Closed Bug 587735 Opened 10 years ago Closed 10 years ago
Register Alloc From Set in terms of ls Reg or ms Reg
Once bug 584935 lands, the platform-specific implementations of nRegisterAllocFromSet could be rewritten in terms of lsReg or msReg, to ease maintenance. We might also get a slight speed bump from them having been inlined.
The speedup is 2.4% for the Tamarin's "performance" testsuite on SH4. Thanks for this suggestion!
Attachment #473989 - Flags: review?(edwsmith)
Comment on attachment 473989 [details] [diff] [review] SH4 backend: redefine Assembler::nRegisterAllocFromSet() in terms of lsReg Cool. Just curious, how does GCC implement __builtin_ctz() on SH4? I didn't think the CPU had a bit-index primitive.
Attachment #473989 - Flags: review?(edwsmith) → review+
Please, could you apply this change? Thanks in advance.
Whiteboard: fixed-in-nanojit → fixed-in-nanojit, fixed-in-tracemonkey
Status: NEW → RESOLVED
Closed: 10 years ago
Resolution: --- → FIXED
Whiteboard: fixed-in-nanojit, fixed-in-tracemonkey → fixed-in-nanojit,fixed-in-tracemonkey,fixed-in-tamarin
You need to log in before you can comment on or make changes to this bug.