For arrays [N]T where T is not a pure memory type, we currently generate separate algorithms for each distinct value of N. Instead, we could emit simple stub algs that load N into a register/argument, and then jump to a common []T implementation.
Unclear if this shows up often enough in practice to be beneficial though.
For arrays [N]T where T is not a pure memory type, we currently generate separate algorithms for each distinct value of N. Instead, we could emit simple stub algs that load N into a register/argument, and then jump to a common []T implementation.
Unclear if this shows up often enough in practice to be beneficial though.
Related #17029.
The text was updated successfully, but these errors were encountered: