[llvm-bugs] [Bug 28279] New: Re-write "lemon" so that it doesn't fork nor spend most of its time on printf

via llvm-bugs llvm-bugs at lists.llvm.org
Thu Jun 23 05:52:52 PDT 2016


https://llvm.org/bugs/show_bug.cgi?id=28279

            Bug ID: 28279
           Summary: Re-write "lemon" so that it doesn't fork nor spend
                    most of its time on printf
           Product: Test Suite
           Version: trunk
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P
         Component: Programs Tests
          Assignee: unassignedbugs at nondot.org
          Reporter: renato.golin at linaro.org
                CC: llvm-bugs at lists.llvm.org
    Classification: Unclassified

A recent review has pointed out that lemon is a broken design:

http://reviews.llvm.org/D21593

Not only it has to fork 20 times to work around running the test with 20
different inputs, but it also spends "60-70% of the time is spend in fputs and
fputc".

>From what I can see, there's nothing stopping the code from operating from a
constant array instead of command line arguments. Even if it's called
differently for different targets / training dataset, it still can be a single
argument which chooses between different hard-coded choices.

Also, the fputs is probably not needed at all, and refactoring like we've done
before (accumulate the data in a small group of variables, keep them in memory,
etc) should be much more reliable and stress the compiler more than the current
situation.

Of course, larger arrays versus reduction have very different cost problems, so
we should stick to whatever is closer to the original intent, not the hacked up
implementation.

-- 
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/20160623/b82130f7/attachment.html>


More information about the llvm-bugs mailing list