Control

Control on the AMD Ryzen 7 5700G + Intel Arc B370 configuration presents a stark picture of performance limits, with the measured frame rates indicating that this combination is heavily constrained by the graphics subsystem across all tested resolutions and settings. The data shows a system that is capable of producing playable results only at the lowest settings and resolution, while higher demands quickly push it into single-digit and low-twenties territory that falls short of smooth gameplay standards.

Resolution Scaling

The progression from 1920x1080 to 3840x2160 reveals a steep and consistent performance penalty that scales almost linearly with pixel count, pointing directly at the GPU as the primary bottleneck. At High settings, the average frame rate drops from 37 FPS at 1080p to 28 FPS at 1440p, a reduction of roughly 24%, and then further to 19 FPS at 4K, which represents a 49% decline from the 1080p baseline. This pattern is not an anomaly; it repeats across every settings preset, indicating that the Intel Arc B370's rendering throughput is exhausted well before the CPU can even begin to strain.

Examining the Low settings row makes the scaling trend even more explicit. The system delivers 59 FPS at 1080p, which drops to 45 FPS at 1440p, and finally lands at 30 FPS at 4K. The relative drop from 1080p to 4K here is about 49%, nearly identical to the High settings percentage, which suggests that the resolution penalty is consistent regardless of the visual load. This uniformity is a classic symptom of a GPU-bound scenario, where the frame rate is determined by the sheer number of pixels being shaded and rasterized, not by the complexity of the scene or the settings applied.

The Medium settings data, which includes minimum and maximum frame times, further corroborates this analysis. At 1080p Medium, the average is 42 FPS with a minimum of 36 FPS and a maximum of 48 FPS. Moving to 1440p Medium, the average falls to 32 FPS, with the minimum dropping to 27 FPS. At 4K Medium, the average collapses to 21 FPS, and the minimum plunges to 18 FPS. The fact that the minimum frame rates at 1440p and 4K are so close to the averages indicates that the GPU is consistently saturated, leaving little headroom for variance; the system is not stuttering intermittently but rather running at a sustained, low ceiling.

What this scaling pattern says about the limiting component is unambiguous. The CPU, an AMD Ryzen 7 5700G with 8 cores and 16 threads, is a capable processor that sits in the 79th percentile of all CPUs, yet it is never given a chance to become the bottleneck. The frame rate deltas between resolutions are far too large to be caused by CPU limitations, which typically manifest as similar FPS across resolutions when the CPU is the constraint. Instead, the data shows the Arc B370, which ranks in the 5th percentile of all GPUs, is the definitive limiting factor. The system is effectively pushing a high-end processor through a severely underpowered graphics pipeline, and the resolution scaling confirms that the GPU's fill rate and shading capacity are the sole determinants of performance.

GPU Role

The Intel Arc B370 is an integrated graphics solution based on the Xe3-LPG architecture, and its role in this game is defined by a fundamental resource deficiency. The GPU operates with a base clock of 300 MHz and a boost clock of 2400 MHz, but these clocks are rendered almost irrelevant by the fact that it relies on system shared memory. The memory size, type, bus width, and bandwidth are all listed as "System Shared" or "System Dependent," meaning the GPU has no dedicated VRAM and must compete with the CPU for access to the same DDR4 memory pool, which is limited to a dual-channel bandwidth of 51.2 GB/s.

This memory configuration is the likely culprit behind the low frame rates observed. With only 1280 shading units, 40 texture mapping units, and 20 raster output units, the Arc B370's raw compute is modest, but the shared memory bottleneck is what truly cripples performance in a game like Control, which is known for heavy texture streaming and complex visual effects. The measured pixel rate of 48.00 GPixel/s and texture rate of 96.00 GTexel/s are theoretical maximums that are unlikely to be achieved in practice when the memory subsystem is starved for bandwidth. The 6.144 TFLOPS of FP32 compute is a nominal figure that does not translate into playable frame rates because the data cannot be fed to the compute units fast enough.

The GPU's benchmark scores reflect this reality. In the 3DMark Steel Nomad DX12 test, the Arc B370 scores 1184, placing it in the 5th percentile of all GPUs. Its nearest rivals in this benchmark are the ATI Mobility Radeon HD 5570 with a score of 1186 (a delta of -0.2%), the ATI Radeon HD 5770 with 1190 (delta -0.5%), and the AMD Radeon HD 7650M with 1192 (delta -0.7%). These are all legacy parts from over a decade ago, and the fact that the Arc B370 lands squarely in their company confirms that its performance class is that of an entry-level iGPU from a previous era. The data shows that the GPU's clocks and architecture are not the issue; the issue is that the hardware simply does not have the execution resources or memory bandwidth to handle a modern action-adventure title at playable settings.

