Skip to content

Aarch64/ARM64-native compiled executable outputs broken tokens only #2

@FlyingFathead

Description

@FlyingFathead

Hi there!

I was wondering if there's a Little Endian/Big Endian byte ordering type issue when compiling vzgpt natively on a ARM64 (i.e. Raspberry Pi) type device?

If compiled on a native aarch64/arm64 system, the program only outputs broken tokens when running as a native-compiled arm64/aarch64 binary (= the program runs "fine", but all of vzgpt's output is garbled into broken Unicode characters).

This issue only appears on aarch64 compiles and when running on that type of architecture.

I have used this as a workaround: if you compile vzgpt to a x86_64/amd64 executabe and run it over Box64 (https://github.com/ptitSeb/box64) on an ARM64/Aarch64 system, the output is pretty much okay, albeit slower than it would run native due to the bottleneck caused by the userspace emulation in between.

Given that the garbled output issue is only present in the aarch64/arm64 native compile, this would lead me to believe it's a little/big endian byte order issue.

Is there a proposed solution to get vzgpt run natively as an aarch64/arm64 executable?

Are you planning on releasing a proper configfile soon that would perhaps take this issue into account?

Thanks and all the best with your project.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions