~mcf/cproc#58: 
Inefficient copies of large aggregate types

Currently we implement struct assignment by copying in increments according to the struct alignment.

However, this is not what we want for very large structs. It causes bad QBE behavior, and produces huge code.

Other compilers emit calls to memcpy when in a hosted environment. We should figure out a good cut-off and do the same.

Status
REPORTED
Submitter
~mcf
Assigned to
No-one
Submitted
1 year, 25 days ago
Updated
1 year, 25 days ago
Labels
performance