[llvm-bugs] [Bug 26837] New: Availability of SSE detrimental to code performance

[via llvm-bugs]

https://llvm.org/bugs/show_bug.cgi?id=26837

            Bug ID: 26837
           Summary: Availability of SSE detrimental to code performance
           Product: libraries
           Version: 3.8
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P
         Component: Backend: X86
          Assignee: unassignedbugs at nondot.org
          Reporter: tulipawn at gmail.com
                CC: llvm-bugs at lists.llvm.org
    Classification: Unclassified

The following Rust playpen snippet (possible to see IR or ASM interactively)

http://is.gd/FrGoDm

produces the following benchmark results, depending on codegen options:


Pentium2

test folds1 ... bench:         736 ns/iter (+/- 5)
test folds2 ... bench:         731 ns/iter (+/- 1)

Pentium3 (or +sse)

test folds1 ... bench:         932 ns/iter (+/- 2)
test folds2 ... bench:         933 ns/iter (+/- 8)

Pentium4 (or +sse2)

running 2 tests
test folds1 ... bench:         249 ns/iter (+/- 1)
test folds2 ... bench:         733 ns/iter (+/- 6)

It seems availability of SSE2 enables successful vectorisation, whereas of SSE 
makes the code slower. The latter shouldn't happen.

-- 
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/20160304/e94b7715/attachment.html>