How well does graph superoptimization transfer? Adapting tensor-graph superoptimizer to a Triton-based NPU
For automatic optimization systems, performance portability across increasingly diverse AI accelerators remains a challenge. We investigate how well graph-level superoptimization transfers across architectures by adapting Mirage from its CUDA-oriented design to the Huawei Ascend 910B3 through a Triton-based compilation stack. We evaluate the adapted system on transformer-relevant workloads, including matrix multiplication, softmax, RMSNorm+MatMul, gated MLP, FNN, and attention. The results show that graph-level optimization can successfully transfer and outperform available reference implementations by up to 87.0% on fused workloads. However, performance is highly dependent on tensor shape and workload structure, with substantial regressions for other configurations. We identify target-specific tensor partitioning, execution-resource modeling, backend lowering and fusion, and search and verification costs as key limitations. Our findings demonstrate that a portable graph representation and programming model do not imply a portable optimization space, motivating modular target-specific mechanisms within otherwise reusable compiler infrastructure.
Authors
- Weronika Hryniewska-Guzik (ORCID: https://orcid.org/0000-0003-2903-6050)
- Maciej Kokociński (ORCID: https://orcid.org/0000-0003-4640-525X)
- Daniel Murawski (ORCID: https://orcid.org/0009-0005-5881-2620)
- Tadeusz Kobus (ORCID: https://orcid.org/0000-0002-7601-1844)
- Jakub Schmiegel (ORCID: https://orcid.org/0009-0004-0313-5178)
- Maciej Świetlik
- Mateusz Comblik
Publication Details
- Journal
- arXiv (Cornell University)
- Published
- 2026-11-17
- DOI
- https://doi.org/10.48550/arxiv.2611.22844
- Primary Topic
- Parallel Computing and Optimization Techniques
- Type
- preprint