Steamworld Build

CPU Role — cores, clocks, and how they relate to this game's results

The Intel Core Ultra 7 265 is a 20-core, 20-thread desktop processor built on the Arrow Lake architecture and manufactured on a 3 nm process by TSMC. Its base clock is 2.40 GHz with a boost clock of 5.30 GHz, and it carries 30 MB of shared L3 cache alongside 3 MB of L2 per core. The processor's 65 W TDP and dual-channel DDR5 memory support with 102.4 GB/s bandwidth position it as a modern high-end desktop part, though its role in Steamworld Build is more nuanced than raw throughput.

Benchmark data places this CPU in the 93rd percentile among all tested processors. Its Cinebench R23 multi-core score is 42,216, while the single-core score is 5,960. PassMark multi-thread performance is 49,682, and single-thread performance is 4,689. These figures indicate substantial headroom for simulation-heavy tasks, which is relevant because city builders like Steamworld Build often stress CPU logic for pathfinding, resource distribution, and population simulation.

In this game's measured results, the CPU's influence is most visible at lower resolutions where frame rates are extremely high. At 1080p Low, the combo reaches 255 FPS average — a figure that suggests the CPU is not a bottleneck at that setting, as the GPU can push frames well beyond typical display refresh rates. However, the gap between settings at 1080p is telling: dropping from Ultra (99 FPS) to Low (255 FPS) is a 156 FPS swing, which implies the GPU is handling most of the visual load while the CPU maintains consistent simulation updates.

The CPU's 20 threads are all physical — no hyperthreading — which is a design choice that favors predictable scheduling for game engines. In Steamworld Build, where many entities update simultaneously, this can be advantageous. The processor's nearest rivals include the AMD EPYC 4464P (average score 64,823, delta -0.3%), the Intel Core Ultra 7 265F (64,438, delta +0.3%), and the AMD EPYC 7343 (64,202, delta +0.7%). These deltas are tiny, indicating that the 265 sits in a tightly clustered performance band. The data shows the 265 is effectively within 1% of these server-class parts in aggregate benchmarks, but for Steamworld Build specifically, the game's frame rate behavior across resolutions points more heavily to GPU scaling than to CPU headroom.

FAQ — 4-6 Q&A pairs answerable from FACT PACK data

Q: What is the highest average frame rate this combo achieves in Steamworld Build?

A: The highest measured average is 255 FPS at 1920x1080 with Low settings. This is the only configuration exceeding 250 FPS in the dataset.

Q: How does the Ultra preset compare to High at 1440p?

A: At 2560x1440, Ultra produces 63 FPS average, while High produces 93 FPS. That is a 30 FPS difference, or roughly 48% higher frame rate on High compared to Ultra.

Q: Is the CPU or GPU more likely to limit performance at 4K Ultra?

A: At 3840x2160 Ultra, the average is 33 FPS. Given that the CPU's Cinebench R23 multi-core score is 42,216 (93rd percentile), the data suggests the GPU is the primary limiter at this resolution and preset, since the CPU has ample headroom for simulation workloads.

Q: What is the only setting with recorded minimum and maximum frame times?

A: The Medium preset at all three resolutions includes min and max values: 1080p (min 166, max 225), 1440p (min 106, max 143), and 4K (min 56, max 76). No other preset has complete min/max data.

Q: How does the combo rank among all tested combinations for this game?

A: The combo rank is 1,199 out of 2,924 tested combos, placing it in the upper portion of the distribution — roughly the 59th percentile overall.

Q: Does the CPU's performance percentile predict consistent frame rates across resolutions?

A: Not directly. The CPU sits in the 93rd percentile, but frame rates drop by more than half from 1080p Low (255 FPS) to 4K Low (86 FPS). This indicates that resolution scaling is dominated by GPU workload, not CPU capability.

Resolution Scaling — how FPS drops from 1080p to 4K and what it says about the limiting component

The measured FPS data reveals a clear pattern of resolution-driven scaling that identifies the GPU as the primary constraint in Steamworld Build. Moving from 1920x1080 to 3840x2160 at any given setting produces a dramatic reduction in frame rate, consistent with a rendering-bound workload.

At Low settings, the drop is from 255 FPS at 1080p to 162 FPS at 1440p, then to 86 FPS at 4K. That is a 66% reduction from 1080p to 4K. At Medium settings, the progression is 196 FPS → 124 FPS → 66 FPS, a 66% drop. At High settings, it is 146 FPS → 93 FPS → 49 FPS, a 66% drop. At Ultra settings, the progression is 99 FPS → 63 FPS → 33 FPS, a 67% drop. The consistency of this ~66-67% reduction across all four presets is striking.

This uniform scaling strongly implies that the limiting component scales predictably with pixel count — which is the GPU. If the CPU were the bottleneck, we would expect to see diminishing frame rate differences at higher resolutions (since the CPU workload per frame stays roughly constant), but instead the frame rate drops proportionally to the number of pixels. The 4K resolution has exactly four times the pixels of 1080p, and the frame rates are consistently about one-third of the 1080p values, not one-fourth. This indicates there is some overhead that doesn't scale perfectly with pixel count, but the dominant effect is clearly GPU-bound rendering.

The fact that even at Low settings the 4K frame rate (86 FPS) is below the 1080p Ultra frame rate (99 FPS) reinforces the interpretation: resolution is the dominant variable, and the GPU's 8 GB GDDR6 memory and 448.0 GB/s bandwidth are being taxed by the higher pixel throughput. The CPU, with its 20 cores and high single-thread score, does not show symptoms of being the limiting factor — it maintains enough throughput to feed the GPU at every resolution tested.

Measured FPS Breakdown — resolution by resolution, settings by settings, exact numbers

The dataset provides average FPS for all twelve combinations of four settings and three resolutions, with min/max values only for Medium presets.

1920x1080:

• Low: 255 FPS average
• Medium: 196 FPS average (min 166, max 225)
• High: 146 FPS average
• Ultra: 99 FPS average

2560x1440:

• Low: 162 FPS average
• Medium: 124 FPS average (min 106, max 143)
• High: 93 FPS average
• Ultra: 63 FPS average

3840x2160:

• Low: 86 FPS average
• Medium: 66 FPS average (min 56, max 76)
• High: 49 FPS average
• Ultra: 33 FPS average

The Medium preset's min/max ranges provide insight into frame time consistency. At 1080p, the range is 166 to 225 FPS — a spread of 59 FPS. At 1440p, the range narrows to 106 to 143 FPS — a spread of 37 FPS. At 4K, the range is 56 to 76 FPS — a spread of 20 FPS. The absolute spread shrinks as resolution increases, but the relative spread (max/min ratio) grows: 225/166 = 1.36 at 1080p, 143/106 = 1.35 at 1440p, and 76/56 = 1.36 at 4K. This consistency suggests the game's frame pacing is stable relative to the average, regardless of resolution.

The step from High to Ultra is consistent across resolutions. At 1080p, the drop is 146 to 99 FPS (32% reduction). At 1440p, it is 93 to 63 FPS (32% reduction). At 4K, it is 49 to 33 FPS (33% reduction). The Ultra preset imposes a near-identical relative cost at every resolution, indicating that the extra visual features scale linearly with GPU load. Conversely, the step from Low to Medium costs 196/255 = 77% at 1080p, 124/162 = 77% at 1440p, and 66/86 = 77% at 4K — again, perfectly uniform scaling.

GPU Role — VRAM, clocks, and how they relate to this game's results

The NVIDIA GeForce RTX 3070 is an Ampere-architecture GPU built on Samsung's 8 nm process, featuring the GA104 chip with 17,400 million transistors on a 392 mm² die. It has 5,888 shading units, 184 texture mapping units, and 96 raster output units. The card's boost clock is 1725 MHz, with a base of 1500 MHz. Memory is 8 GB of GDDR6 on a 256-bit bus, delivering 448.0 GB/s bandwidth. The GPU's FP32 performance is 20.31 TFLOPS, and it supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4.

The RTX 3070 sits in the 61st percentile among all GPUs in the benchmark database. Its PassMark G3D score is 22,214, and its 3DMark Steel Nomad DX12 score is 3,162. The nearest rivals are the AMD Radeon RX 7600 XT (average score 17,083, delta +0.7%), the NVIDIA GeForce GTX 690 (17,037, delta +1%), and the AMD Radeon HD 7970M (17,019, delta +1.1%). The RTX 3070's average benchmark score is 17,208, placing it within 1.5% of these competitors — a tightly grouped performance tier.

For Steamworld Build specifically, the GPU's role is dominant. The measured frame rates demonstrate that the RTX 3070 is the limiting component across all settings and resolutions. At 4K Ultra, the 33 FPS average is below playable thresholds for many users, while at 1080p Low, the 255 FPS ceiling suggests the GPU is capable of much higher output when pixel count is low. The 8 GB VRAM capacity is a relevant consideration for a city builder where texture detail and draw distances scale with population density, though the benchmark data does not isolate VRAM usage.

