all of them.

Q: What is the GPU’s percentile ranking among all tested cards?

A: The GeForce GTX 1070 sits at the 47th percentile, meaning it performs better than 47% of all GPUs in the database. Its average benchmark score is 9780, placing it just below the median.

Q: Which resolution shows the smallest performance gap between settings?

A: At 1920x1080, the gap between Ultra (70 FPS) and Low (128 FPS) is 58 FPS. At 4K, the same gap is only 23 FPS (51 FPS Low vs 28 FPS Ultra). The tighter spread at 4K indicates the GPU is saturated at all settings, reducing the impact of quality changes.

Measured FPS Breakdown

The measured FPS data provides a complete picture of this combo’s performance across three resolutions and four settings. Starting at 1920x1080, the system delivers its best results. Low settings produce an average of 128 FPS, which is exceptionally smooth. Medium settings average 97 FPS, with a measured minimum of 82 FPS and maximum of 111 FPS, indicating that even in demanding scenes, the frame rate stays above 80. High settings average 74 FPS, and Ultra averages 70 FPS. These numbers show that 1080p is the sweet spot for this pairing, with all settings exceeding 60 FPS on average and even Ultra remaining playable.

Moving to 2560x1440, the performance drops noticeably. Low settings average 88 FPS, which is still smooth. Medium settings average 66 FPS, with a minimum of 56 FPS and maximum of 76 FPS. This minimum is below the 60 FPS target, suggesting occasional stutters in heavy scenes. High settings average 51 FPS, and Ultra averages 48 FPS. Neither High nor Ultra consistently maintains playable frame rates, so users at 1440p should stick to Medium or Low for a smoother experience.

At 3840x2160, the results are marginal across the board. Low settings average 51 FPS, which is the only preset that approaches playability. Medium averages 39 FPS, with a minimum of 33 FPS and maximum of 45 FPS. High averages 30 FPS, and Ultra averages 28 FPS. These numbers indicate that 4K is not a practical resolution for this GPU in Thief, as even Low settings dip below 60 FPS, and higher settings become a slideshow. The data shows a clear hierarchy: 1080p for all settings, 1440p for Low/Medium, and 4K only at Low with compromises.

Resolution Scaling

The FPS drop from 1080p to 4K is dramatic and reveals the limiting component. At Low settings, the average FPS falls from 128 at 1080p to 88 at 1440p, then to 51 at 4K. This represents a 31% reduction from 1080p to 1440p, and a further 42% reduction from 1440p to 4K. Overall, the drop from 1080p to 4K is a 60% loss in average FPS. At Medium settings, the progression is 97 FPS at 1080p, 66 FPS at 1440p, and 39 FPS at 4K—a 60% loss from 1080p to 4K as well. High settings go from 74 FPS at 1080p to 51 FPS at 1440p and 30 FPS at 4K, a 59% drop. Ultra goes from 70 FPS at 1080p to 48 FPS at 1440p and 28 FPS at 4K, a 60% drop.

The consistency of this ~60% reduction across all settings is telling. If the CPU were the bottleneck, the scaling would be less uniform because the CPU load would increase less with resolution. Instead, the frame rate scales almost linearly with pixel count, which is the signature of a GPU-bound system. The 256.3 GB/s memory bandwidth and the card’s fill rate are being saturated as the pixel load increases. The data suggests that even at Low settings, the GTX 1070’s compute and memory resources are fully utilized at 4K, leaving no headroom for higher quality presets. This is a GPU-limited scenario through and through.

Another way to read the scaling is to look at the ratio between 1080p and 4K at each setting. For Low, the 4K FPS is 39.8% of the 1080p value (51/128). For Medium, it’s 40.2% (39/97). For High, it’s 40.5% (30/74). For Ultra, it’s 40% (28/70). This remarkably tight range (39.8% to 40.5%) indicates that the GPU’s performance ceiling is being hit uniformly, regardless of the visual complexity. The CPU’s 14 cores and 14 threads are more than sufficient for Thief’s single-threaded and lightly threaded workloads, so the processor is never the constraint.

Settings Recommendations

Based on the measured rows, the best experience for this combo in Thief depends heavily on the target resolution. At 1080p, Medium settings offer the optimal balance. The average FPS of 97 is well above the 60 FPS threshold, and the minimum of 82 FPS ensures that even the most demanding scenes remain smooth. The jump from Medium to High costs 23 FPS (from 97 to 74), which is a significant drop for marginal visual gains, while the jump from High to Ultra costs only 4 FPS (from 74 to 70). If a user prioritizes quality and can tolerate occasional dips below 70 FPS, Ultra is viable, but Medium is the recommended default for a consistently fluid experience.

At 1440p, Medium settings are again the best choice, but with more caveats. The average FPS of 66 is just above 60, but the minimum of 56 FPS means there will be moments of perceptible slowdown. Low settings at 88 FPS average are smoother and would be the choice for competitive or fast-paced play, but for a stealth game like Thief, where pacing is slower, Medium’s higher fidelity is worth the occasional frame dips. High and Ultra are not recommended at 1440p, as both average below 55 FPS, which is below the playability threshold for most users.

At 4K, no setting delivers a fully smooth experience. Low settings average 51 FPS, which is the best possible result, but it still falls short of 60 FPS. Medium averages 39 FPS, which is playable if the user is tolerant of stutter, but High and Ultra are effectively unplayable at 30 FPS and 28 FPS respectively. For 4K, the recommendation is to use Low settings and accept the frame rate, or to lower the resolution to 1440p for a better experience. The data clearly shows that the GTX 1070 was not designed for 4K gaming in this title, and the measured results reflect that limitation.

CPU Role

The Intel Core Ultra 5 245 is a 14-core, 14-thread processor based on the Arrow Lake architecture, built on a 3 nm process. Its base clock is 3.50 GHz, with a boost clock of 5.10 GHz. The CPU has a 24 MB shared L3 cache, with 192 KB of L1 and 3 MB of L2 per core. This is a modern, high-performance desktop part, as evidenced by its 90th percentile ranking among all CPUs. Its benchmark scores are strong: Cinebench R23 multicore reaches 32924, single-core reaches 4648, and Passmark multithread scores 38706. These numbers indicate a processor with substantial compute headroom.

In the context of Thief, the CPU’s role is secondary. The game was released in 2014, and its engine is not heavily multithreaded, meaning the Core Ultra 5 245’s 14 cores are likely underutilized. The single-core performance, as measured by Cinebench R23 single-core score of 4648, is more relevant, and it is more than sufficient for the game’s logic and AI. The CPU’s boost clock of 5.10 GHz ensures that any single-threaded workloads are handled quickly. The data shows that the CPU is never the bottleneck in this pairing, as the FPS scaling with resolution is consistent with GPU limitations.

The CPU’s memory support for DDR5 with a dual-channel bus and 102.4 GB/s bandwidth is also ample for Thief’s data demands. The game’s assets are not excessively large, and the CPU’s cache hierarchy, including the 24 MB L3 cache, helps keep frequently accessed data close to the cores. The nearest rivals to the Core Ultra 5 245, such as the AMD Ryzen 9 7900 and Intel Core i5-14600K, are within 0.8% in average benchmark score, confirming that this CPU is competitive with other top-tier parts. For Thief, the CPU’s contribution is to provide a stable platform that never introduces frame stutter or CPU-bound FPS caps, allowing the GTX 1070 to operate at its full potential, which is the limiting factor in this combination.