A polyhedral transformation technique identifies hidden matrix multiplications in programs and substitutes them with specialized CGRA kernels, delivering up to 9.1x speedups on benchmarks with various CGRA sizes.
An mlir-based compilation framework for control flow management on cgras
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Exploiting pre-optimized kernels with polyhedral transformations for CGRA compilation
A polyhedral transformation technique identifies hidden matrix multiplications in programs and substitutes them with specialized CGRA kernels, delivering up to 9.1x speedups on benchmarks with various CGRA sizes.