Steamworld Build

Steamworld Build is a city builder where simulation logic and population scale often dictate performance, but the measured results for this specific combination of AMD Ryzen 9 9950X3D and Intel Arc B390 paint a complex picture of bottlenecking. The data indicates a system where a top-tier 16-core processor is paired with an integrated graphics solution that severely limits frame rates, particularly at higher resolutions. The benchmark results show that this pairing ranks near the bottom of all tested combos, a direct consequence of the GPU’s limited throughput.

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

The AMD Ryzen 9 9950X3D is a 16-core, 32-thread processor based on the Zen 5 architecture, codenamed Granite Ridge. It operates with a base clock of 4.30 GHz and a boost clock of 5.70 GHz. With a substantial 128 MB of L3 cache, this CPU is designed for high-throughput workloads. Its benchmark scores are exceptional; it achieves a Cinebench R23 multicore score of 42018 and a single-core score of 2259. In Geekbench, it scores 26736 multicore and 3064 single-core. The processor’s Passmark multithread score is 70255, and its 3DMark max threads score is 17184. These figures place the CPU in the 95th percentile of all processors, indicating it is among the fastest available.

However, in Steamworld Build, the CPU’s role appears secondary to the GPU. The measured frame rates show little variation that can be attributed to CPU performance scaling. For instance, at 1080p with Low settings, the average FPS is 106, while at 1080p with High settings, it drops to 60. This drop is significant, but the absolute numbers suggest the CPU is not the limiting factor; if the CPU were the bottleneck, we would expect more consistent performance across settings, as lowering graphical fidelity would not affect the simulation thread. The data suggests that the CPU is far ahead of the GPU’s capabilities, leaving significant headroom unused.

The processor’s nearest rivals include the AMD Ryzen 7 PRO 9755, which has an average score of 75738, a mere 0.1% difference. Similarly, the AMD Ryzen 7 PRO 9755X3D scores 75716, also a 0.1% delta. The Intel Core Ultra 9 285 is 0.4% behind with a score of 75488. These tiny deltas confirm that the 9950X3D is at the apex of CPU performance. For a city builder like Steamworld Build, where large populations can strain simulation logic, this CPU’s 32 threads and high cache size could theoretically handle massive cities. Yet, the benchmark results show that the GPU fails to deliver playable frame rates at higher settings, meaning the CPU’s potential is never realized in this configuration.

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

The Intel Arc B390 is an integrated graphics processor (IGP) based on the Xe3-LPG architecture, built on a 3 nm process. It is part of the Arc Graphics-M generation for Panther Lake. The GPU has a base clock of 300 MHz and a boost clock of 2500 MHz. Critically, its memory is "System Shared," meaning it uses the system’s DDR5 memory, with bandwidth that is "System Dependent." This is a fundamental limitation for gaming performance. The GPU features 1536 shading units, 48 texture mapping units, and 24 raster output units, along with 12 ray tracing cores. Its theoretical peak performance is 7.680 TFLOPS for FP32, and it supports DirectX 12 Ultimate.

The GPU’s benchmark score is very low. In 3DMark Steel Nomad DX12, it scores just 1482, placing it in the 9th percentile of all GPUs. Its nearest rivals include the NVIDIA GeForce GT 520MX (score 1463, 1.3% slower), the NVIDIA GeForce 800M (score 1460, 1.5% slower), and the NVIDIA GeForce GT 625 OEM (score 1446, 2.5% slower). These are legacy, low-end parts, indicating the Arc B390 is not a gaming-capable discrete solution. The measured FPS in Steamworld Build confirms this. At 1080p Ultra, the average FPS is only 41, which is below the threshold for smooth gameplay. At 4K Ultra, the average drops to 14 FPS, making the game unplayable.

The GPU’s shared memory architecture is a critical bottleneck. Unlike a dedicated VRAM pool, the integrated GPU must compete with the CPU for memory bandwidth. In a simulation-heavy game like Steamworld Build, which requires constant data updates for the simulation, this contention likely causes severe performance degradation. The data shows that the GPU is the primary constraint, as lowering settings from Ultra to Low at 1080p increases FPS from 41 to 106, demonstrating that the rendering load is the limiting factor, not the CPU or the game’s simulation logic.

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

Given the GPU’s limitations, the choice of settings is critical for achieving a playable experience. The measured data provides clear guidance. At 1080p, the Low preset yields an average FPS of 106, which is excellent for a city builder, providing smooth scrolling and responsive interactions. The Medium preset at 1080p averages 81 FPS, with a minimum of 69 and a maximum of 93, which is also very playable and offers better visual fidelity. The High preset at 1080p drops to 60 FPS, which is borderline acceptable but may feel less fluid. The Ultra preset at 1080p averages 41 FPS, which is below the ideal threshold for a game that requires frequent camera movement and menu interaction.

At 1440p, the situation worsens. The Low preset averages 67 FPS, which is playable but noticeably lower than 1080p. Medium averages 51 FPS (min 44, max 59), which is on the lower edge of acceptability. High averages 38 FPS, and Ultra averages just 26 FPS, both of which are too low for a smooth experience. At 4K, all presets are below playable thresholds, with Low averaging 35 FPS, Medium 27 FPS, High 20 FPS, and Ultra 14 FPS.

