[Issue 9190] New: Vector operations are not optimized for x86_64 architecture

d-bugmail at puremagic.com d-bugmail at puremagic.com
Thu Dec 20 12:49:59 PST 2012


http://d.puremagic.com/issues/show_bug.cgi?id=9190

           Summary: Vector operations are not optimized for x86_64
                    architecture
           Product: D
           Version: D2
          Platform: x86_64
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: druntime
        AssignedTo: nobody at puremagic.com
        ReportedBy: lomereiter at gmail.com


--- Comment #0 from Artem Tarasov <lomereiter at gmail.com> 2012-12-20 12:49:53 PST ---
>From what I see in druntime/src/rt/arrayint.d, there are some heavy
optimizations for x86 architecture but not for x86_64.

For instance, in my code replacing a vector operation which resulted in a call
to _arrayExpSliceAddass_k, with a loop, gave me 2x performance boost with -O
-release -inline flags (and even more with LDC compiler).

As a temporary solution (until optimized versions will be written), I suggest
to enhance the following condition by taking target architecture into account:
https://github.com/D-Programming-Language/dmd/blob/master/src/arrayop.c#L304

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------


More information about the Digitalmars-d-bugs mailing list