Thief

# How This Combo Ranks

The pairing of the Intel Core Ultra 9 290K Plus with the AMD Radeon RX 580 lands at rank 2617 out of 2851 tested combos in Thief. That places this configuration in the bottom 8% of all measured systems, a result driven almost entirely by the graphics card's age rather than the processor's capability. The RX 580's average benchmark score of 12928 puts it at the 53rd percentile among all GPUs, while the Core Ultra 9 290K Plus sits at the 96th percentile among all CPUs — a stark mismatch that defines this combo's standing.

In practical terms, this is a system where the CPU is operating several leagues above the GPU's head. The processor's nearest rival, the Intel Core Ultra 9 285K, scores 83807 on average (a mere 0.2% delta), while the EPYC 4584PX and EPYC 9135 trail by 1.1% and 1.2% respectively. On the GPU side, the RX 580's closest competitor is the NVIDIA GeForce RTX 3050 Ti Mobile at a -0.1% delta, with the GTX 1660 SUPER just 0.4% ahead. None of these comparisons matter much in Thief, though, because the measured frames tell a story of GPU-bound stagnation across nearly every resolution and setting combination.

The rank itself reflects that the combo's performance ceiling is low relative to the broader field, but it also masks how consistent the experience is. At 1080p Low, the system delivers 83 FPS, which is genuinely playable. But at 4K Ultra, it collapses to 18 FPS. The delta between best and worst measured outcomes is roughly 4.6x, and that spread is almost entirely attributable to the RX 580's fixed hardware limits. The CPU's 24 cores and 24 threads, clocked at 3.70 GHz base and 5.80 GHz boost, never become the bottleneck in any measured row — the data shows the GPU saturating well before the processor breaks a sweat.

For context, this combo's rank of 2617 means that roughly 234 other configurations tested worse, but the overwhelming majority of systems in the database outperform it. The takeaway is straightforward: this is a legacy GPU strapped to a flagship-class CPU, and Thief's measured results reflect that imbalance at every step.

# Settings Recommendations

The measured FPS rows offer a clear hierarchy of playability across Thief's four preset levels. At 1080p, the Low preset delivers 83 FPS average, which is the only configuration in the entire dataset that approaches high-refresh territory. Medium at 1080p drops to 63 FPS with a minimum of 54 FPS, still smooth enough for most gaming scenarios. High at 1080p falls to 48 FPS, and Ultra bottoms out at 46 FPS — both playable but noticeably less fluid.

Stepping up to 1440p, the situation tightens considerably. Low produces 57 FPS average, which remains acceptable, but Medium falls to 43 FPS with a minimum of 37 FPS. High at 1440p yields 33 FPS, and Ultra drops to 31 FPS — both dipping into choppy territory. At 4K, every preset becomes marginal: Low manages 33 FPS, Medium 25 FPS, High 19 FPS, and Ultra 18 FPS. None of these 4K figures meet the 60 FPS threshold that modern gamers expect, and even the 30 FPS baseline is only barely crossed by the Low preset.

The data suggests that the optimal experience for this combo is 1080p Medium. It offers a 63 FPS average with a 54 FPS minimum, which keeps frame pacing consistent enough for the stealth gameplay that Thief demands. The jump from Low to Medium costs 20 FPS on average, but the visual uplift is substantial — and the minimum frame rate of 54 FPS ensures the experience stays above the stutter threshold. At 1080p High, the 48 FPS average is still workable, but the loss of 15 FPS from Medium for what is typically a modest visual gain makes it less compelling.

For users who prioritize frame rate above all else, 1080p Low at 83 FPS is the clear winner, but it sacrifices the game's atmospheric lighting and shadow work, which are central to Thief's stealth mechanics. The Medium preset at 1080p represents the best balance of visual fidelity and responsiveness in the measured data. At higher resolutions, users should expect to drop settings aggressively — 1440p Low at 57 FPS is the only viable option above 1080p, and even that falls short of a locked 60 FPS.

