Onimusha: Way of the Sword Benchmark

Where to get the benchmark

The official Capcom benchmark page is the primary destination for this tool. The Steam store page also surfaces a benchmark entry alongside the full game and the free demo, and lists the official performance targets and upscaling notes.

Because it is separate from the game, you can measure your PC before buying — which is the entire point of Capcom shipping one.

How to run it so the result means something

Match the benchmark settings to how you will actually play: same resolution, same preset, same upscaling mode, same ray tracing state. A run at 1080p Low tells you nothing about 1440p with ray tracing enabled.

Run it more than once. The first pass on a fresh install often includes shader compilation, so the second and third runs are more representative of steady-state performance.

What numbers to write down

Average frame rate alone hides the problems that ruin a parry-based action game. Record 1% lows as well, because a dip during a deflect window is what costs you a fight.

Note VRAM usage and any stutter around scene transitions. PC Gamer reported occasional brief hitching around loading and cutscene transitions even on a high-end configuration, so a small amount is not necessarily a fault in your setup.

Reference results from reviewers

PC Gamer measured frequent 120 FPS on an RX 9070 XT with an i5-13600K at maximum settings with FSR 4 Quality. TechRadar likewise described stable performance on high-end PC hardware and confirmed DLSS and FSR frame generation, ray tracing and ultrawide support.

TechRadar also reported roughly 40 FPS on a handheld, achieved by lowering power and image settings. That is the practical floor to keep in mind if you play on a portable PC.

Third-party GPU benchmarks

A widely viewed RTX 4060 Ti 16 GB run with an i5-13400 covers 1440p Ultra with ray tracing High, Low and Off, plus DLAA and DLSS Quality, Balanced and Performance with frame generation. An RTX 5080 run covers 4K and 1440p with DLAA, DLSS 4.5 Quality, multi-frame generation and ray tracing.

Both were captured on the demo build with specific CPUs, memory and driver versions. Use them to judge relative scaling between settings, not to predict your exact frame rate.

Turning a benchmark result into settings

If you land below your target, the cheapest gains usually come from lowering ray tracing first and upscaling quality second, before touching texture or shadow detail. The official targets on the Steam page assume upscaling is already enabled.

If you comfortably exceed your target, raise settings one step at a time and re-run rather than jumping to maximum. Re-testing after driver updates and launch patches is worth the few minutes it costs.

Benchmark, demo, or both?

The two tools answer different questions. The benchmark measures your hardware against a fixed scene, while the demo shows how the game actually plays during combat, including the moments when effects and enemies pile up.

Running both is the thorough approach: the benchmark for a repeatable number, the demo for whether that number feels good in a fight. Neither costs anything.

If you only have time for one and you are unsure about your GPU, run the benchmark first. There is no point enjoying a demo you cannot run smoothly at the settings you want.

Reading your result against the official targets

The published targets are roughly 30 FPS at 1080p Low with upscaling on the minimum specification, and about 60 FPS at 1080p Medium with upscaling on the recommended one. Both assume upscaling is on, so compare like with like.

If your benchmark result comfortably exceeds the recommended target at 1080p, moving to 1440p is a reasonable next test. If it barely meets it, stay at 1080p and spend the headroom on stability instead.

Re-run after driver updates and after launch patches. Performance on a pre-release build is not a permanent property of your PC.

Settings that move the needle

Ray tracing is typically the most expensive single option, and the RTX 4060 Ti test covers High, Low and Off specifically because that spread matters. Turning it down is usually the first lever to pull.

Upscaling quality is the second lever. DLSS and FSR modes trade image sharpness for frame rate in predictable steps, and frame generation raises the displayed rate without improving input latency.

CPU performance presets exist as well, which matters if your bottleneck is the processor rather than the GPU. A benchmark run with GPU usage below full load is a hint that this is your situation.

What a benchmark cannot tell you

A fixed benchmark scene is repeatable, which is its value, but it does not reproduce every load the game will put on your hardware. Dense combat with several enemies and effects can behave differently from a scripted camera path.

It also cannot predict how launch patches or driver updates will change performance. Treat any result as a snapshot of that build on that day.

For a feel-based check, run the free demo afterwards. It contains real fights, which is exactly what the benchmark leaves out.

Where the benchmark fits in a buying decision

The sequence that wastes the least money is straightforward: run the benchmark, then the demo, then read a review of the PC version. Each step answers a different question and all three are free.

If the benchmark falls short at the settings you want, you still have options — lower ray tracing, adjust upscaling, or wait for hardware. None of those decisions require owning the game first.