Issue 184313
Summary [RISCV] The instructions of satd_4x4 increases a lot from LLVM 21 to LLVM 22
Labels regression:22
Assignees
Reporter wangpc-pp
    See: https://godbolt.org/z/3zhWv9r3q

Though we generate unit segment loads instead of strided loads, we cost a lot to shuffle the vectors.
_______________________________________________
llvm-bugs mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs

Reply via email to