Thief

The AMD Ryzen 7 5800XT paired with the AMD Radeon RX 570 delivers a distinctive performance profile in Thief, a 2014 stealth title that remains demanding at higher resolutions. The measured data reveals a system where the GPU is the dominant constraint, particularly as resolution climbs, while the CPU’s strong multi-threading capabilities provide ample headroom that the game rarely exploits. This analysis examines the resolution scaling, GPU characteristics, CPU influence, and settings recommendations based strictly on the measured FPS rows.

Resolution Scaling

The FPS drop from 1080p to 4K is steep and consistent, signaling that the Radeon RX 570’s rendering pipeline is the primary bottleneck. At 1080p Low, the combo averages 75 FPS, which is a smooth, playable baseline. Moving to 1440p Low drops that to 51 FPS — a 32% reduction. At 4K Low, the average plunges to 30 FPS, exactly half of the 1080p Low figure. This pattern is mirrored across all settings. For High, the progression is 43 → 30 → 17 FPS across 1080p, 1440p, and 4Kp, respectively. The Medium preset shows 57 FPS at 1080p, 39 FPS at 1440p, and 23 FPS at 4Kp. Ultra follows the same trajectory: 41 FPS, 28 FPS, 17 FPS.

The scaling factor is nearly linear with pixel count, which is a classic sign of a GPU-bound workload. From 1080p to 4Kp (four times the pixels), the Low preset’s FPS drops from 75 to 30 — a 60% reduction, not the 75% that perfect scaling would suggest. The smaller-than-linear drop indicates that some portion of the frame time is fixed, likely CPU-side overhead, but the overall trend is overwhelmingly dominated by the GPU. At 1440p, the mid-point, the system still delivers playable 30 FPS on High, but 4K is only viable at Low settings. The data implies that Thief at 4K is far beyond the RX 570’s fill-rate capabilities, even with a powerful eight-core CPU feeding it.

GPU Role

The Radeon RX 570 is built on the Polaris 20 chip with a GCN 4.0 architecture, fabricated on a 14 nm process. It operates with a base clock of 1168 MHz and a boost clock of 1244 MHz, with memory running at 1750 MHz (7 Gbps effective) across a 256-bit bus. That configuration yields 224.0 GB/s of memory bandwidth and 32 ROPs, with a peak pixel rate of 39.81 GPixel/s and a texture rate of 159.2 GTexel/s. The card has 2048 shading units, 128 TMUs, and 32 ROPs. In the context of Thief, these specifications explain the 4K collapse — the 32 ROPs are simply insufficient for 8.3 million pixels at high settings.

The GPU’s 4 GB of GDDR5 memory is a notable constraint at higher resolutions. At 4K Ultra, the average FPS is 17, matching the High preset’s 17 FPS, which suggests the game is hitting a memory bandwidth or capacity wall rather than a pure shader-bound limit. The 224.0 GB/s bandwidth is substantial for its era but not enough for 4K textures and effects. Interestingly, the RX 570’s nearest rivals in the benchmark database include the AMD Radeon RX 6800M (which is 0.4% slower), the AMD Radeon RX 470 (0.8% faster), and the NVIDIA Quadro RTX 8000 (1.2% faster). This places the RX 570 in a narrow performance band, all within ±1.2% average score — meaning these results are representative of a mid-range GPU tier, not an outlier.

The GPU’s architecture lacks modern ray tracing cores or tensor cores, but that is irrelevant for a 2014 title like Thief, which relies on traditional rasterization. The FP32 compute rate of 5.095 TFLOPS is modest, yet the game’s geometry and lighting are not particularly compute-heavy. The data shows that the GPU’s pixel throughput, not its compute capability, is the limiting factor at 4K. At 1080p, the RX 570 can exceed 60 FPS only at Low settings, demonstrating that even a capable CPU cannot overcome the GPU’s fill-rate ceiling.

FAQ

*Q: What is the highest average FPS this combo achieves in Thief?*

A: The highest measured average FPS is 75, achieved at 1080p with Low settings. No other setting or resolution exceeds this in the measured data.

