Inside

Inside is a 2016 puzzle platformer that, on paper, should not challenge a modern flagship mobile CPU paired with a top-tier desktop GPU. Yet the measured frame rates across resolutions and settings reveal a title that is surprisingly demanding at its upper bounds, while also showing distinct scaling behaviors that highlight the balance between the Intel Core Ultra 9 185H and the NVIDIA GeForce RTX 5090. The data shows a configuration capable of extreme performance at 1080p, but one that encounters a hard ceiling at 4K Ultra that is more about engine and driver limits than raw hardware throughput.

Resolution Scaling

The frame rate drop from 1080p to 4K is steep and non-linear, varying significantly by preset. At Low settings, the average FPS falls from 337 at 1920x1080 to 218 at 3840x2160, a reduction of 119 FPS or roughly 35%. This is a substantial hit, but the absolute numbers remain high enough to suggest the GPU is still the primary driver. At Medium settings, the drop is less severe in absolute terms but similar in percentage: from 311 FPS at 1080p to 175 FPS at 4K, a reduction of 136 FPS (about 44%). The 1440p results sit almost exactly midway, with 302 FPS at Low and 278 FPS at Medium, indicating a smooth scaling curve that is heavily influenced by pixel count.

The High and Ultra presets tell a different story. At High, the 1080p result is 241 FPS, which drops to 174 FPS at 1440p and then to 109 FPS at 4K. The incremental penalty from 1440p to 4K is 65 FPS, which is proportionally larger than the drop from 1080p to 1440p (67 FPS). This suggests that at High settings, the GPU is beginning to hit a memory bandwidth or shading throughput limit as resolution increases. At Ultra, the trend is even more pronounced: 230 FPS at 1080p, 165 FPS at 1440p, and only 104 FPS at 4K. The scaling from 1080p to 4K at Ultra is a 126 FPS loss, or about 55%, which is a significant degradation.

The limiting component is clear from the data. The RTX 5090, with its 32 GB of GDDR7 memory and 512-bit bus, is not memory-starved at any resolution. The bottleneck at 4K Ultra is more likely the game's engine, which was designed in 2016 and may not be optimized for the massive parallelism of a Blackwell 2.0 GPU. The fact that 4K Ultra (104 FPS) is only 5 FPS lower than 4K High (109 FPS) while 4K Medium (175 FPS) is vastly higher indicates a non-linear performance cliff between Medium and High presets, not a smooth degradation. This points to a specific rendering feature enabled at High and above that disproportionately impacts the GPU at higher resolutions, rather than a simple pixel-fill limitation.

Settings Recommendations

The measured rows provide a clear hierarchy of performance versus visual fidelity. For players seeking the highest frame rates, Low settings at 1080p deliver 337 FPS, which is more than sufficient for any display. However, the jump to Medium at 1080p yields 311 FPS, a reduction of only 26 FPS (about 8%), while likely offering a substantial visual improvement. This makes Medium the optimal preset for 1080p, as it retains near-maximum performance while avoiding the steep penalties of High (241 FPS) and Ultra (230 FPS) which cost 70 and 81 FPS respectively.

At 1440p, the recommendation shifts. Medium produces 278 FPS, which is still exceptionally high. High drops to 174 FPS, a 104 FPS penalty that is hard to justify for a puzzle platformer. Ultra at 165 FPS is only 9 FPS lower than High, suggesting that the Ultra preset's additional effects are less costly than the jump from Medium to High. For a balanced experience at 1440p, Medium is again the clear winner, as it sits 104 FPS above High and 113 FPS above Ultra. The data indicates that High and Ultra are not well-optimized at this resolution, likely due to the same feature cliff seen at 4K.

At 4K, the situation becomes more nuanced. Medium delivers 175 FPS, which is playable on high-refresh displays. High and Ultra drop to 109 FPS and 104 FPS respectively, which are still smooth but represent a significant performance cost. Given that the difference between High and Ultra at 4K is only 5 FPS, there is no reason to choose High over Ultra if visual quality is the priority. The best experience per the measured rows at 4K is Medium, as it provides a 66 FPS advantage over High while likely avoiding the problematic rendering path. For users with 120Hz or 144Hz 4K displays, Medium is the only preset that consistently exceeds 144 FPS, making it the definitive choice. Low settings at 4K (218 FPS) are unnecessary unless targeting very high refresh rates, as the visual degradation is unlikely to be worth the 43 FPS gain over Medium.

GPU Role

The NVIDIA GeForce RTX 5090 is the dominant component in this configuration, and the benchmark results reflect its raw capabilities. The GPU's 32 GB of GDDR7 memory at 1.79 TB/s bandwidth is never a limiting factor in a 2016 game like Inside, even at 4K. The data shows that the GPU is capable of pushing 337 FPS at 1080p Low, which is far beyond any display's refresh rate. The GPU's boost clock of 2407 MHz and its 104.8 TFLOPS of FP32 performance are more than adequate for this title's simple geometry and lighting.

However, the GPU's role becomes more complex at higher presets. The performance cliff between Medium and High at all resolutions (e.g., 4K: 175 to 109 FPS) suggests that a specific rendering feature, likely related to post-processing or volumetric effects, is disproportionately expensive on this architecture. The RTX 5090's 170 RT cores and 680 tensor cores are largely idle in a 2D side-scroller, but the fixed-function units for rasterization (176 ROPs, 680 TMUs) are heavily utilized. The pixel rate of 423.6 GPixel/s should theoretically handle 4K at high fill rates, but the observed 104 FPS at Ultra indicates that the game's draw calls or driver overhead are the bottleneck, not the GPU's theoretical limits.

The VRAM capacity is a non-issue; 32 GB is far beyond the requirements of a 2016 title that likely uses less than 4 GB even at 4K. The clocks are also not a factor, as the GPU's boost behavior is consistent. The data suggests that the RTX 5090 is overkill for this game, and the measured FPS is limited by software inefficiencies rather than hardware constraints. The 92,200 million transistors on a 750 mm² die are simply not fully utilized in this workload, as evidenced by the massive frame rate gains from simply lowering the preset from High to Medium.

How This Combo Ranks

The Intel Core Ultra 9 185H and RTX 5090 combination ranks 698th out of 3,716 tested combos for Inside, placing it in the top 19% of all configurations. This is a strong result, but not a top-tier ranking, which is surprising given the hardware. The ranking suggests that many other combos, likely with desktop-grade CPUs, achieve higher frame rates in this game. The combo's percentile is not provided, but the rank implies that there are over 3,000 other combos that perform better, which is unusual for such a powerful GPU.

The CPU's nearest rivals in synthetic benchmarks provide context. The Core Ultra 9 185H has an average benchmark score of 35,670, which is nearly identical to the AMD Ryzen AI 7 PRO 350 (35,719, a -0.1% delta) and the Intel Core 7 250H (35,728, a -0.2% delta). This indicates that the CPU is not a performance outlier in general compute tasks. However, in gaming, the CPU's 16 cores and 22 threads are more than sufficient for Inside, which is a single-threaded or lightly-threaded title. The ranking of 698 suggests that the combo is held back by the CPU's memory bandwidth (89.6 GB/s) or PCIe Gen 5 x8 link, but the data does not support this conclusion given the GPU's dominance.

The GPU's nearest rivals in synthetic benchmarks are all data-center or professional parts, with the RTX 5090 having a 0.3% delta over the NVIDIA Tesla P100 PCIe 16 GB. This is irrelevant to gaming performance, but it underscores that the RTX 5090 is a compute monster. In Inside, the combo's rank is likely influenced by the game's age and lack of multi-threading optimization, meaning that the CPU's 3.90 GHz base and 5.10 GHz boost clocks are not the limiting factor. The rank confirms the GPU's ability to brute-force its way to high frame rates, but the engine's single-threaded nature means that faster per-core performance would yield better results.

CPU Role

The Intel Core Ultra 9 185H is a 16-core, 22-thread processor based on the Meteor Lake architecture, built on Intel's 7 nm process. In Inside, the CPU's role is minimal, as the game's puzzle platformer design does not demand high core counts. The data shows that the CPU's single-thread performance, as measured by a Cinebench R23 single-core score of 1810.5, is adequate for the game's logic and physics, but the frame rates are entirely GPU-bound at lower resolutions. At 1080p Low, the 337 FPS result is so high that the CPU is likely processing frames as fast as the GPU can render them, but not bottlenecking.

The CPU's multi-thread capability (Cinebench R23 multi-core score of 17,984.5) is irrelevant for this title, as Inside does not scale across 16 cores. The L3 cache of 24 MB shared is sufficient for the game's data, and the 5.10 GHz boost clock ensures that any single-threaded tasks are handled quickly. The CPU's 45 W TDP and mobile form factor (BGA 2049 socket) are not factors in performance, as the system is likely a laptop or a compact desktop with adequate cooling.

The key observation is that the CPU's performance, relative to its nearest rivals, is almost identical. The Core Ultra 9 185H's average benchmark score is within 0.7% of the AMD Ryzen 7 7700X, a desktop part. This parity suggests that in a game like Inside, the CPU is not a differentiator. The measured FPS differences across combos are more likely due to the GPU or memory configuration. The data indicates that the CPU is never the limiting factor in this combo, as the FPS drops from 4K Medium to High (175 to 109) are far too large to be caused by the CPU. The CPU's 22 threads are idle during most of the game, and its role is limited to feeding the GPU with draw calls. The 89.6 GB/s memory bandwidth is more than enough for a game with simple textures, so the CPU's memory subsystem is not a constraint.

Measured FPS Breakdown

The complete set of measured FPS data reveals consistent patterns across resolutions and presets. At 1920x1080, the average FPS ranges from 337 (Low) to 230 (Ultra), with Medium at 311 and High at 241. The Medium preset has a recorded minimum of 264 FPS and a maximum of 357 FPS, indicating a stable frame time distribution. The High and Ultra presets lack min/max data, but the averages suggest they are less stable due to the rendering feature cliff.

At 2560x1440, the averages are 302 (Low), 278 (Medium), 174 (High), and 165 (Ultra). The Medium preset shows a minimum of 236 FPS and a maximum of 319 FPS, which is a wider range than at 1080p, suggesting that the GPU is working harder and frame times are more variable. The drop from Medium to High at 1440p is 104 FPS, the largest single-preset jump in the entire dataset, highlighting the problematic High preset.

At 3840x2160, the averages are 218 (Low), 175 (Medium), 109 (High), and 104 (Ultra). The Medium preset has a minimum of 148 FPS and a maximum of 201 FPS, which is a 53 FPS spread, indicating significant variability even at this resolution. The High and Ultra results are nearly identical (109 vs 104), suggesting that the Ultra preset adds minimal visual fidelity for a 5 FPS cost. The data shows that the game is playable at 4K on Medium with an average of 175 FPS, but High and Ultra are best avoided due to the disproportionate performance loss. The 1080p Low result of 337 FPS is the highest average recorded, and it is likely limited by the CPU's ability to issue draw calls rather than the GPU's rendering speed. The 4K Ultra result of 104 FPS is the lowest, and while it is still smooth, it represents a 69% reduction from the 1080p Low result.

FAQ

Q: What is the highest average FPS recorded for this combo, and at what settings?

A: The highest average FPS is 337, achieved at 1920x1080 with Low settings. The second-highest is 311 FPS at 1080p Medium, and the third is 302 FPS at 1440p Low.

Q: Why is there a large performance drop between Medium and High settings at all resolutions?

A: The data shows a consistent cliff: at 4K, Medium averages 175 FPS while High drops to 109 FPS; at 1440p, Medium is 278 FPS and High is 174 FPS; at 1080p, Medium is 311 FPS and High is 241 FPS. This suggests a specific rendering feature enabled at High that is disproportionately expensive on this hardware.

Q: Does the RTX 5090's 32 GB VRAM ever become a limiting factor in this game?

A: No. The measured FPS shows no indication of VRAM limitations, as the GPU's 1.79 TB/s bandwidth and 32 GB capacity are far beyond the requirements of a 2016 puzzle platformer. The performance drops are due to other rendering bottlenecks.

Q: How does this combo rank compared to other tested configurations?

A: The combo ranks 698th out of 3,716 tested combos for Inside, placing it in the top 19%. This is a strong result, but not a top-tier ranking, likely because the game does not scale with the CPU's 16 cores.

Q: Is the Intel Core Ultra 9 185H a bottleneck in this game?

A: The data suggests no. The CPU's single-core performance (Cinebench R23 score of 1810.5) is sufficient, and the frame rates are GPU-bound. The CPU's nearest rival, the AMD Ryzen AI 7 PRO 350, has a nearly identical average benchmark score (-0.1% delta), indicating that CPU choice has minimal impact on this title's performance.

Q: What is the best preset for a 4K display with a 144Hz refresh rate?

A: Medium is the only preset that consistently exceeds 144 FPS at 4K, with an average of 175 FPS and a minimum of 148 FPS. Low averages 218 FPS, but High (109 FPS) and Ultra (104 FPS) are below the 144Hz threshold, making Medium the optimal choice for high-refresh 4K displays.