# Resolution Scaling

The FPS degradation from 1080p to 4K reveals where this combo's bottleneck lives. At 1080p Low, the system produces 83 FPS. Moving to 1440p Low drops that to 57 FPS — a 31% reduction. At 4K Low, the average falls to 33 FPS, which is 60% lower than the 1080p figure. The scaling is nearly linear with pixel count, which is the classic signature of a GPU-bound workload. The RX 580's 256-bit memory bus and 256.0 GB/s bandwidth are simply overwhelmed as resolution climbs.

The same pattern holds across every preset. At Medium, 1080p yields 63 FPS, 1440p yields 43 FPS (a 32% drop), and 4K yields 25 FPS (a 60% drop from 1080p). At High, the numbers are 48, 33, and 19 FPS respectively — a 31% drop from 1080p to 1440p and a 60% drop from 1080p to 4K. Ultra follows suit: 46 FPS at 1080p, 31 FPS at 1440p, and 18 FPS at 4K.

This consistency is telling. The percentage losses are nearly identical across all four presets, which means the limiting factor is not the settings themselves but the raw pixel throughput of the GPU. The Core Ultra 9 290K Plus, with its 36 MB of shared L3 cache and 115.2 GB/s memory bandwidth, is more than capable of feeding frames at any resolution. The data shows that the CPU's role in this game is essentially static — it is never the constraining component.

If the system were CPU-bound, we would expect to see FPS remain relatively flat as resolution increases, with the GPU having headroom to spare. Instead, the measured results show FPS falling in near-perfect proportion to pixel count, confirming that the RX 580 is the sole performance limiter. At 4K, the GPU's 32 ROPs and 6.175 TFLOPS of FP32 compute are simply insufficient for Thief's rendering demands, regardless of how powerful the processor behind it might be.

# Measured FPS Breakdown

At 1920x1080, the combo delivers its best results. Low settings produce an average of 83 FPS, the highest figure in the entire dataset. Medium averages 63 FPS with a minimum of 54 FPS and a maximum of 72 FPS, showing a reasonably tight frame distribution. High drops to 48 FPS average, and Ultra falls to 46 FPS. The gap between Low and Ultra at 1080p is 37 FPS, or roughly 45% — a significant range that gives users real choices based on their preference for visual quality versus smoothness.

At 2560x1440, the numbers compress downward. Low averages 57 FPS, which is the only 1440p preset that stays above 50. Medium produces 43 FPS with a minimum of 37 FPS and a maximum of 50 FPS, indicating occasional dips below what most players would consider smooth. High averages 33 FPS, and Ultra averages 31 FPS. The spread between Low and Ultra at 1440p is 26 FPS, a narrower range than at 1080p because the GPU's limits become more pronounced at higher pixel counts.

At 3840x2160, the combo struggles to maintain playability. Low averages 33 FPS, Medium averages 25 FPS with a minimum of 22 FPS and a maximum of 29 FPS, High averages 19 FPS, and Ultra averages 18 FPS. The minimum frame rate of 22 FPS at 4K Medium is the lowest recorded in the dataset, and even the best 4K result (33 FPS at Low) falls short of a consistent 30 FPS. The difference between 1080p Low and 4K Low is 50 FPS, underscoring how quickly the RX 580's performance evaporates as resolution climbs.

The data also reveals that the difference between High and Ultra at any given resolution is minimal. At 1080p, the gap is 2 FPS; at 1440p, it is 2 FPS; at 4K, it is 1 FPS. This suggests that Ultra's additional rendering features impose almost no extra cost on this particular GPU, likely because the card is already saturated at High. The meaningful performance jumps occur between Low, Medium, and High, with Medium typically offering the best compromise of frame rate and visual quality.

