Overpower

The pairing tested here combines a 2018 mid-range six-core processor with NVIDIA's 2025 flagship graphics card. Overpower, an action game released in August 2023, runs at frame rates above 100 FPS in every measured configuration, yet the data reveals a CPU-limited combination rather than a GPU-bound one. The Ryzen 5 2600's six cores and 12 threads, relying on Zen architecture with a 65 W TDP, deliver a PassMark multi-thread score of 13,160, while the RTX 5090 brings a 32 GB GDDR7 buffer and a 1.79 TB/s memory bandwidth. Overpower's 1080p-to-4K scaling is unusually flat, pointing straight at the processor as the bottleneck.

CPU Role

The AMD Ryzen 5 2600 is a desktop processor from the Ryzen 5 2000 series, built on the Zen+ (Pinnacle Ridge) architecture at GlobalFoundries' 12 nm node. It houses 6 cores and 12 threads with a base clock of 3.40 GHz and a boost clock of 3.90 GHz. Its power envelope is 65 W TDP and it supports DDR4 memory via a dual-channel interface with a peak bandwidth of 46.9 GB/s. The CPU includes a 96 KB L1 cache per core, 512 KB of L2 per core, and 16 MB of shared L3 cache. It launched on 2025-04-18 with a launch MSRP of $199. In benchmark tests it returned a Cinebench R15 multi-core score of 1,248 and single-core score of 157.3, a Geekbench 5 multicore of 5,648 and single-core of 1,154, and a PassMark average benchmark score of 21,484. Its percentileVsAllCpus is 75, meaning it outperforms 75% of all tested CPUs, with a PassMark multi-thread result of 13,160 and a single-thread score of 2,239.

Overpower is a fast-paced action game that benefits from high CPU performance and low thread latencies. But the measured FPS scaling in Overpower is discouraging for a CPU with these specifications. At 1280x720 the overall average framerate across settings is small; the frame rate ranges from 148 FPS at Low to 113 FPS at Ultra. At 4K, the same settings produce 135 and 103 FPS, respectively. That is a tiny drop despite a boost from 1080p to 3840×2160, and that flatness is the signature of a CPU that reaches its throughput ceiling before the GPU takes over. The 6-core, 12-thread design with a 3.90 GHz boost ceiling cannot feed the RTX 4090 floral at high particle and physics moments, especially with 4K frames.

The CPU sits against a quartet of nearest rivals: the Intel Core Ultra 5 228V with an average score 21,440 (0.2% lower than the Ryzen 5 2600's benchmark average of 21,484), the Intel Core i9-11900H at 21,367 (+0.5%), Intel Xeon D-1746TER at 21,635 (-0.7%), and AMD EPYC 9454 at 21,223 (+1.2%). So, within this grouping, the Ryzen 5 2600 is essentially average in general compute, but its single-thread capability historically lags all those chips. Yet those empirical competition metrics are peripheral: the 6-core count and low clock speed of this chip is still adequate for Overpower in a single-thread- resolved, low-resolution settings. Overpower's own demands seen in the FPS data indicate a higher neutral CPU load that the 2nd-gen Ryzen 5 is prone to saturating.

How This Combo Ranks

In the full Overpower database, this specific DLLS: "Overpower comboRankInGame lists any combs that are hosted in 2023." Here, a combo consists of a Ryzen 5 2600 paired with the RTX 5090. The measured FPS data ranks it at position 1,024 out of 3,713 total combinations that have been tested—meaning 1,023 combos deliver higher average frame rates. Considering that both components are solid, this is not a neck-deep ranking. The CPU at the 75th percentile and the GPU at the 92th percentile globally, but the combined result lands roughly in the 27th percentile of this game’s tested combos. It clearly isn't the best mix of this GPU, and the CPU is the pull.

Note that the CPU's nearest rivals are all close in score, so the RM-2600's raw compute within 9%-1.2% of its peers. Still, in Overpass a 6.4% average deltas gives the user 148: you do not get the FPS that the GPU would be able to push on your machine. And the nearestRivals of the GPU are some old GPUs: the NVIDIA Tesla P100 with avgScore 79,605 (0.3% difference) and the RTX 5090 `avgBenchmarkScore` 79,842. So the 5090 outscores over those by basically 0.3%. The 50-class silicon's raw benchmarks are amongst the highest in the database—its percentile 75% all GPUs—but the combo rank is not as could be. The bottleneck is the CPU.

Resolution Scaling

Look at the measured FPS across resolutions. All tested frames are above 100 in every setting. If the GPU where the primary constraint, you would see a large relative performance drop as resolution raises. For the RTX 5090 with 32 GB of GDDR7 memory and 512-bit bus, the rate for each preset is reduced.

Actual measured numbers for the Full HD (or 1080p 1920×1080) vs 2560×1440 vs 4K (3840×2160) are:

• Low: 148 → 142 → 135
• Medium: 123 → 116 → 112
• High: 120 → 112 → 110
• Ultra: 113 → 106 → 103

The pattern across these 12 measures indicates Typical scaling: from 1080p to 4K you lose 18%, exactly 7 frames per preset. The average is about 10% less from 1080p to 4K, which is very small for a factor that fills +3× the pixels. The resolution isn't relevant at 1080p; the FPS only goes from 148 to 113 on the same GPU because the CPU is forcing a stutter.

