Gpu Optimization News
The latest Gpu Optimization coverage curated for game developers.
TileGS is a new tile-local depth binning approach for 3D Gaussian Splatting that cuts raster traversal work without changing output quality. On a 9-scene benchmark, the No-GW variant sped up …
HiGS splits 3D Gaussian Splatting into coarse macro-tiles for partitioning and fine render tiles for rasterization, which tackles the usual bottleneck where a few dense tiles stall the frame. In …
Tensor Cores may finally make 3D Gaussian Splatting practical for latency-sensitive rendering. This paper reports a 1.65× end-to-end speedup by moving the rasterization bottleneck onto Tensor-Core-friendly matrix ops and running …
BlitzGS pushes city-scale Gaussian splatting toward practical training times by cutting the active workload at three levels: GPU sharding, densification control, and per-view culling. The result is near order-of-magnitude speedups …
The exploration of performance-energy trade-offs in 3D Gaussian Splatting is crucial for developers working with constrained GPU budgets. This research reveals how under-clocking and power caps can simulate various GPU …
Recent advancements in spatiotemporal variance-guided filtering (SVGF) have led to a significant performance boost for real-time path tracing. Developers can achieve a 1.3–2.5x speedup without compromising filtering quality, thanks to …