Sparse matrix computations are pivotal to advancing high-performance scientific applications, particularly as modern numerical simulations and data analyses demand efficient management of large, ...

Abstract: Real-time movie recommendation systems must efficiently handle large amounts of sparse user-item interaction data while maintaining great prediction accuracy. Conventional collaborative ...

Multiplying the content of two x-y matrices together for screen rendering and AI processing. Matrix multiplication provides a series of fast multiply and add operations in parallel, and it is built ...

High-performance matrix multiplication remains a cornerstone of numerical computing, underpinning a wide array of applications from scientific simulations to machine learning. Researchers continually ...

Abstract: Large language models (LLMs) rely on self–attention for contextual understanding, demanding high-throughput inference and large–scale token parallelism (LTPP). Existing dynamic sparsity ...

CUDA-L2 is a system that combines large language models (LLMs) and reinforcement learning (RL) to automatically optimize Half-precision General Matrix Multiply (HGEMM) CUDA kernels. CUDA-L2 ...