The impact of the shared memory is most evident when looking at the FPS deltas between settings. At 1080p, dropping from Ultra to Low yields a jump from 32 FPS to 59 FPS, an 84% improvement. This suggests that the GPU is highly sensitive to the reduction in texture quality and shading complexity, which in turn reduces the amount of data that needs to be pulled from system memory. The same pattern holds at 1440p, where Low (45 FPS) is 80% faster than Ultra (25 FPS), and at 4K, where Low (30 FPS) is 76% faster than Ultra (17 FPS). This consistent scaling indicates that the GPU's performance is directly tied to the memory traffic generated by the game's assets, reinforcing the conclusion that VRAM capacity and bandwidth are the critical constraints.

How This Combo Ranks

Within the benchmark database, this specific pairing of the AMD Ryzen 7 5700G and the Intel Arc B370 ranks 2657 out of 2888 tested combos for Control. This places it in the bottom 8% of all configurations, which is a poor result that aligns with the GPU's low percentile ranking. The combo rank is a holistic measure that accounts for the interaction between the CPU and GPU, and in this case, it highlights a significant mismatch in capabilities.

The CPU component of this combo is not the problem. The Ryzen 7 5700G has an average benchmark score of 27051, which places it in the 79th percentile of all CPUs. Its nearest rivals include the Intel Core i9-9900K (score 27097, delta -0.2%), the AMD Ryzen AI 7 445 (score 26936, delta 0.4%), the Intel Core i5-14400T (score 27166, delta -0.4%), and the Intel Core i7-1370P (score 26900, delta 0.6%). These deltas are all within a fraction of a percent, indicating that the 5700G is performance-competitive with a range of modern and previous-generation processors. In synthetic tests, it scores 12830.5 in Cinebench R23 multi-core and 1494 in single-core, which are respectable figures that show it can handle demanding workloads.

However, when paired with the Arc B370, the CPU's potential is entirely wasted. The GPU's 5th percentile ranking drags the entire combo down to the 2657th position. The data shows that the combo rank is almost entirely determined by the GPU, as the CPU's nearest rivals in the overall benchmark rankings do not appear in the game-specific ranking context. This is not a balanced system; it is a case where a mid-range desktop processor is shackled to a graphics solution that is barely capable of running a game from 2019 at low settings. The rank of 2657 out of 2888 is a clear indictment of the GPU's role, and it suggests that any other modern discrete GPU, even a low-end one, would yield a significantly better combo ranking.

Settings Recommendations

Given the measured FPS data, the only viable settings preset for this configuration is Low, and even then, only at 1920x1080. At 1080p Low, the system achieves an average of 59 FPS, which is the only row in the entire dataset that approaches a smooth 60 FPS target. This is the recommended configuration for any playable experience, as it provides a frame rate that is within the realm of acceptable responsiveness for an action-adventure game.

Every other settings and resolution combination falls short of this threshold. At 1080p Medium, the average drops to 42 FPS, which is playable but not ideal, and the minimum of 36 FPS indicates that there will be noticeable hitches during intense scenes. At 1080p High, the average falls to 37 FPS, and at Ultra, it drops to 32 FPS. These are all below the 60 FPS target and would result in a less fluid experience, but they are at least technically playable in the sense that the game is running.

The situation becomes untenable at higher resolutions. At 1440p, the best result is 45 FPS on Low, which is marginal, and the Medium, High, and Ultra presets all fall below 32 FPS. At 4K, no preset achieves even 30 FPS, with the best being 30 FPS on Low and the worst being 17 FPS on Ultra. For a game like Control, which features fast-paced combat and physics-based environments, frame rates below 30 FPS are generally considered unplayable due to input lag and visual stutter.

Therefore, the measured rows clearly indicate that the intended use case for this combo is 1080p gaming at Low settings. The user should accept that the game will not look its best, but it will run at a frame rate that is functional. If the user insists on higher visual fidelity, the 1080p Medium preset at 42 FPS is the next best option, but it comes with a 29% performance penalty compared to Low. There is no scenario in which 1440p or 4K is recommended for this game, as the GPU simply does not have the resources to maintain playable frame rates at those resolutions, regardless of the settings used.

