Thief

FAQ

Q: What is the measured average FPS for Thief at 4K Ultra with this hardware combination?

A: The measured average FPS is 87 at 3840x2160 with Ultra settings. This is the lowest average across all tested settings at 4K, confirming that Ultra presets impose the heaviest load on the RTX 5090 D in this title.

Q: How does the AMD Ryzen 5 5500 compare to its closest rival in average benchmark score?

A: The Ryzen 5 5500 has an average benchmark score of 19593, which is 0.1% lower than the AMD Ryzen AI 5 430 (score 19617) and 0.4% lower than the Intel Core i5-12500 (score 19668). It is 0.5% higher than the Intel Core i7-10700F (score 19499).

Q: What is the GPU's percentile ranking among all GPUs?

A: The NVIDIA GeForce RTX 5090 D sits in the 92nd percentile among all GPUs, with an average benchmark score of 77712. Its closest rival, the AMD Radeon RX 6650M XT, scores 76904, putting the RTX 5090 D 1.1% ahead.

Q: Which resolution and settings combination yields the highest average FPS in Thief?

A: The highest measured average FPS is 150, achieved at 1920x1080 with Low settings. This is followed by 136 FPS at 1080p Medium and 132 FPS at 1440p Low.

Q: Does the RTX 5090 D's 32 GB VRAM show any benefit in Thief's measured results?

A: The data shows no VRAM-related bottleneck across any resolution or settings. At 4K Ultra, the average FPS is 87, and at 4K Low it rises to 111. The 32 GB GDDR7 memory and 1.79 TB/s bandwidth exceed what Thief requires, as FPS scales consistently with settings rather than hitting memory limits.

Q: What is the combo rank of this CPU-GPU pairing in Thief?

A: The combination ranks 658 out of 2851 tested combos, placing it in the top 23% of all configurations for this game. This reflects a strong but not top-tier position, constrained by the CPU's performance ceiling.

GPU Role

The NVIDIA GeForce RTX 5090 D is the dominant component in this pairing. Its architecture, Blackwell 2.0, is built on a 5 nm process at TSMC, with 92,200 million transistors on a 750 mm² die. The GPU operates at a base clock of 2017 MHz and boosts to 2407 MHz, with memory clocked at 1750 MHz (28 Gbps effective). This memory configuration, 32 GB of GDDR7 on a 512-bit bus, delivers 1.79 TB/s of bandwidth — a figure that dwarfs the demands of a 2014 stealth title like Thief.

The GPU's raw compute capabilities are substantial: 21,760 shading units, 680 TMUs, and 176 ROPs. Its FP32 throughput is 104.8 TFLOPS, and it maintains a 1:1 FP16 ratio at 104.8 TFLOPS. Pixel rate is 423.6 GPixel/s, and texture rate reaches 1,636.8 GTexel/s. These specifications place the RTX 5090 D in the 92nd percentile among all GPUs, with an average benchmark score of 77712. Its nearest rival, the AMD Radeon RX 6650M XT, scores 76904, meaning the RTX 5090 D leads by 1.1%. Against the AMD Radeon RX 6850M XT (score 78940), it trails by 1.6%.

In Thief, the GPU's role is clearly dominant. At 4K Ultra, the average FPS is 87, and the drop from 1080p Ultra (121 FPS) to 4K Ultra (87 FPS) represents a 28% reduction. This scaling pattern is consistent with a GPU-bound scenario at higher resolutions. The 32 GB VRAM never becomes a limiting factor, as evidenced by the smooth FPS progression across settings at each resolution. The GPU's high pixel rate and texture rate ensure that even at 4K, the game's rendering pipeline is fed without stutter.

The RTX 5090 D's performance percentile (92nd) and its position relative to rivals underscore that this GPU is overqualified for Thief's engine. The game's 2014 release date means its graphical demands are modest by modern standards, yet the GPU still delivers a measurable performance spread across settings — from 87 FPS at 4K Ultra to 150 FPS at 1080p Low — indicating that the GPU is never the sole constraint, but it remains the primary driver of FPS at higher resolutions.

Resolution Scaling

The measured FPS data reveals how Thief scales from 1080p to 4K, and the pattern indicates a shift in the limiting component. At 1080p, the average FPS across settings ranges from 121 (Ultra) to 150 (Low). At 1440p, this range narrows to 107 (Ultra) to 132 (Low). At 4K, the range compresses further to 87 (Ultra) to 111 (Low).

