lijinlei@kvark if need to update an existing buffer every frame, map_write or copy_buffer_to_buffer which is the best choice? as gpuweb docs mentioned, user can reusing upload buffers to reduces overhead, is it possible or appropriate in wgpu-rs?
lijinlei: we moved to the matrix, btw
look up gfx or wgpu rooms and let's continue there!
to answer your question, re-using upload buffers is pretty much blocked on #9, so creating a new upload buffer and copying from it is the way to go, for now
Great! Beware though, latest wgpu requires that any resources you use in a pass are still alive at the end of a pass recording. In Rust this is enforced by the type system for free.
Also, we have moved this room to #wgpu:matrix.org
almarklein Hi! We've started building higher-level things (scene graph, render engine) on top of wgpu-py, and have some technical questions about implications of certain wgpu features. Where would be the best place to ask?
Hi! This room is deprecated. We moved to #wgpu:matrix.org
Hi, has anyone already try to run webgpu on IOS devices?
@baonghi1201: hey! this room is deprecated, see the topic
Hello. Are there any barriers in webgpu ? I wrote to storage texture using atomic operations. And sometimes they are not visible. Thanks.
So I've finished the tutorial at https://sotrh.github.io/learn-wgpu, but now I want to render simple dots in 3d space whose movement follows some rules. What keywords do I need to search to find some good examples of what that entails?
Hello, I am just getting started with WGPU and am curious what the best practices are for managing many webGPU resources. For example, all the tutorials I have found will make one shader that does one thing, but what If I need many shaders that operate independent of one another? Which resources are completely unique to receiving a shader, and which resources can be shared? For Example: to the best of my knowledge (correct me if i'm wrong at any point) the Instance, Adapter, Device and Queue, and Command Encoder are shared between WGPU resources. While the Buffers, BindGroupLayouts, BindGroups, ShaderModules and Pipelines are unique to processing a specific shader. If this is correct, would it be wise to make a wrapper for those unique components and another wrapper for a collection of the unique values as well?