Therefore, the best experience is achieved at 1080p with either Low or Medium settings. The Medium preset offers a 31% FPS reduction compared to Low (81 vs 106), but the visual upgrade may be worth it for some users. However, for a stable, high-refresh experience, the Low preset is the safest recommendation. The data suggests that the Ultra preset should be avoided entirely, as it provides less than half the frame rate of Low at any resolution. For users with 1440p monitors, the Low preset is the only viable option, while 4K is not a realistic target for this hardware combination.

How This Combo Ranks — use comboRankInGame vs other tested combos

This specific combination of CPU and GPU ranks 2573 out of 2924 tested combos in Steamworld Build. This places it in the bottom 12% of all tested systems. This low rank is almost entirely attributable to the GPU. The CPU is a top-tier part, ranking in the 95th percentile of all processors, but the GPU is in the 9th percentile. This creates a massive performance imbalance. The combo rank reflects the overall system performance, which is dragged down by the GPU.

The data shows that the CPU’s performance headroom is wasted in this configuration. For instance, the CPU’s Cinebench R23 multicore score of 42018 is more than adequate for any game, but the GPU’s 3DMark score of 1482 is insufficient for modern gaming. The 2573rd rank out of 2924 means that most other systems, even those with much weaker CPUs, will outperform this combo in Steamworld Build because they have a discrete GPU with dedicated memory. The benchmark results indicate that the CPU contributes little to the final frame rate, as the GPU saturates well before the CPU can become a limiting factor. This is a classic case of a bottleneck, where the system’s weakest link determines the overall performance.

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

The resolution scaling data clearly identifies the GPU as the limiting component. At Low settings, the average FPS drops from 106 at 1080p to 67 at 1440p, a 37% reduction. From 1440p to 4K, it drops from 67 to 35, a 48% reduction. This pattern of performance degradation with increasing resolution is typical of a GPU bottleneck. When the resolution increases, the number of pixels the GPU must render grows exponentially, and the GPU’s fill rate and memory bandwidth become the primary constraints. The CPU’s workload, which is primarily related to game logic and simulation, does not scale with resolution.

The drop from 1080p Low to 4K Low results in a 71 FPS difference (106 to 35), which is a 67% reduction. This is a stark indicator of the GPU’s limited capabilities. If the CPU were the bottleneck, we would expect the FPS to remain relatively flat across resolutions, as the CPU’s workload would be unchanged. However, the data shows a steep decline, confirming that the Intel Arc B390’s rendering capabilities are overwhelmed at higher resolutions. The fact that the GPU uses system shared memory further exacerbates this issue, as the memory bandwidth available to the GPU is dependent on the system’s memory controller and other workloads. The data indicates that to maintain a playable frame rate, users must remain at 1080p, and even then, they must lower settings to achieve smooth performance.

FAQ

Q: Is the AMD Ryzen 9 9950X3D a good match for the Intel Arc B390 in this game?

A: No. The benchmark data shows a severe imbalance. The CPU is in the 95th percentile of processors, while the GPU is in the 9th percentile. The combo ranks 2573 out of 2924, indicating the GPU severely limits the system.

Q: What is the best settings preset for this hardware at 1080p?

A: The Low preset provides an average of 106 FPS, which is the smoothest experience. The Medium preset averages 81 FPS and is also viable, but High (60 FPS) and Ultra (41 FPS) are not recommended for consistent fluidity.

Q: Can this system run Steamworld Build at 4K?

A: No. At 4K, the best average FPS is 35 on Low settings, with Medium at 27 FPS, High at 20 FPS, and Ultra at 14 FPS. All of these are below playable thresholds.

Q: How much faster is the 9950X3D compared to its nearest rival?

A: The nearest rival, the AMD Ryzen 7 PRO 9755, has an average score of 75738, which is a 0.1% difference. The Intel Core Ultra 9 285 is 0.4% slower.

Q: What does the GPU’s "System Shared" memory mean for performance?

A: It means the GPU uses the system’s DDR5 memory instead of dedicated VRAM. The bandwidth is "System Dependent," which is a critical bottleneck, as the GPU must compete with the CPU for memory access.

Q: What is the frame rate drop from 1080p to 4K at Medium settings?

A: At Medium settings, the average FPS drops from 81 at 1080p to 51 at 1440p, and then to 27 at 4K. This represents a 67% total reduction from 1080p to 4K.

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

The measured FPS data provides a comprehensive view of the system’s performance across all settings and resolutions.

At 1920x1080, the system achieves its best results. The Low preset averages 106 FPS. The Medium preset averages 81 FPS, with a minimum of 69 and a maximum of 93. The High preset averages 60 FPS. The Ultra preset averages 41 FPS.

At 2560x1440, performance drops significantly. The Low preset averages 67 FPS. The Medium preset averages 51 FPS, with a minimum of 44 and a maximum of 59. The High preset averages 38 FPS. The Ultra preset averages 26 FPS.

At 3840x2160, the system struggles. The Low preset averages 35 FPS. The Medium preset averages 27 FPS, with a minimum of 23 and a maximum of 31. The High preset averages 20 FPS. The Ultra preset averages 14 FPS.

These numbers illustrate a clear trend: the system is only playable at 1080p. The gap between Low and Ultra at 1080p is 65 FPS (106 to 41), which highlights the GPU’s sensitivity to graphical settings. The data also shows that the minimum FPS is only recorded for the Medium preset at each resolution, suggesting that the frame pacing may be inconsistent at other settings, potentially leading to stutter. The benchmark results are definitive: this is a 1080p, low-to-medium settings configuration.