Closed
Description
This is a meta issue for tracking which bytecodes have been specialized, need to be specialized, need to be improved, or aren't worth specializing.
Completed
- LOAD_GLOBAL Port LOAD_GLOBAL from old opcache to PEP 659 adaptive interpreter. #51
- BINARY_MULTIPLY
Done, but could be improved
- LOAD_ATTR Port LOAD_ATTR from old opcache to PEP 659 adaptive interpreter #52
- BINARY_OP
- BINARY_SUBSCR Implement BINARY_SUBSCR specialization "family" for PEP 659 adaptive interpreter #59
- STORE_ATTR Implement STORE_ATTR specialization "family" for PEP 659 adaptive interpreter #53
- LOAD_METHOD Specialize LOAD_METHOD #81
- STORE_SUBSCR
In progress
- CALL_NO_KW
To do
- CALL_KW
Needs more data
Before more optimizing these, we need to know how well they will specialize
- FOR_ITER Consider FOR_ITER family for specialization? #67
- FORMAT_VALUE bpo-33092: The bytecode for f-string formatting is inefficient. #61
Not going to do
- All bytecodes whose behavior does not depend on the value(s) operated on, e.g. LOAD_FAST