SageMaker Processing Job (Spark) vs. AWS Glue Job (Spark) - any baseline comparison?

curious has anyone done a baseline comparison of cost/performance (using same dataset and spark transformations) b/w SageMaker Processing Job (Spark container) vs an AWS Glue Job (Spark) ?

Looking at pricing SM Processing Job w/ Spark using 1x ml.m5.xlarge instance (4 vCPU and 16 GB of memory) = $0.23/hr vs AWS Glue Job (priced by DPU, where/ 1 DPU provides 4 vCPU and 16 GB of memory (~ 1 ml.m5.xlarge ?) comes out to $0.44 per DPU-Hour for Apache Spark