Automated Operator Fusion for GPU-Accelerated SQL: Generation and Evolutionary Refinement
This work targets the middle ground between per-operator hand-written kernels and bespoke implementations synthesized for an entire workload or execution engine, and finds that strong code models can often generate buildable and equivalent fused Triton kernels, and that these kernels can improve end-to-end runtime when...
CoreyLammie, Yotam Perlitz, Andrea Giovannini et al.
· 0 citations