[AMDGPU] Emit a waitcnt instruction after each memory instruction#79236

Merged

jwanggit86 merged 16 commits intollvm:mainfrom jwanggit86:waitcnt-after-each-mem-insn

Apr 10, 2024

+1,686

Commits on Apr 4, 2024

[AMDGPU] Emit a waitcnt instruction after each memory instruction
Jun Wang
committed
Combined insertions of waitcnt with existing SIMemoryLegalizer code.
Jun Wang
committed
Merge code for precise mem with the existing SICacheControl classes.
Jun Wang
committed
Add support for GFX12.
Jun Wang
committed
Some small changes based on code review.
Jun Wang
committed
Code formatting.
Jun Wang
committed
Change the option from amdgpu-precise-memory-op to precise-memory
Jun Wang
committed
Move implementation from SIMemoryLegalizer to SIInsertWaitcnts.
Jun Wang
committed
Minor code change.
Jun Wang
committed
Use getAllZeroWaitcnt() when creating the Wait obj. Some changes
Jun Wang
committed
Use update_llc_test_checks.py on insert_waitcnt_for_precise_memory.ll.
Jun Wang
committed
Replace testcases test_load_store() and test_load_store_as5() with
Jun Wang
committed
Check if mem instruciton is already immediately followed by a
Jun Wang
committed
Change iterator update operator from post-inc to pre-inc.
Jun Wang
committed
With llvm#87539 , previous commit that checks for the instruction immediately
Jun Wang
committed

Commits on Apr 9, 2024

Add testcase that covers flat_atomic_swap, an atomic without return.
Jun Wang
committed