The scaling from 1080p to 1440p is modest. For High settings, FPS drops from 123 to 108, a 12% reduction. From 1440p to 4K, the same settings see a further drop to 91 FPS, a 16% reduction. The cumulative drop from 1080p High to 4K High is 26% (123 to 91 FPS). For Low settings, the drop from 1080p to 4K is 26% as well (150 to 111 FPS). This consistency across settings suggests that the GPU is handling the resolution scaling predictably, with no sudden cliff that would indicate a VRAM or bandwidth bottleneck.

The limiting component analysis is nuanced. At 1080p, the CPU's role is more pronounced. The Ryzen 5 5500 has a base clock of 3.60 GHz and a boost clock of 4.20 GHz, with 6 cores and 12 threads. Its single-thread performance, as measured by Cinebench R23, is 2319, and its multi-core score is 16429. These figures are modest by 2025 standards, and in a game from 2014, the CPU's single-thread capability becomes the ceiling at lower resolutions. The fact that 1080p Ultra (121 FPS) is only slightly below 1440p Ultra (107 FPS) — a 12% drop — suggests that the CPU is holding back the GPU at 1080p, preventing the GPU from fully stretching its legs.

At 4K, the GPU becomes the clear limiter. The 87 FPS at 4K Ultra versus 121 FPS at 1080p Ultra shows a 28% drop, which is larger than the 1080p-to-1440p drop of 12%. This steepening curve indicates that at 4K, the RTX 5090 D's rendering pipeline is fully saturated, and the CPU's influence diminishes. The GPU's 104.8 TFLOPS FP32 throughput is being taxed more heavily at 4K, and the data reflects this.

The Medium settings row provides the only min/max FPS data across all resolutions. At 1080p Medium, the average is 136 FPS with a min of 115 and max of 156. At 1440p Medium, the average is 120 FPS with min 102 and max 138. At 4K Medium, the average is 100 FPS with min 85 and max 115. The spread between min and max narrows as resolution increases (41 FPS at 1080p, 36 FPS at 1440p, 30 FPS at 4K), indicating that frame pacing becomes more consistent as the GPU becomes the bottleneck, reducing CPU-induced variance.

Measured FPS Breakdown

The measured FPS data for Thief with the AMD Ryzen 5 5500 and NVIDIA GeForce RTX 5090 D is as follows:

1920x1080 (1080p):

• Low: 150 FPS average
• Medium: 136 FPS average, min 115, max 156
• High: 123 FPS average
• Ultra: 121 FPS average

2560x1440 (1440p):

• Low: 132 FPS average
• Medium: 120 FPS average, min 102, max 138
• High: 108 FPS average
• Ultra: 107 FPS average

3840x2160 (4K):

• Low: 111 FPS average
• Medium: 100 FPS average, min 85, max 115
• High: 91 FPS average
• Ultra: 87 FPS average

At 1080p, the spread between Low and Ultra is 29 FPS (150 vs 121), which is a 19% difference. This relatively small gap indicates that even at the highest settings, the CPU is limiting performance. The jump from Ultra (121) to High (123) is only 2 FPS, suggesting that the Ultra preset's additional effects are not heavily taxing the GPU at this resolution.

At 1440p, the spread between Low and Ultra widens to 25 FPS (132 vs 107), a 19% difference again. The High-to-Ultra gap is just 1 FPS (108 vs 107), showing that the GPU is beginning to take over as the limiter, but the CPU still constrains the upper end.

At 4K, the spread between Low and Ultra grows to 24 FPS (111 vs 87), a 22% difference. The High-to-Ultra gap remains 4 FPS (91 vs 87), but the Low-to-High gap is 20 FPS (111 vs 91), indicating that the GPU's workload scales significantly with settings at this resolution. The 4K Ultra result of 87 FPS is the lowest measured value, yet it still exceeds 60 FPS, confirming that this combination is playable at maximum settings in Thief.

The Medium settings rows provide the only min/max data. At 1080p, the min of 115 and max of 156 show a 41 FPS range. At 1440p, the range narrows to 36 FPS (102 to 138). At 4K, the range is 30 FPS (85 to 115). This narrowing range as resolution increases is a classic sign of the GPU becoming the bottleneck, as frame time variance from the CPU diminishes.

Settings Recommendations

For 1080p gaming, the data suggests that High settings (123 FPS) offer the best balance. The jump from High to Ultra (121 FPS) yields only a 2 FPS drop, so Ultra is also viable. However, Low (150 FPS) and Medium (136 FPS) sacrifice visual fidelity for marginal FPS gains that are already well above refresh rate thresholds. The 1080p Medium row shows a min of 115 FPS, which is still smooth, but High's consistent 123 FPS average with no min data suggests stability.

At 1440p, High settings (108 FPS) are the recommended choice. The drop from High to Ultra (107 FPS) is negligible, so Ultra is equally viable. The 1440p Medium row shows a min of 102 FPS and max of 138, indicating strong frame pacing. Low (132 FPS) offers a 24 FPS advantage over High, but the visual compromise in a stealth game, where shadow and texture detail affect gameplay readability, makes High the better pick.

At 4K, the data points to High settings (91 FPS) as the optimal preset. Ultra (87 FPS) is only 4 FPS lower and remains playable, but the 4K Medium row shows a min of 85 FPS, which is close to the Ultra average. This suggests that at 4K, the GPU is the primary constraint, and the difference between Medium and Ultra is small. The 4K Low result (111 FPS) is 20 FPS higher than High, but the visual downgrade is substantial in a game that relies on ambient lighting for stealth mechanics.

The best overall experience across all resolutions is achieved at High settings. At 1080p, High delivers 123 FPS; at 1440p, it delivers 108 FPS; and at 4K, it delivers 91 FPS. This consistency — a 26% drop from 1080p to 4K — indicates that High settings scale predictably with resolution, avoiding the steep cliffs seen at Ultra (121 to 87 FPS, a 28% drop) and Low (150 to 111 FPS, a 26% drop). For players prioritizing frame rate over fidelity, Medium at 4K (100 FPS) provides a comfortable buffer above 60 FPS with a min of 85, making it a strong fallback.

CPU Role

The AMD Ryzen 5 5500 is a 6-core, 12-thread processor from the 5000 series, based on the Zen 3 architecture with the Cezanne codename. It operates on the AMD Socket AM4 platform with a base clock of 3.60 GHz and a boost clock of 4.20 GHz. The CPU has a 16 MB L3 cache, 512 KB L2 per core, and 64 KB L1 per core. Its memory support is DDR4 dual-channel with a bandwidth of 51.2 GB/s. The CPU's TDP is 65 W, and it is manufactured on a 7 nm process at TSMC.

In Thief, the CPU's role is most apparent at 1080p. The game, released in 2014, was designed for CPUs of that era, and its engine typically relies on a few heavily threaded tasks. The Ryzen 5 5500's single-thread performance, as measured by Cinebench R23 (2319 points) and Geekbench single-core (1730), is adequate but not exceptional. Its multi-thread scores are stronger: Cinebench R23 multi-core (16429), Geekbench multi-core (7976), and Passmark multithread (19330).

The benchmark data shows that at 1080p, the CPU limits the GPU. The RTX 5090 D, despite being in the 92nd GPU percentile, cannot push beyond 150 FPS at 1080p Low. This is not a GPU limitation — the same GPU at 4K Low produces 111 FPS, and at 1080p Ultra it produces 121 FPS. The 29 FPS gap between Low and Ultra at 1080p is smaller than the 24 FPS gap at 4K, indicating that the CPU is capping the upper FPS ceiling at lower resolutions.

The CPU's percentile ranking is 73 among all CPUs, with an average benchmark score of 19593. Its nearest rivals are closely matched: the AMD Ryzen AI 5 430 (19617, 0.1% higher), Intel Core i5-12500 (19668, 0.4% higher), Intel Core i7-10700F (19499, 0.5% lower), and Intel Core i5-11600KF (19723, 0.7% higher). This clustering shows that the Ryzen 5 5500 is positioned in a competitive mid-range segment, and its performance in Thief reflects this — it is sufficient to maintain 60+ FPS even at 4K Ultra, but it does leave performance on the table at 1080p.

The CPU's boost clock of 4.20 GHz is the key figure for Thief's single-threaded tasks. The game's stealth mechanics, AI routines, and scene management often run on fewer threads, and the 4.20 GHz boost helps maintain responsiveness. However, the 3.60 GHz base clock means that under sustained multi-threaded loads, the CPU may not sustain peak frequencies. The measured FPS data at 1080p High (123 FPS) versus 1440p High (108 FPS) shows a 12% drop, which is less than the 1440p-to-4K drop of 16% (108 to 91 FPS). This suggests that the CPU's influence fades as resolution increases, but it never completely disappears — even at 4K, the CPU's single-thread performance contributes to the 87 FPS Ultra result.

The combo rank of 658 out of 2851 (top 23%) reflects this balance. The CPU is not a bottleneck at 4K, but it prevents the GPU from achieving its full potential at 1080p. For Thief, the Ryzen 5 5500 provides a stable foundation, and the data shows that upgrading the CPU would yield diminishing returns at higher resolutions, while the GPU remains the primary driver of FPS at 4K.