# GPU Role

The AMD Radeon RX 580 is the defining component in this combo, and its specifications explain the measured results. Built on the Polaris 20 chip with GCN 4.0 architecture on a 14 nm process, the card packs 2304 shading units, 144 texture mapping units, and 32 ROPs. Its base clock of 1257 MHz boosts to 1340 MHz, while the 8 GB of GDDR5 memory runs at 2000 MHz (8 Gbps effective) across a 256-bit bus, yielding 256.0 GB/s of bandwidth. The GPU's pixel rate is 42.88 GPixel/s and its texture rate is 193.0 GTexel/s, with FP32 compute rated at 6.175 TFLOPS.

These numbers place the RX 580 at the 53rd percentile among all GPUs, with an average benchmark score of 12928. Its nearest rivals — the RTX 3050 Ti Mobile, GTX 1660 SUPER, Radeon 740M, and FirePro W5100 — all sit within 0.6% of its score, confirming that the card occupies a narrow performance band. In Thief, that band translates to the 18-83 FPS range measured across all resolution and settings combinations.

The GPU's 8 GB of VRAM is not the limiting factor in this game; Thief's 2014 release date means its texture requirements are modest by modern standards. Instead, the card's raw compute throughput and memory bandwidth cap its performance. At 4K, the 32 ROPs become a severe bottleneck — the 42.88 GPixel/s pixel rate is simply insufficient to fill a 3840x2160 frame at acceptable speeds. The data shows this clearly: 4K results range from 18 to 33 FPS, with no preset breaking past the 30 FPS mark.

The RX 580's DirectX 12 support (12_0) and Vulkan 1.3 capability are present, but Thief's original release predates widespread adoption of these APIs, meaning the game likely runs on older rendering paths that favor raw fill rate over modern features. The card's 185 W TDP and dual-slot cooler are irrelevant to the benchmark data, but they do reflect the hardware's era — a mid-range GPU from 2017 that was never designed for 4K gaming.

# CPU Role

The Intel Core Ultra 9 290K Plus is a 24-core, 24-thread processor based on the Arrow Lake Refresh architecture, manufactured on a 3 nm process at TSMC. Its base clock of 3.70 GHz boosts to 5.80 GHz, and it carries 36 MB of shared L3 cache plus 3 MB of L2 per core. The CPU's memory support is dual-channel DDR5 with a bandwidth of 115.2 GB/s, and it connects via PCIe Gen 5 with 20 CPU lanes. This is a flagship-class desktop processor, and its benchmark scores reflect that: Cinebench R23 multicore hits 51731, single-core 7303, and Passmark multithread scores 60860.

In Thief, this CPU is effectively idle relative to its capabilities. The game's stealth gameplay and 2014-era engine are not heavily threaded, and the measured FPS data shows no evidence of CPU limitation at any resolution or settings level. The 83 FPS at 1080p Low, for example, is far below what this processor could feed a more powerful GPU — the bottleneck is entirely on the RX 580's side. The CPU's 96th percentile ranking among all processors (average benchmark score 84003) stands in stark contrast to the GPU's 53rd percentile, creating a massive performance asymmetry.

The CPU's nearest rivals — the Core Ultra 9 285K at 0.2% delta, EPYC 4584PX at 1.1%, and EPYC 9135 at 1.2% — all score similarly, meaning the 290K Plus is at the very top of the desktop CPU hierarchy. But in this combo, that power goes unused. Thief's frame rates are governed by the GPU's 6.175 TFLOPS of compute and 256.0 GB/s of bandwidth, not by the CPU's 5.80 GHz boost clock or 36 MB of cache. If the RX 580 were replaced with a modern high-end GPU, the processor would likely deliver vastly higher frame rates, but the measured data only reflects what this specific pairing can achieve.

The CPU's 24 threads do not translate to a meaningful advantage in Thief, which likely utilizes only a handful of cores. The processor's single-thread performance — Cinebench R15 single-core score of 736 and Passmark single-thread of 4823 — is more relevant, but even that exceeds the game's demands at the frame rates measured. The data suggests that the Core Ultra 9 290K Plus is not the limiting factor in any configuration tested; it is the RX 580 that dictates every result in this dataset.