This should be implemented in Benchmark dialect and BenchmarkToAffine conversion. Note that our lowering is template-based and just for benchmarking, which means the operation parameters are not necessary to be too comprehensive and detailed. They should just reflect the most important characteristics of the operation.
This should be implemented in Benchmark dialect and BenchmarkToAffine conversion. Note that our lowering is template-based and just for benchmarking, which means the operation parameters are not necessary to be too comprehensive and detailed. They should just reflect the most important characteristics of the operation.