WebGPU Resource Binding: Static Bind Groups vs. Dynamic Offsets
Learn how to optimize WebGPU performance by choosing between static Bind Groups and Dynamic Offsets for uniform data, including alignment requirements and implementation steps.
Loading…
ReadMeFeed / Community knowledge
#gpu-optimization
Real questions. Useful conversations. Find the people who know your stack.
Eliminating GPU Starvation with tf.data Pipelines
Stop wasting GPU cycles. Learn how to use tf.data to overlap CPU preprocessing with GPU execution using prefetching, parallel mapping, and autotuning.
tensorflowReducing Global Memory Bottlenecks with OpenCL Local Memory Tiling
Learn how to use OpenCL Local Memory (LDS) and tiling patterns to reduce global memory latency and improve kernel performance in compute-heavy applications.
openclOptimizing 2D Rendering in MonoGame with SpriteBatch
Learn how to optimize 2D rendering in MonoGame by mastering SpriteBatch, managing SpriteSortMode for performance, and using SamplerState.PointClamp for crisp pixel art.
monogameOptimizing Custom Mesh Data with Three.js BufferGeometry
Learn how to use Three.js BufferGeometry and InterleavedBuffers to optimize custom mesh rendering and reduce CPU-to-GPU data transfer overhead.
threejsGPU Resource Disposal Limits in Complex Scene Graphs
Manual Memory Management in Three.js Three.js requires explicit resource management for assets resident on the GPU. While JavaScript handles CPU-side memory via garbage collection, GPU-resident data such as BufferGeometry , Material , and Texture objects must be manually released using the .dispose() method. Resource Orphanage Constraints Removing a mesh fro
BufferGeometry dynamic attribute update strategy in Three.js r152
Dynamic Vertex Attribute Updates Three.js BufferGeometry stores vertex data in TypedArrays that are uploaded to the GPU as BufferAttributes. When attributes change every render frame, developers must choose a WebGL usage pattern and decide whether to replace the whole buffer or update only modified ranges. The needsUpdate flag triggers a full buffer upload,
NaN Gradients during Mixed Precision Training in Keras
Resource Optimization via Mixed Precision To reduce VRAM overhead and lower compute costs for low-traffic workloads, the Keras Mixed Precision API is used to execute compute-heavy operations in float16. This approach aims to minimize the memory footprint of activations, potentially allowing the use of smaller GPU instances. Numerical Stability Constraints Wh
kerasSpriteBatch and GraphicsDevice: Impact of SpriteSortMode on GPU State Changes
MonoGame uses the SpriteBatch class to interface with the GraphicsDevice , employing a deferred rendering system to minimize the overhead of GPU state changes. The efficiency of this pipeline depends heavily on the SpriteSortMode selected during the Begin() call. When utilizing SpriteSortMode.Deferred , the framework batches draw calls until End() is invoked