Measured FPS Breakdown

The complete set of measured FPS data provides a granular view of the system's performance envelope. At 1920x1080, the results are as follows: Low settings yield an average of 59 FPS, Medium yields 42 FPS with a minimum of 36 FPS and a maximum of 48 FPS, High yields 37 FPS, and Ultra yields 32 FPS. The progression from Low to Ultra shows a steady decline of 27 FPS, or approximately 46%, which demonstrates the significant cost of enabling higher visual features on this GPU.

At 2560x1440, the performance drops across the board. Low settings produce an average of 45 FPS, Medium produces 32 FPS with a minimum of 27 FPS and a maximum of 37 FPS, High produces 28 FPS, and Ultra produces 25 FPS. The delta between Low and Ultra is 20 FPS, a 44% reduction from the Low baseline. Notably, the 1440p Low result (45 FPS) is only slightly better than the 1080p High result (37 FPS), which illustrates that the GPU's resolution scaling is more punishing than its settings scaling.

At 3840x2160, the system is clearly out of its depth. Low settings average 30 FPS, Medium averages 21 FPS with a minimum of 18 FPS and a maximum of 25 FPS, High averages 19 FPS, and Ultra averages 17 FPS. The 4K Low result is the only one that hits the 30 FPS mark, and it is still below the playable threshold for most users. The spread between Low and Ultra at 4K is only 13 FPS, which suggests that the GPU is so heavily pixel-bound that the settings have a diminished relative impact; it is struggling to fill the screen regardless of the visual complexity.

The Medium settings data, which is the only preset with full min/max statistics, shows an interesting pattern. At 1080p, the average is 42 FPS with a 36 FPS minimum and 48 FPS maximum, indicating a 12 FPS frame time variance. At 1440p, the average is 32 FPS with a 27 FPS minimum and 37 FPS maximum, a 10 FPS variance. At 4K, the average is 21 FPS with an 18 FPS minimum and 25 FPS maximum, a 7 FPS variance. This tightening of the frame time variance as resolution increases is a sign that the GPU is becoming more consistently saturated; at 4K, it is always working at its maximum capacity, leaving little room for fluctuations. In contrast, at 1080p, there is more headroom, which allows for occasional bursts of higher performance but also more pronounced dips.

CPU Role

The AMD Ryzen 7 5700G plays a supporting role in this configuration, and the benchmark data indicates that it is more than capable of meeting the demands of Control, but its performance is never fully utilized. The CPU is based on the Zen 3 architecture and codename Cezanne, built on a 7 nm process by TSMC. It features 8 cores and 16 threads, with a base clock of 3.80 GHz and a boost clock of 4.60 GHz. This is a modern, multi-threaded processor that scores 12830.5 in Cinebench R23 multi-core and 1494 in single-core, placing it in the 79th percentile of all CPUs.

The CPU's nearest rivals in the overall benchmark rankings are all comparable or slightly faster parts. The Intel Core i9-9900K scores 27097, which is 0.2% higher than the 5700G's 27051. The AMD Ryzen AI 7 445 scores 26936, which is 0.4% lower. The Intel Core i5-14400T scores 27166, 0.4% higher, and the Intel Core i7-1370P scores 26900, 0.6% lower. These deltas are all within statistical noise, confirming that the 5700G is a solid mid-range performer that should not bottleneck any reasonably paired GPU.

However, in this specific combo, the CPU is effectively idle. The GPU's performance ceiling is so low that the CPU never has to work hard to feed it frames. The measured FPS at 1080p Low (59 FPS) is the only scenario where the CPU might be approaching a relevant role, but even then, the GPU is likely still the limiting factor. The CPU's 16 threads and 16 MB of L3 cache are more than sufficient for a game like Control, which typically scales well up to 8 cores, but they are irrelevant when the GPU can only render 30-40 frames per second.

The data shows that the CPU's role is to provide a stable platform, which it does admirably, but it cannot compensate for the GPU's deficiencies. If a more powerful discrete GPU were paired with this CPU, the 5700G would likely be able to push much higher frame rates, given its 79th percentile ranking. But as it stands, the CPU is a bystander in this benchmark, and the results are almost entirely a reflection of the Intel Arc B370's limited capabilities. The CPU's 65 W TDP and 10,700 million transistors are indicative of a well-rounded processor, but they do nothing to alter the fundamental reality that this combo is GPU-bound in every measurable scenario.