gfx-rs/wgpu

StagingBelt memory usage problems in certain scenarios

Open

#1,441 建立於 2021年5月8日

在 GitHub 查看
 (0 留言) (3 反應) (0 負責人)Rust (9,290 star) (703 fork)batch import
good first issuetype: bug

描述

Recently I decided to try out Iced but run into some memory problems while testing an example, as can be seen in this issue: https://github.com/hecrj/iced/issues/786 I narrowed it down to the wgpu-rs StagingBelt, and the lack of a way to cleanup unused free Chunks from the belt. The issue with the Game of Life example was that I only added more geometry to the screen, so my buffers got larger, never fitting inside the previous Chunks and so creating new Chunks while the old Chunks stayed unused forever. I managed to fix the issue by implementing a naive GC: https://github.com/eduardobraun/wgpu-rs/commit/a62dd38538b41e2890a0cecd59e692f1d208d6e6

TLDR: StagingBelt needs a way to reclaim memory from free Chunks.

貢獻者指南

技術棧
rust
領域
performance
議題類型
performance
難度面向新貢獻者的預計實作難度，1 表示很小改動，5 表示專家級工作。
4
預計時間有經驗貢獻者完成調查、實作、測試並準備 pull request 的粗略時間範圍。
1-2 days
活動狀態議題目前的可參與程度：新鮮、活躍、陳舊、阻塞或等待維護者輸入。
stale
清晰度議題是否清楚說明預期改動、驗收標準和下一步。
clear
前置要求
Rust programmingwgpu rs basicsmemory management concepts
新手友善度1-100 的估計分數，表示該議題對首次貢獻者的友善程度。
30
研究方向
Examine the linked commit a62dd38538b41e2890a0cecd59e692f1d208d6e6 which implements a naive GC for StagingBelt. Understand the current memory allocation strategy in the StagingBelt module of wgpu rs. Identify where free chunks accumulate and design a cleanup mechanism to reclaim memory. Consider integrating a periodic or explicit cleanup method that can be called to release unused chunks.