At the high end, the RTX 5090 has enough raw performance to push 135 FPS at 4K with Low. At 1080p Low anne: 148 FPS—so the gain from reducing resolution from 4K to 1080p is only 13 FPS. This type of small uplift is an obvious consequence of CPU capped frame generation. The difference in memory bandwidth of the card (1.79 TB/s) and its clock 2017/2407 MHz doesn't impact that. The game’s physics or simulation logic saturates the Ryzen 5’s pipelines.

The Medium setting yields the only available min\-max data for all resolutions: at 1080p, the minimum is 105 and the maximum is 142 (that is a ~135% jump). At 1440p the min is 101, max 136; 4K min 96, max 129. The very wide min/max spacing means that while the average stays 123+ whatever, there are frame dips that approximate to a 17–23% loss below frame averages. In Overpower, single‐direction streams have sufficient CPU strain to see stable periods.

Measured FPS Breakdown

The data set provides exact averages for 12 combinations of resolution (1. Each resolution and set used a bitpack). Below is the full observed table.

1920×1080 (FHD)

• Low: 148 avg
• Medium: 123 avg (min 105, max 142)
• High: 120 avg
• Ultra: 113 avg

2560×1440 (QHD)

• Low: 125
• Medium: 116 (min 101, max 136)
• High: 112
• Ultra: 108

3840×2160 (UHD)

• Low: 135
• Medium: 109 (min 96, max 129)
• High: 110
• Ultra: 103

These numbers are the product, and they are as follows. At 1080p, every preset results in at least 113 FPS, plenty for any monitor. At 1440p, the low preset averages 135. The 4K results vary between 103 and 135. A full inspection of the four settings at each resolution reveals that the variance across settings at 1080p is 34 FPS (from Low to Medium—low 158, medium 123), which is a 26% reduction, minor compared to unlimited. At 4K this spread is slightly smaller: 32 FPS between low and ultra, again with the CPU dominating.

Notice that the 1440p row has a 101 FPS minimum on Medium, and 4K Medium has a 96 FPS minimum, still safe for FPS games. Since the only min/max reported for medium, we use that a smooth experience with this combo requires avoiding "low" since even the minimum is still not a-th.

GPU Role

The NVIDIA GeForce RTX 5090 is Blackwell 2.0 architecture on a GB202 die, using TSMC 5 nm technology, with 92,200 million transistors on a 750 mm² die. It has 32 GB of GDDR7 memory on a 512-bit interface, bandwidth 1.79 TB/s, and a base clock of 2017 MHz, a boost of 2407 MHz. It carries 21,760 shading units, 680 TMUs, 176 ROPs, 170 ray tracing cores and 680 Tensor cores. The compute power is rated 104.8 TFLOPS FP32, pixel fill rate 423.6 GPixel/s, texture rate 1,636.8 GTexel/s. Its TDP is 575 W, coming with a 16-pin power connector and suggested 950 W PSU. It launches in 2024 with a price of $1,999 USD (the only launch MSRP we mention). Among all GPUs, it is in the 92nd percentile with a 3DMark Steel Nomad score of 18,355, Geekbench OpenCL 334,370, Vulkan 376,728, and PassMark G3D of 39,650 (average over all benchmark results: 79,842). Its nearest rivals are surprisingly different: the NVIDIA Tesla P100 16 GB (average score 79,605, delta +0.3%), the Tesla P100 12 GB (79,396, +0.6%), AMD Radeon RX 6850M XT (78,940, +1.1%), AMD Radeon Pro Vega 64X (80,959, -1.4%). No matter that, it is an overwhelming GPU.

But within this specific game, the GPU's high bandwidth minimizes the FPS gap as you reduce resolution. The fastest (Low) has an FPS ceiling of 3840×1080* = 103 to 148. The RTX 5090 doesn't ramp up as the resolution increases, meaning the GPU is never fully sweat; for example, at 1080p Ultra it would 113, and at 1080p Low 148, but even 4K Low 135 remains. The scaling along the memory bandwidth/Compute load doesn't have effect because the simulation is lightness at laz. Real yield, the combination of the GPU gives, falls toward CPU-side.

Settings Recommendations

Which preset gives the best experience for this balanced setup? The measured rows provide two important clues. First, all presets are playable at all resolutions, with a minimum average of 103 FPS (Ultra / 4K). Second, the only min/max data are for the Medium preset, the only setting in each resolution where the system logs the frame spikes (low 96 FPS at 4K, max 129; at 1080p min 105, max 142). That means Medium gives precise info about frame pacing.

If the top priority is frame stability and reading experiences, the Medium preset is the safest choice. It produces an average 144. Based on the measured min values, 1080p Medium holds 105 FPS as the minimum and 4K Medium 96 FPS bottom, which is still an action-game decent experience (though near 96). On the other side, High and Ultra have higher average numbers (120 at 1080p High, 116 at 1440p, 112 at 4K High), but no min details, so we cannot guarantee these averages represent the consistent experience. If quality® is primary, Ultra at 4K averages 103 FPS—above the 60Hz threshold—and at 1080p averages 116 FPS, both passable. Delving into numbers, the /settings recommendation from this data is obvious: use Medium on this particular CPU / GPU combo under any condition, because f that resolution only set yields actual minimum and maximum values, and averages still reach 119-123 FPS on 1080p and 112 at 4K. High affords better quality with ~2-4 frame losses on average, but we cannot assess stability. If you must choose quality, go for High at 1080p / 1440p, the framerate differential of 110-120 is unlikely to hurt; at 4K, choose Medium since the performance gap for Ultra High is ≤ 7 FPS at that average. Do not push the CPU over: the difference between High and Low in FPS is 18-13%, so no amount of settings engineering removes the processor bound. A 65 W six-core is the limiting element, not rasterier.