Q: Does the RX 570’s 4 GB VRAM cause issues at 4K?

A: The data suggests yes. At 4K Ultra, the average FPS is 17, identical to 4K High (17 FPS), and at 4K Medium it is 23 FPS. The 4 GB VRAM, with 224.0 GB/s bandwidth, appears to be a bottleneck as resolution increases, with 4K results collapsed to a narrow 17–30 FPS range across settings.

*Q: Is the AMD Ryzen 7 5800XT capable of feeding the RX 570 in Thief?*

A: Yes. The CPU has 8 cores and 16 threads, with a boost clock of 4.80 GHz and a Cinebench R23 multicore score of 23794 and single-core score of 3359. The FPS drops from 1080p to 4K are consistent with a GPU limit, not a CPU bottleneck, as the frame rate scales inversely with pixel count.

Q: How does the RX 570 compare to its nearest GPU rivals in terms of average benchmark score?

A: The RX 570’s average benchmark score is 28766. The closest rivals are the AMD Radeon RX 6800M (28874, 0.4% slower), the AMD Radeon RX 470 (28996, 0.8% faster), and the NVIDIA Quadro RTX 8000 (28421, 1.2% faster). The differences are minor, within 1.2%.

Q: What is the difference between Low and Ultra at 1080p?

A: At 1080p, Low yields 75 FPS, Medium 57 FPS, High 43 FPS, and Ultra 41 FPS. Going from Low to Ultra reduces FPS by 34 FPS (45%), while High to Ultra is only a 2 FPS drop, suggesting diminishing returns from High to Ultra.

Q5: Is 4K playable with this combo?

A: Not really. The highest 4K average is 30 FPS at Low settings, which is borderline playable but below typical smoothness targets. At Medium the average drops to 23 FPS, and at High or Ultra it is 17 FPS. The 4K results are uniformly low, making 1440p the highest playable resolution for most settings.

Settings Recommendations

Given the measured data, the best experience depends on the target resolution. At 1080p, the Medium preset offers a strong balance: 57 average FPS, with a minimum of 48 and maximum of 65. This is the only preset at 1080p that stays above 40 FPS while providing higher fidelity than Low. The High preset drops to 43 FPS, which is still playable, but the jump to Ultra only adds a 2 FPS loss (41 FPS), making Ultra a reasonable choice for visual quality over High. Low provides the highest headroom at 75 FPS, but the visual degradation is significant for a stealth game where shadows and lighting matter.

At 1440p, the Low preset is the only one that reaches above 40 FPS, averaging 51 FPS. Medium delivers 39 FPS (min 33, max 45), which is borderline playable but may feel stuttery in fast scenes. High settles at 30 FPS, and Ultra at 28 FPS. For 1440p, the data recommends Low for smooth play, or Medium if the user can tolerate occasional dips to 33 FPS.

At 4K, no preset is truly playable. Low yields 30 FPS, which is the only setting that approaches a console-like frame rate, but the frame pacing is not measured. Medium, High, and Ultra average 23, 17, and 17 FPS respectively. There is no practical recommendation for 4K beyond low settings for screenshots, but not for action sequences. The clear conclusion is that this combo is best served by 1080p Medium or 1440p Low, depending on resolution preference.

Measured FPS Breakdown

1920x1080 (1080p)

• Low: 75 FPS average
• Medium: 57 FPS average, 48 min, 65 max
• High: 43 FPS average
• Ultra: 41 FPS average

2560x1440 (1440p)

• Low: 51 FPS average
• Medium: 39 FPS average, 33 min, 45 max
• High: 30 FPS average
• Ultra: 28 FPS average

3840x2160 (4K)

• Low: 30 FPS average
• Medium: 23 FPS average, 19 min, 26 max
• High: 17 FPS average
• Ultra: 17 FPS average

The data reveals a consistent hierarchy: at every resolution, Low > Medium > High > Ultra, but the gap between High and Ultra is narrow. At 1080p, Ultra is only 2 FPS slower than High (43 vs 41). At 1440p, the difference is also 2 FPS (30 vs 28). At 4K, High and Ultra are identical at 17 FPS. This suggests that the Ultra preset’s extra visual effects are not more demanding than High at higher resolutions, or that the GPU is already saturated at High, making Ultra free from a performance perspective. The Medium preset’s min/max frametimes (e.g., 48-65 at 1080p) show a 17 FPS variance, indicating some frame pacing inconsistency, but the average is still solid.

