[llvm-bugs] [Bug 32802] New: [BranchFolding] Investigate reducing the default value of tail-merge-size

via llvm-bugs llvm-bugs at lists.llvm.org
Wed Apr 26 07:09:14 PDT 2017


https://bugs.llvm.org/show_bug.cgi?id=32802

            Bug ID: 32802
           Summary: [BranchFolding] Investigate reducing the default value
                    of tail-merge-size
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: Windows NT
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Transformation Utilities
          Assignee: unassignedbugs at nondot.org
          Reporter: llvm-dev at redking.me.uk
                CC: andrew.v.tischenko at gmail.com, davide at freebsd.org,
                    filcab at gmail.com, jlpaquet at uvic.ca,
                    llvm-bugs at lists.llvm.org, matze at braunis.de

llvm\lib\CodeGen\BranchFolding.cpp:

// Heuristic for tail merging (and, inversely, tail duplication).
// TODO: This should be replaced with a target query.
static cl::opt<unsigned>
TailMergeSize("tail-merge-size",
          cl::desc("Min number of instructions to consider tail merging"),
                              cl::init(3), cl::Hidden);

Currently the tail-merge-size for duplicate BB tail merging is set to a default
value of 3 instructions, which prevents us from merging all the BBs that we
could otherwise (Bug #27136 and D30572). Additionally there appears to be no
testing of this feature at all.

Is the value purely for compile time reasons or does it have to have a value
that must correctly interact with tail-dup-placement-threshold?

Does the TODO comment suggesting making it target specific make sense?

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20170426/f6766042/attachment.html>


More information about the llvm-bugs mailing list