Why AnyPS5’s counter fell from 91% to 42% while implemented functions grew
ItsOOGWAY · PC guides & recorded tests
Why AnyPS5’s counter fell from 91% to 42% while implemented functions grew
AnyPS5's system-library counter dropped from 91% to around 42%, while the counted implemented functions increased. This update explains the bigger tracked workload, recent merged code changes, and what the game reports actually demonstrate.
The key checkpoint: The denominator expanded much faster than implemented coverage. A system-library percentage is not a game-compatibility score or a universal FPS forecast.
Before you start
- This is the 22 October 2026 reporting snapshot, not today’s live percentage.
- Keep packaged release, current main and contributor branches distinct.
- Judge game results by the exact build, scene and hardware.
Resources from this video
Links, downloads & documentation
Every original URL from the video description is collected here, including recorded releases, setup discussions and related videos. Use the version-specific notes when following a recorded build.
Watch the complete video
Open this video on YouTube ↗. Screenshots below link to their exact recorded moments.
Walkthrough 01
Define the reporting snapshot
This update is based on the project pages and contributor reports checked on October twenty-second. Let's separate three things: the size of the tracked workload, changes that have reached the main project, and results from individual game tests. They answer different questions. The latest packaged release in this snapshot is version zero point one point three, published on October twentieth. Some changes we're discussing landed afterward. Updating a release package, building current main, and using a contributor's test branch do not necessarily give you the same code.
Walkthrough 02
Read the numerator and denominator together
First, the drop. The history shown here goes from three thousand, six hundred and eighty-three implemented functions out of four thousand and twenty-nine declared, to four thousand, six hundred and seventy out of eleven thousand and thirty-two. That's roughly ninety-one percent becoming forty-two percent. The implemented side grew by nine hundred and eighty-seven. The declared total grew by seven thousand and three. Think of it as a checklist that became much longer. Completing more items doesn't keep the percentage high if the list expands even faster.
These are functions across libraries, not seven thousand new libraries. And the historical chart recalculates older code with the current counting rules. It is useful for comparison, but it isn't a time machine showing exactly what every old badge displayed.
Walkthrough 03
Separate export declarations from working behavior
Here is the concrete example in the recording. This merged change adds declarations for system-library exports. The important word is declarations. Some entries are placeholders that explicitly stop with an unimplemented error if called. Resolving a function's name and reproducing its behavior are separate jobs. Making that missing behavior visible helps map the work; it doesn't magically make a game compatible. But there is another update worth watching, because adding names is only one part of the job.
Walkthrough 04
Interpret behavioral and shader changes carefully
A separate merged change reports corrections to the signatures or behavior of seventeen hundred and two existing exported functions, plus eighty-five typed wrappers. A signature describes how a function receives arguments and returns a result. Having a name available is not enough if the caller and the implementation disagree about that contract. This is why a growing implemented count still needs behavioral testing. The change's description also says it wasn't tested after its rebase, so a large patch count should not be presented as a large set of proven gameplay fixes.
The graphics work is moving too. One merged shader change addresses a particular image-table pattern that stopped Guardians of the Galaxy. Another targets unnecessary CPU transfers when refreshing certain texture arrays. Those are different kinds of progress: handling a shader correctly, and avoiding work that shouldn't have to happen. Neither gives us a universal FPS multiplier. To find out what a player gets, we need the corresponding game test. And this is where the update gets more interesting than the percentage.
Walkthrough 05
Compare actual game-test reports
The recorded Silksong discussion contains a report of playable sixty-FPS gameplay on a named Windows build. Astro Bot reports show rendered levels, but performance and even successful startup still depend heavily on the build, scene and machine. Ghost of Yotei has reached the open world on a contributor test branch, while its reported world performance remains around half a frame per second. Those are very different stages. One green percentage cannot describe all three. That also explains why the shader counter showing one hundred percent does not mean the graphics system is finished.
Instruction coverage is one measurement. Correct images, memory handling, synchronization and acceptable frame times still have to come together in real games.
Walkthrough 06
Track evidence that can be reproduced
So what should you actually follow from here? Start with the exact game and title ID. Check the date, source build, operating system and hardware. Then ask what happened: did it only launch, reach a menu, enter a level, or sustain playable gameplay? Finally, check whether the result came from main or a contributor's branch with extra changes. A screenshot and a patch list become much more useful when that context stays attached.
If your result differs
| Situation | Next useful check |
|---|---|
| Reading 42% as lost code | The recorded implemented count grew from 3,683 to 4,670 while declarations grew from 4,029 to 11,032. |
| A 100% shader counter looks final | Instruction coverage does not establish correct rendering, synchronization or acceptable frame times. |
| A branch result is attributed to a release | Keep source branch and packaged version attached to every result. |
Comments
Post a Comment