Graphics Programming 419


Focus on Your Algorithm—NVIDIA CUDA Tile Handles the Hardware

  • introduces CUDA Tile, a tile-based programming model launching with CUDA 13.1
  • provides developers with cuTile Python for accessible GPU programming and CUDA Tile IR for building custom DSL or libraries


VK_EXT_present_timing: the Journey to State-of-the-Art Frame Pacing in Vulkan

  • introduces VK_EXT_present_timing extension to enable precise frame pacing control in Vulkan
  • provides timing feedback from previous frames and allows specifying target presentation times for future frames


Minimalist ray-tracing leveraging only acceleration structures

  • presents a low-quality ray tracing mode using ray queries and data stored in acceleration structures
  • eliminates ray pipelines, shader binding tables, and texture access by storing average diffuse color in TLAS instanceCustomIndex
  • uses VK_KHR_ray_tracing_position_fetch to compute face normals from triangle vertex positions


GSWT: Gaussian Splatting Wang Tiles

  • proposes a tile-based framework for synthesizing large-scale terrains from 3D Gaussian Splatting exemplars
  • builds on Wang Tiles with boundary constraints to enable seamless tiling of Gaussian fields


AMD GPU Debugger

  • presents a walkthrough of implementing a GPU debugger for AMD hardware
  • explains how to use registers and trap handlers to pause GPU execution and inspect register state
  • demonstrates stepping through shaders, examining variables, and integrating with SPIR-V compilation through RADV’s ACO compiler


libGPUCounters Machine Readable Specification

  • provides a machine-readable specification for Arm Mali GPU performance counters starting from Mali-G71
  • offers Python wrappers that resolve symbolic counter references for any supported GPU
  • enables structured iteration through counters based on various orders


A series of tricks and techniques I learned doing tiny GLSL demos

  • shares techniques learned from creating four tiny GLSL demos constrained to 512 characters
  • covers isometric projection setup, smooth minimum operations, and various domain warping techniques for procedural generation


[video] New GPU Profiler and RHI Submission Pipeline | Unreal Fest Stockholm 2025

  • presentation covering improvements to Unreal Engine’s GPU profiler
  • discusses the redesigned RHI submission pipeline architecture and implementation
  • shows how to leverage these tools for performance analysis, optimization and crash debugging


[video] Better Tile Breakup - Terrain Shaders - Episode 7

  • video tutorial on techniques to reduce tiling artifacts in terrain textures
  • demonstrates methods to break up repetitive patterns in large-scale terrain rendering
  • part of an ongoing series on terrain shader development


Thanks to Leonardo Etcheverry for support of this series.


Would you like to see your name here too? Become a Patreon of this series.