Thief

Thief

AVERAGE FPS
76
good

This combination provides smooth gameplay with an average of 76 FPS, suitable for most gaming scenarios.

Average FPS: resolution vs quality settings

UltraHighMediumLow
4K Ultra HD 34 35 47 61
1440p QHD 57 60 79 104
1080p Full HD 84 88 115 152

Cell color: green is 120+ FPS, teal is 60+, amber is 30+, red is below 30.

Every measured configuration

3840x2160 Low 61
3840x2160 Medium 47
3840x2160 High 35
3840x2160 Ultra 34
2560x1440 Low 104
2560x1440 Medium 79
2560x1440 High 60
2560x1440 Ultra 57
1920x1080 Low 152
1920x1080 Medium 115
1920x1080 High 88
1920x1080 Ultra 84

Thief with Intel Core Ultra 5 235 + AMD Radeon RX 5700 XT, In-Depth Analysis

The data-driven profile of the Intel Core Ultra 5 235 paired with an AMD Radeon RX 5700 XT in Thief (2014) shows a system that can deliver smooth 1080p and 1440p play, but at 4K the overhead becomes present. The Core Ultra 5 235 has 14 cores and 14 threads, and its base clock of 3.40 GHz, boost of 5.00 GHz, and 24 MB of shared L3 cache reflect an architecture and manufacturing on a 3 nm process. Meanwhile, the RX 5700 XT packs 8 GB of GDDR6 memory on a 256-bit bus, yielding 448.0 GB/s bandwidth, alongside a 1605 MHz base and 1905 MHz boost clock. Together, they produce the measured frames per second (FPS) arrays that follow—each value directly from the benchmark data, with no extrapolation.

CPU Role

The Intel Core Ultra 5 235 is a 14-core / 14-thread desktop processor with a straightforward set of characteristics: base frequency 3.40 GHz, boost up to 5.00 GHz, and a 65 W TDP. Its architecture is Arrow Lake (code name Arrow Lake-S) and it is manufactured on a 3 nm node at TSMC. While the game’s production is from 2014, the CPU’s single-thread and multi-thread capabilities are nonetheless modern. The Cinebench R23 multi-core score is 14769, and the single-core score is 2085. In the PassMark single-thread test it reaches 4516, and in the integer math synthetic it scores 87948. These numbers consistently place the CPU in the upper echelon of current processors (its CPUcomboPercentileVsAllCpus is 89). Among nearest rivals from the database, the AMD Ryzen AI 9 HX 375 (avg score 46030) and the Intel Core i9-13900HX (avg score 46098) both sit within 0.1% of the Ultra 5’s average score of 46062. That proximity means that for predictable thread workloads, the Ultra 5 is effectively indistinguishable from those chips.

How does this translate into Thief? Because the title is not CPU-heavy, a single-core strong processor will unlock high FPS at lower resolutions and lower graphics settings. At 1080p with Low quality, the system delivers an average of 152 FPS. This high mark signals that the CPU can feed the engine efficiently for the time period, and it is not the stranglehold. Compare 1080p High, at 88 FPS; the GPU is the part that starts to limit. The CPU’s impeccable single‑thread performance (Cinebench R23 single‑core 2085) explains why it can achieve 152 FPS in the lightest-configuration row. The 14 threads also help background tasks, but in this game the difference from 152 to 84 (Ultra) at 1080p is attributable to GPU cost, not CPU.

In summary, the CPU’s role here is to supply the frame at the lower resolutions and to out touch in the ―scenarios where the GPU is underloading. The measured data indicates that the Core Ultra 5 235 does not bottleneck in any of the measured rows; all four settings at 1080p are above 84 FPS, which is above the refresh rate of many monitors. The CPU’s 24 MB of L3, 3 MB dedicated L2, and 192 KB per core L1 all help to move data quickly, but the main relief is that the CPU is more than enough to drive the older engine on the RX 5700 XT.

FAQ

Q: What is the average FPS at 1080p with Medium quality settings?

A: The dataset lists 115 FPS for that row, with a reported range of 98–133 FPS.

Q: Which resolution and preset yields the highest measured FPS?

A: The 1080p Low preset, which records an average of 152 FPS. No other row exceeds 115 FPS, giving this combo that ability.

Q: The fastest FPS at 4K across the settings?

A: At 2160p Low, the average is 61 FPS. Medium averages 47 and High/Ultra sit at 35 and 34 FPS, respectively.

Q: Is the GPU or CPU limiting factor at 4K?

A: Look at the resolution scaling between 1080p and 4K on the same preset. For High, the 1080p figure is 88 FPS, while the 4K figure is 35 FPS. This drop of 53 FPS is far larger than any difference seen at 1080p between low and high, pointing toward the GPU as the primary binding factor at 4K.

Q: How does the combo rank among all tested systems in Thief?

A: The overall combination rank is 2731 out of 3720 (comboRankInGame). That places it in the lower, non‑top percentile, but it is still within the playable range for 1440p.

How This Combo Ranks

The measured data places this exact CPU+GPU pair at rank 2731 out of 3720 tested combinations in Thief. To understand this, rank is 1‑based, so 2731 means that 2730 systems are better, and 989 are worse. That puts it in the 26.6% percentile of the test population: only 26.6% of systems are behind it. The ranking is determined by the average FPS across all measured configurations, so a system that excels at that particular game will rank higher. Thief is a stealth game with a known moderate graphical load, and the RX 5700 XT is a 2019 midrange card that clearly does not match the 3nm CPU’s capability when pushed to 4K. The CPU is a top ‑tier, but the GPU is the limiting factor, resulting in a rank that hovers in the lower quartile.

Notably, the combo’s measured FPS at 1080p with High settings (88) is only just above the 60 FPS target, and at 1440p High it drops to 60. At 4K only the Low preset reaches 61; all other settings are under 60. This pattern supports the lower placement in ranking, as most competing systems will deliver higher frame rates at higher resolution.

In contrast, the CPU alone has a score that places it in the top 11% (89.0 percentile), while the GPU sits at the 59.5 percentile of all GPUs. The gap between the two components explains why the overall rank is not better. If the GPU were to be upgraded, this system’s rank would rise dramatically, but as configured the GPU is the bottleneck.

Measured FPS Breakdown

The measurement table below presents every recorded FPS value from the dataset, with averages and, where available, the minimum and maximum frame rates.

| Resolution | Settings | Avg FPS | Min FPS | Max FPS |

|--------------|----------|-------|---------|---------|

| 3840×2160 | High | 35 | - | - |

| 3840×2160 | Low | 61 | - | - |

| 3840×2160 | Medium | 47 | 40 | 54 |

| 3840×2160 | Ultra | 34 | - | - |

| 2560×1440 | High | 60 | - | - |

| 2560×1440 | Low | 104 | - | - |

| 2560×1440 | Medium | 79 | 67 | 91 |

| 2560×1440 | Ultra | 57 | - | - |

| 1920×1080 | High | 88 | - | - |

| 1920×1080 | Low | 152 | - | - |

| 1920×1080 | Medium | 115 | 98 | 133 |

| 1920×1080 | Ultra | 84 | - | - |

All values come directly from the measured rows. Some presets do not list min/max numbers, so they are shown as "‑". The other clear pattern is that Medium is the only preset at 4K that has a defined min/max, and it shows 40–54 FPS, indicating that even at 2160p the 47 average does not dip below 40 on average, but it also cannot maintain a stable 60.

At 1440p, the middle ground, the average ranged from 57 (Ultra) to 104 (Low). The Medium preset offers 79 with a range 67–91, which is the tightest envelope of any 1440p row, suggesting a stable experience. At 1080p, Low 152 and Medium 115 both exceed 60, while Ultra 84 also remains above 60; all preset values at 1080p are ≥84, which is above the typical 60‑frame quench.

Resolution Scaling

The generic between 1080p and 4K is profound. For example, at High settings, the FPS falls from 88 at 1920×1080 to 35 at 3840×2160—a drop of 53 FPS. At Low settings, the plummet is from 152 to 61–a 91‑frame difference. The scaling from 1080p to 1440p is less dramatic; the High99 row moves from 88 to 60 (‑28), the Low from 152 to 104 (‑48), and Medium from 115 to 79 (‑36). The Ultra preset at 1440p is already at 57, which is near the margin to dip below 60; only 1080p Ultra is at 84.

These numbers point strongly to the GPU being the performance limiter once resolution rises. The CPU retains a great deal of headroom, because the loss between 1080p and 4K is far greater than what would be expected from CPU‑bound work. If, for example, the CPU were the constraint, the frame rate would scale more evenly; the further large differences indicate that the processor is not the saturating device. This is typical of a Khronos RAN 1.0 GPU like the RX 5700 XT, which, even at 2019, cannot exceed the structural wastage of pushing 4K without low- and ultra-quality reduction.

At 4K, every setting except Low falls below 60 FPS, suggesting that the GPU is at capacity. The 4K High and Ultra values (35 and 34) are essentially identical, implying that the GPU is already rendering everything at its near‑max, so the raise in preset load from High to Ultra has negligible impact. That is first‑sign of vGPU‑bound scenario.