The uniform scaling of FPS with resolution — approximately 66-67% reduction from 1080p to 4K across all presets — is consistent with a GPU that is bandwidth-bound at higher resolutions. The 448.0 GB/s memory bandwidth and 256-bit bus are competitive for the era, but 4K pixel throughput demands significantly more fill rate. The pixel rate of 165.6 GPixel/s and texture rate of 317.4 GTexel/s are the specifications that likely govern the 4K performance ceiling, which the data shows is around 86 FPS at Low settings.

Settings Recommendations — which preset gives the best experience per the measured rows

Given the measured data, the Medium preset offers the most balanced experience across all three resolutions. At 1080p, Medium delivers 196 FPS average with a min of 166 and max of 225 — comfortably above any display refresh rate and with a tight frame time distribution. At 1440p, Medium reaches 124 FPS average (min 106, max 143), which is well-suited for high-refresh-rate monitors. At 4K, Medium provides 66 FPS average (min 56, max 76), which is the only preset that maintains playable performance at 4K while still offering meaningful visual uplift over Low.

High settings are a reasonable choice for users prioritizing visual fidelity at 1080p or 1440p, where the 146 FPS and 93 FPS averages respectively are still strong. However, at 4K High the 49 FPS average dips below the 60 FPS threshold, making it less ideal for smooth gameplay. Ultra settings should be avoided at 4K entirely — the 33 FPS average is likely to feel stuttery — and even at 1440p the 63 FPS average is marginal for fast-paced interactions.

Low settings are only recommended for maximum frame rate at 1080p, where the 255 FPS average is the highest in the dataset. At 1440p Low (162 FPS) and 4K Low (86 FPS), the frame rates are high but the visual trade-off may not be worth it, especially since Medium provides a substantial quality improvement for a modest FPS cost: at 1080p, Medium is 196 FPS vs 255 FPS Low (23% reduction); at 1440p, Medium is 124 FPS vs 162 FPS Low (23% reduction); at 4K, Medium is 66 FPS vs 86 FPS Low (23% reduction).

The most striking recommendation from the data is that Medium provides consistent 77% of Low's frame rate at every resolution, making it a predictable and reliable choice. The min/max data for Medium also shows stable frame pacing, with relative spreads of 1.35-1.36 across all resolutions, indicating no resolution-specific stuttering issues.

How This Combo Ranks — use comboRankInGame vs other tested combos

The Intel Core Ultra 7 265 paired with the NVIDIA GeForce RTX 3070 achieves a combo rank of 1,199 out of 2,924 tested combinations in Steamworld Build. This places the combo in the 59th percentile of all tested systems — meaning roughly 41% of tested combos perform better, and 59% perform worse.

This rank is notable given the individual component positions. The CPU sits in the 93rd percentile among all CPUs, while the GPU sits in the 61st percentile among all GPUs. The combo's overall rank of 59th percentile is much closer to the GPU's percentile than the CPU's, which is consistent with the resolution scaling analysis showing the GPU as the limiting factor. In a game where CPU simulation matters more, we would expect the combo rank to align more closely with the CPU's 93rd percentile.

The gap between the CPU's high percentile and the combo's mid-tier rank suggests that Steamworld Build's performance envelope is constrained by graphics processing rather than logic processing. This interpretation is supported by the fact that the RTX 3070's nearest rivals — the RX 7600 XT, GTX 690, and HD 7970M — all have average benchmark scores within 1.5% of the 3070, indicating that the GPU tier is competitive but not top-tier.

For context, the combo rank of 1,199 out of 2,924 means there are 1,198 combos that perform better and 1,725 that perform worse. The distribution is not uniform — the fact that the combo sits just above the midpoint suggests that many high-end CPU + mid-range GPU combos land in a similar performance band. The data shows that upgrading the GPU would likely improve the combo rank more than upgrading the CPU, since the current CPU has significant headroom as evidenced by its 93rd percentile standing and strong Cinebench scores.

The measured FPS data supports this ranking logic. At 1080p Ultra, the combo achieves 99 FPS, which is respectable but not exceptional for a high-end system. At 4K Ultra, the 33 FPS average is below the 60 FPS target that many users consider the baseline for smooth gameplay. These numbers are consistent with a combo rank that is solidly mid-pack rather than top-tier, reflecting the GPU's 61st percentile position as the primary determinant of the system's overall standing in this game.