PC performance and settings

Day-one update and Nvidia driver warning

S-GAME’s October 28 letter announces an approximately 5 GB day-one update, with ray tracing and path tracing performance improvements and bug fixes. Preloaded copies need the update before play; a fresh download includes it according to the developer. The letter does not state an executable version or a PS5 download size. Kingy has not installed or benchmarked the update.

S-GAME’s October 28 letter supersedes its earlier blanket advice to install the latest graphics driver: it reports instability with Nvidia 621.01 and recommends 617.42, using a clean installation, until a subsequent Nvidia update. This is dated developer advice, not a Kingy-tested fix or a recommendation to downgrade every GPU. Check current S-GAME and Nvidia release notes for your hardware before changing drivers; this review has not validated a later fix.

Sources: Reason and Madness — day-one update and launch options ↗ (checked November 2, 2026)

November 4 patch: shader compilation and controller fixes

S-GAME’s version 1.0302256 notes (November 4) say the PC shader compilation process was updated to cut startup times, and ray tracing shaders are no longer compiled at startup by default. Players who had Ray Tracing disabled need a one-time shader recompile on the first launch after the patch; players who had it enabled can play without recompiling. If you turn Ray Tracing on, its shaders compile separately at the next restart. The notes also list less stutter with a Bluetooth controller, NVIDIA Reflex working on Auto with V-Sync on, a fix for hair rendering on some AMD cards with certain drivers, and a warning that some third-party mods are incompatible. Kingy has not installed the patch or measured startup time.

Sources: Patch Notes, Version 1.0302256 ↗ (checked November 4, 2026)

What has been tested

Kingy has not run Phantom Blade Zero. The requirements guide records publisher hardware targets, while this page offers a not yet tested measurement method. A third-party benchmark should be assessed by its build, hardware and method.

Establishing your baseline

Use the same conditions for each measurement:

• Write your hardware, driver, OS and resolution before the first launch.
• If a built-in benchmark exists, run it three times in the same session and record the median result.
• Record one representative scene (first city street, one crowd fight) and note the frame counter range across 60 seconds.
• Change exactly one setting at a time for A/B comparisons — never a cluster.

What to distrust day one

Before relying on a posted GPU result or settings preset, check for the game version, hardware, method and benchmark scene. Without those details, it may not predict performance on your setup.