skills.rest
gregoryraymond
gregoryraymondCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

vram-debugging

Diagnose VRAM leaks on AMD RADV GPUs in Bevy projects using vram-bench.

Updated Apr 21, 2026
Search Tags:#performance#debugging#bevy#memory-leaks#vram#gpu-memory#radv
One-click install
npx skills add https://github.com/gregoryraymond/claude-agents-and-skills --skill vram-debugging
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: vram-debugging
Source: https://github.com/gregoryraymond/claude-agents-and-skills/tree/main/skills/vram-debugging
Command: npx skills add https://github.com/gregoryraymond/claude-agents-and-skills --skill vram-debugging

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

VRAM leaks can cause GPU memory to grow unboundedly, leading to crashes and degraded performance. This skill provides a structured guide to diagnose and prevent VRAM leaks in Bevy projects running on AMD RADV GPUs, including practical monitoring and root-cause patterns.

Core Features & Use Cases

  • VRAM monitoring: read and interpret GPU memory counters, detect unbounded growth, and assess leak rates.
  • Root-cause pattern guidance: common patterns that trigger memory churn in Bevy (e.g., per-frame resource mutations) and how to fix them.
  • Performance debugging workflow: step-by-step approach to reproduce, measure, and validate fixes in layers using vram-bench and Bevy resources.

Quick Start

Run the vram-bench tool to profile memory usage and identify leaking layers.

Frequently Asked Questions about vram-debugging

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I debug VRAM leaks in a Bevy game on AMD RADV?▼

Debug VRAM leaks in Bevy on AMD RADV by running the vram-bench tool to profile memory usage and reading GPU memory counters under /sys/class/drm/card*/device to detect unbounded growth.

What causes per-frame memory churn in Bevy?▼

Per-frame memory churn in Bevy is typically caused by patterns involving per-frame resource mutations. Identifying and refactoring these mutation patterns stops unexpected memory growth and stabilizes GPU memory allocation.

How do I monitor GPU memory usage on Linux for RADV drivers?▼

Monitor GPU memory usage on RADV drivers by reading and interpreting the memory counters located under /sys/class/drm/card*/device to assess leak rates and detect unbounded VRAM growth over time.

Can I use vram-bench to measure and validate VRAM leak fixes?▼

Yes, vram-bench profiles memory usage to identify leaking layers in Bevy projects. You can use it throughout the debugging workflow to reproduce leaks, measure memory bounds, and validate applied fixes.

Does this VRAM debugging workflow support non-AMD GPUs?▼

No, this VRAM debugging workflow specifically diagnoses VRAM leaks on AMD RADV GPUs in Bevy-based projects, relying on memory counters under /sys/class/drm/card*/device that are specific to the RADV driver.

Why does GPU memory grow unboundedly until my Bevy project crashes?▼

GPU memory grows unboundedly due to VRAM leaks from patterns like per-frame resource mutations in Bevy. Using vram-bench helps identify the leaking layers causing degraded performance and eventual crashes.

Related Skills

Explore More Software Engineering Skills→
RuanBuitendag42RuanBuitendag42

gaming-linux

Configure Arch Linux gaming with AMD RX 6800 drivers, tools, and Hyprland rules.

Community
Advanced
gregoryraymondgregoryraymond

geometry-debugging

Diagnose terrain and coastline geometry issues in Bevy pipelines.

Community
Advanced
eng-cceng-cc
13

optimization-performance

Profile CPU/GPU metrics and optimize frame times across platforms.

Community
Advanced
tksuorantksuoran
896

erhe-quest-validation

Enable Vulkan validation layers in erhe Quest builds to detect GPU hangs.

Community
Intermediate