I'm Dhaval Panchal. I build renderers, physics engines, and GPU-accelerated graphics tools in C++. This is the engineering log.
Custom math libraries, hand-written shaders, measured benchmarks. No wrappers, no black boxes. All code is open-source and the writeups cover real architectural decisions.
Engineering retrospectives in the blog. Project breakdowns on the projects page.
Building an N-body physics engine (Stardust) with data-oriented design, custom memory allocators, and cache-aware data layouts. Production-grade C++: systems architecture, RAII ownership, and performance-measured design decisions.
Developing Gravity Rift, a custom OpenGL 4.6 deferred renderer. G-buffer pipeline, Cook-Torrance PBR, PCF shadow maps, SSAO, HDR with ACES tonemapping, instanced rendering, and compute-shader particle systems.
Writing a CPU path tracer with SAH BVH, then porting it to CUDA. Kernel dispatch for parallel ray casting, coalesced memory access, occupancy tuning, and full FLOP benchmarks. End goal: a measured GPU vs CPU speedup comparison on real scene data.