Settings Recommendations

From the measured data, the following are the most viable presets for each resolution:

  • 1920×1080 – All presets are above 60 FPS. However, the Ultra preset yields 84 FPS, which is still high and offers the best visual richness. For the highest frame, Low (152) is available, but Ultra is recommended for quality if the monitor supports variable refresh.
  • 2560×1440 – Medium preset delivers 79 FPS with a min of 67, which is a strong balance. The High preset sits exactly at 60, but can stutter in busy scenes; the Medium 79 allows a small headroom. The Ultra preset at 57 is below 60 but might be acceptable for a slow‑paced stealth game; still, the data shows the Medium is the best combination of fps and quality.
  • 3840×2160 – At 4K, Low produces 61 FPS, just above the 60 criteria. Medium averages 47 with a 40–54 range, which is below the standard v-sync. High (35) and Ultra (34) are not playable for competitive‑focused players. Therefore, the recommendation is to use Low on 4K if the user insists on this resolution. For a more significant visual boost than Low while maintaining >60, they must step down to 1440p with High (60) or Ultra (57). The Usable best across all resolution is 1080p Ultra (84 FPS) or 1440p Medium (79 FPS) for smooth play.

The GPU’s 8 GB frame buffer is not a limiting factor at these settings, but the raw compute (9.754 TFLOPS FP32) and pixel fill rate of 121.9 GPixel/s become crucial when the scene is dense. The dataset does not show any memory‑related stutter since all rows are average FPS, but the performance curve indicates the GPU’s 20‑ALU architecture (2560 shading units) is working near its ceiling at 4K.

GPU Role

The AMD Radeon RX 5700 XT represents the RDNA 1.0 architecture, built on TSMC 7nm, with 10,300 million transistors on a 251 mm² die. It runs up to a 1905 MHz boost clock and features 8 GB of GDDR6 memory clocked at 1750 MHz (14 Gbps effective) on a 256‑bit bus. That configuration yields 448.0 GB/s of memory bandwidth. The GPU is no slouch; it launches from mid‑2019 and is still capable of beating many older cards. Its metrics for this system: PassMark G3D score of 16277, with a percentileVsAllGpus of 59; the nearest rivals include the AMD Radeon Pro 5600M (avgScore 16351, delta 0.1%) and the NVIDIA RTX PRO 6000 Blackwell (16408, delta -0.3%), indicating the 5700 XT sits right at the median of GPUs.

For a 2014 game like Thief, the RX 5700 XT is more than adequate at 1080p and 1440p, and only the 4K resolution causes it to hit its limit. The 8 GB VRAM is not an issue at these settings, but the resolution drives the 35 FPS at High 4K. The high compute power of 9.754 TFLOPS is the reason the card can still deliver 60–100+ FPS at 1440p Low, the engine’s complexity is low. The raw pixel fill rate of 121.9 GPixel/s helps the card’s ability to push fill‑rate‑heavier scenes even at High details.

The GPUs are the core limiting component at 4K, as evidenced by the similarity between the High and Ultra results (35 vs 34) and the fact that only the Low preset breaks 60 FPS. This but does not mean the card is obsolete; rather, it indicates that Thief’s engine scales very well with compute throughput and the 5700 XT’s 160 TMUs and 64 ROPs are enough for 1440p. On the whole, the graphics card provides all that is expected for 1080/1440p nostalgia, but at 4K requires the user to lower the preset to Low to remain within playable bounds. The future replacement would be a higher‑rung GPU, but for the present dataset, these measured frames are the performance you can expect from this solution.

Hardware Specifications

Intel Core Ultra 5 235

Cores / Threads 14 / 14
Base Clock 3400 MHz
Boost Clock 5000 MHz
TDP 65W
Socket Intel Socket 1851
View Full CPU Details →

AMD Radeon RX 5700 XT

VRAM 8 GB GDDR6
Base Clock —
Boost Clock 1905 MHz MHz
TDP 225 WW
View Full GPU Details →

FPS Benchmarks by Resolution & Settings

Performance data for Intel Core Ultra 5 235 + AMD Radeon RX 5700 XT in Thief

Resolution Settings Avg FPS
3840x2160 Low 61.0
3840x2160 Medium 47.0
3840x2160 High 35.0
3840x2160 Ultra 34.0
2560x1440 Low 104.0
2560x1440 Medium 79.0
2560x1440 High 60.0
2560x1440 Ultra 57.0
1920x1080 Low 152.0
1920x1080 Medium 115.0
1920x1080 High 88.0
1920x1080 Ultra 84.0

Thief with iUltra 5 235 + Other GPUs

See how Thief performs with the same CPU but different graphics cards.

Thief with RX 5700 XT + Other CPUs

See how Thief performs with the same GPU but different processors.

Other Games with iUltra 5 235 + RX 5700 XT

Explore FPS benchmarks for other games using this same hardware combination.