Is it possible to use native acceleration with single core compute?

Ask the Community
,
1

I want to use native acceleration but the data volume is not massive, so :

@configure(profile=[...],
    backend=ComputeBackend.VELOX)

I would like to use a single node compute, so, the spark profile:

@configure(profile=["KUBERNETES_NO_EXECUTORS", ...])

Is it possible to use both at the same time ?

As it seems that the OFFHEAP Fraction is enforced, such below statement would not work ? For instance:

@configure(profile=["KUBERNETES_NO_EXECUTORS", ..."EXECUTOR_MEMORY_EXTRA_EXTRA_SMALL", "EXECUTOR_MEMORY_OFFHEAP_FRACTION_HIGH"],
    backend=ComputeBackend.VELOX)