i have write some python code to start the performance record, in which can see the memory cost

click the button that can see the memory cost when in sop ,is almost 50MB

but when i cook the hda in topnet hdaprocessor ,it cost more memory,almost 360MB

so why it will cost more memory in hdaprocessor? can it be improved?

Attachments:
my_way_to_get_the_record.png (821.4 KB)
memory_cost_directly.png (78.2 KB)
memory_cost_in_hdaprocessor.png (65.9 KB)