- comprehensive technical breakdown of Solari, a real-time raytracing system integrated into Bevy 0.17
- discusses the motivation and development history
- implements ReSTIR Direct and global illumination with a world-space irradiance cache
- uses DLSS Ray Reconstruction for denoising and upscaling
- presents a look at the performance of the various implementation steps
- explains how triple buffering improves frame pacing and throughput by preventing GPU stalls when VSync is enabled
- provides practical implementation examples for Vulkan, DirectX 12, and OpenGL
- briefly mentions the disadvantages of the techniques and common pitfalls
- blog post introduces a library for implementing cubic Bézier curve representation in just 28 bytes
- provides comparisons in time and quality against other implementations
At Astrobotic, we’re building the future of space robotics, from lunar landers to autonomous systems- and simulation is central to that mission. We’re seeking engineers with strong skills in C++, graphics, and simulation to develop high-fidelity environments that enable spacecraft testing and mission operations.
- presents techniques for reducing BLAS builds for cluster-based continuous level of detail systems
- discusses BLAS Sharing, Caching, and Merging
- releases the implementation of the techniques with a detailed discussion of the implementation choices
- additionally presents performance and memory profiling results
- video exploring the foundational principles and methodologies behind Euclidean geometry
- examines how Euclid’s systematic approach to mathematical proofs laid the groundwork for modern mathematical reasoning
- introductory video covering fundamental color theory concepts for beginners
- explains color relationships, harmony principles, and practical applications in design and digital art
- provides accessible guidance on using the color wheel and understanding color interactions for improved visual compositions
- detailed analysis of how increasing vertex shader exports affects rendering performance on NVIDIA GPUs and AMD integrated GPUs
- demonstrates that drawcall costs can significantly increase when exporting more data to the vertex shader stage
- uses Nsight Graphics profiling to show how bottlenecks and memory allocation patterns impact vertex-to-pixel data flow
- introduces robust-kbench, a comprehensive benchmark for evaluating CUDA kernel performance and correctness across varied scenarios
- presents an agentic framework that uses large language models to discover, verify, and optimize CUDA kernels automatically
Thanks to Stephen Hill for support of this series.
Would you like to see your name here too? Become a Patreon of this series.
