tritonBLAS: Triton-based Analytical Approach for GEMM Kernel Parameter Selection
By: Ryan Swann , Muhammad Osama , Xiaohu Guo and more
Potential Business Impact:
Makes computer graphics run much faster.
We present tritonBLAS, a fast and deterministic analytical model that uses architectural parameters like the cache hierarchy, and relative code and data placement to generate performant GPU GEMM kernels. tritonBLAS explicitly models the relationship between architectural topology, matrix shapes, and algorithmic blocking behavior to predict near-optimal configurations without runtime autotuning. Based on this model, we developed and implemented a lightweight GEMM framework entirely within Triton. We evaluate the performance of tritonBLAS across a diverse set of GEMM problem sizes on modern GPUs. tritonBLAS achieves over 95% of the performance of autotuning solutions, while reducing autotuning time to zero. This makes tritonBLAS a practical drop-in replacement for empirical tuning in production HPC and ML workloads.
Similar Papers
Harnessing Batched BLAS/LAPACK Kernels on GPUs for Parallel Solutions of Block Tridiagonal Systems
Mathematical Software
Speeds up computer calculations for complex problems.
Harnessing Batched BLAS/LAPACK Kernels on GPUs for Parallel Solutions of Block Tridiagonal Systems
Mathematical Software
Speeds up computer math for science and control.
Harnessing Batched BLAS/LAPACK Kernels on GPUs for Parallel Solutions of Block Tridiagonal Systems
Mathematical Software
Makes computers solve hard math problems faster.