这是indexloc提供的服务,不要输入任何密码
Skip to content

copyTextureToBuffer, should we add native API limits or workaround them? #3729

@greggman

Description

@greggman

In Metal, an undocument limit of copying a texture to a buffer is that you pass a bytesPerRow value but that value can not be more than the maxTextureDimension. In other words, if you're copying from an r16 texture and the maxTextureDimension is 16384, then bytesPerRow must be less than or equal to 32768.

I have not checked if other APIs have a similar limit.

Should WebGPU also have that limit or should implementions work around it (for example by copying a row at a time)

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions