Graphics Programming Weekly - Issue 385


Optimizing copy of null descriptors in D3D12

  • The blog post analyzes the CPU overhead caused by copying D3D12 descriptors
  • It demonstrates how to use available APIs to reduce this overhead significantly
  • Discusses how different hardware tiers handle uninitialized descriptors differently
  • Explores how various shader compiler options affect performance outcomes


The DDA Algorithm, explained interactively

  • The blog post provides an interactive explanation of the Digital Differential Analyzer (DDA) algorithm
  • Through step-by-step visualizations, it breaks down how the algorithm traverses a grid
  • Focuses on the 2D case to build intuition before discussing more complex applications


[video] Forward Plus Additional Lights - Custom Lighting Models - Episode 11

  • This video tutorial demonstrates how to extend a forward lighting shader to support multiple dynamic lights
  • Implementation is shown using Unity’s rendering pipeline
  • Explains the performance considerations and optimizations for handling many light sources


Modern SSAO in a modern run-time

  • The blog post details the integration of Screen Space Ambient Occlusion into Use.GPU, a reactive GPU authoring system for the web
  • Showcases the debug pass inspector integration for visual analysis
  • Explains the technical considerations and implementation choices behind their SSAO approach


The Mean-ing of Loss Functions

  • The blog post explores the mathematical and statistical foundations behind common loss functions
  • Starting with squared error loss, it explains the connection between loss functions and probability distributions
  • Provides practical guidance on selecting appropriate loss functions based on the underlying assumptions of your data


RDNA 4's Out-of-Order Memory Accesses

  • The blog post presents experimental results comparing RDNA3 and RDNA4 memory access models
  • Demonstrates how RDNA4 resolves false cross-wave dependencies that exist in RDNA3
  • Includes comparative benchmarks against NVIDIA and Intel hardware to highlight architectural differences


Thanks to Warren Moore for supporting this series


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