CPU Role

The AMD Ryzen 7 5800XT is a Zen 3 Vermeer chip with 8 cores and 16 threads, a base clock of 3.80 GHz, and a boost clock of 4.80 GHz. It has 32 MB of shared L3 cache, 512 KB of L2 per core, and 64 KB of L1 per core. It supports DDR4 memory in a dual-channel configuration, with a theoretical memory bandwidth of 51.2 GB/s. In benchmark terms, it scores 23794 in Cinebench R23 multicore and 3359 in single-core, placing it in the 81st percentile of all CPUs. Its nearest rivals are the AMD Ryzen 7 7840H (29868, 0% delta), the AMD Ryzen 7 8845HS (29955, -0.3%), and the AMD Ryzen 5 9600X (29713, -0.3%).

In Thief, the CPU’s role is secondary. The FPS scaling with resolution is a hallmark of a GPU-bound workload: at fixed settings, FPS drops as resolution increases, while the CPU load remains constant. The Ryzen 7 5800XT’s multi-threading would be overkill for a 2014 title, but the single-core performance (3359 in Cinebench R23) is more relevant for the game’s main thread. The data shows no evidence of CPU limiting — at 1080p Low, the 75 FPS is likely the maximum the RX 570 can achieve, not the CPU. The CPU’s 16 threads are likely underutilized, but the game’s engine is old and not designed for 16-thread scaling.

The 5800XT’s average benchmark score is 29879, and its nearest rival is the Ryzen 7 7840H at 29868 (0% delta). This means the 5800XT is essentially head-to-head with a mobile 8-core part, despite being a desktop chip. In Thief, the CPU’s single-thread performance is what matters, and the 5800XT’s boost to 4.8 GHz ensures it can feed the RX 5700 efficiently. However, the data suggests that even a slower CPU would produce similar FPS, since the GPU saturates first. The CPU’s memory bandwidth of 51.2 GB/s is more than sufficient for the game’s data demands, which are modest.

How This Combo Ranks

The combo of AMD Ryzen 7 5800XT and AMD Radeon RX 570 ranks 3551 out of 3720 tested combinations in Thief. This places it in the bottom 5% of all combos (rank 3551 / 3720). That seems counterintuitive given the CPU’s strength, but the ranking is dominated by the GPU’s performance. The RX 570, with a 74th percentile standing among all GPUs, is a mid-range card, and in Thief at 4K it fails to achieve playable FPS. The bottom 5% ranking is due to the game’s high GPU demands at higher resolutions and the RX 570’s limited pixel throughput.

The ranking is not a reflection of the CPU. The Ryzen 7 5800XT is in the 81st percentile of all CPUs, but Thief is not a CPU-heavy title. The data shows the GPU is the limiting factor, so the combo’s rank is essentially the GPU’s rank. In the database, the RX 570’s nearest GPU rivals (RX 6800M, RX 470, Quadro RTX 8000) all have similar average scores, so swapping the CPU would not change the rank much. The combo’s poor ranking is a direct result of the RX 570’s inability to handle 4K, which is a common scenario for many older GPUs.

The rank of 3551 out of 3720 indicates that this combo is not a top performer for Thief, but it also reflects the game’s heavy requirements relative to the GPU. At 1080p, the combo can deliver 75 FPS at Low and 43 at High, which is acceptable for a stealth game. The 4K results drag the rank down, as the measured data includes all resolutions. The rank would be higher if only 1080p results were considered, but the database includes 4K as a test. The data indicates that the combo is a low-end solution for Thief, best suited for 1080p gaming, and the CPU is not the reason for the low rank. The 81st percentile CPU is wasted on a game that is GPU-bound, and the 3551 rank is a reminder that a balanced system matters more than raw CPU strength in this title.