Align specialization guards and caching with CPython - #7341

Merged

youknowone merged 31 commits into

RustPython:mainfrom

youknowone:specialization

Mar 4, 2026

Commits on Mar 4, 2026

fix jit
youknowone
committed
c6b6842
vm: complete specialized opcode dispatch paths
youknowone
committed
b379239
vm: cache LOAD_GLOBAL with dict entry hints
youknowone
committed
bdeca2b
vm: align adaptive specialization counters with CPython backoff
youknowone
committed
ca3814a
vm: apply cooldown counter on specialization success paths
youknowone
committed
f04bf22
vm: retain LOAD_GLOBAL specializations on misses
youknowone
committed
02a1495
vm: keep attr and call specializations on guard misses
youknowone
committed
13125d8
vm: retain store-attr and store-subscr specializations on misses
youknowone
committed
c7007b5
vm: retain specialization opcodes on generic fallback paths
youknowone
committed
453294d
vm: align jump-backward specialization defaults with CPython
youknowone
committed
b8f6cc2
vm: retain exact-args call specializations on misses
youknowone
committed
7bfa961
vm: retain SEND_GEN specialization on non-coroutine sends
youknowone
committed
c852994
vm: specialize list.append calls like CPython CALL_LIST_APPEND
youknowone
committed
1a1bbf5
vm: set cooldown on LOAD_ATTR_CLASS specialization
youknowone
committed
bdd87f4
vm: specialize bound method object CALL paths
youknowone
committed
696e57a
vm: specialize CALL_KW for bound method objects
youknowone
committed
3dc64cc
vm: use current-state function version for CALL_KW specialization
youknowone
committed
f220a24
vm: align CALL/CALL_KW pyfunction specialization with CPython
youknowone
committed
a66c706
vm: drop call-site identity caches in generic CALL specializations
youknowone
committed
59545f5
vm: align builtin type call specializations with CPython guards
youknowone
committed
f8eebec
vm: align builtin CALL guards with CPython self_or_null semantics
youknowone
committed
58bb1ea
vm: require exact list in CALL_LIST_APPEND fast path
youknowone
committed
157a2c3
vm: align CALL builtin/class specialization flow with CPython
youknowone
committed
cae26d4
vm: tighten len/isinstance CALL specializations to builtin guards
youknowone
committed
0d01f7e
vm: gate CALL_BUILTIN_CLASS on type vectorcall like CPython
youknowone
committed
22cccae
vm: run non-py CALL specializations via direct vectorcall
youknowone
committed
30e8a75
vm: align class-call specialization branching with CPython
youknowone
committed
da36859
Fix CI: disable ForIterGen, tighten CALL guards
youknowone
committed
42d81ff
vm: restore FOR_ITER_GEN specialization and tuple index parity
youknowone
committed
d464119
Add datastack-backed FastLocals for non-generator frames
youknowone
committed
0f044d6
Drop read lock before key_eq in dict get_hint
youknowone
committed
cc3fd09

Navigation Menu

Uh oh!

Align specialization guards and caching with CPython - #7341

Align specialization guards and caching with CPython#7341
youknowone merged 31 commits into
RustPython:mainfrom
youknowone:specialization

Commits on Mar 4, 2026

fix jit

vm: complete specialized opcode dispatch paths

vm: cache LOAD_GLOBAL with dict entry hints

vm: align adaptive specialization counters with CPython backoff

vm: apply cooldown counter on specialization success paths

vm: retain LOAD_GLOBAL specializations on misses

vm: keep attr and call specializations on guard misses

vm: retain store-attr and store-subscr specializations on misses

vm: retain specialization opcodes on generic fallback paths

vm: align jump-backward specialization defaults with CPython

vm: retain exact-args call specializations on misses

vm: retain SEND_GEN specialization on non-coroutine sends

vm: specialize list.append calls like CPython CALL_LIST_APPEND

vm: set cooldown on LOAD_ATTR_CLASS specialization

vm: specialize bound method object CALL paths

vm: specialize CALL_KW for bound method objects

vm: use current-state function version for CALL_KW specialization

vm: align CALL/CALL_KW pyfunction specialization with CPython

vm: drop call-site identity caches in generic CALL specializations

vm: align builtin type call specializations with CPython guards

vm: align builtin CALL guards with CPython self_or_null semantics

vm: require exact list in CALL_LIST_APPEND fast path

vm: align CALL builtin/class specialization flow with CPython

vm: tighten len/isinstance CALL specializations to builtin guards

vm: gate CALL_BUILTIN_CLASS on type vectorcall like CPython

vm: run non-py CALL specializations via direct vectorcall

vm: align class-call specialization branching with CPython

Fix CI: disable ForIterGen, tighten CALL guards

vm: restore FOR_ITER_GEN specialization and tuple index parity

Add datastack-backed FastLocals for non-generator frames

Drop read lock before key_eq in dict get_hint