This is the mail archive of the gcc-prs@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

target/10404: gcc does not use 0 when using the indexed instructions of altivec like stewx


>Number:         10404
>Category:       target
>Synopsis:       gcc does not use 0 when using the indexed instructions of altivec like stewx
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Apr 14 20:56:00 UTC 2003
>Closed-Date:
>Last-Modified:
>Originator:     Andrew Pinski
>Release:        GNU C version 3.4 20030412 (experimental)
>Organization:
>Environment:
powerpc-apple-darwin6.5
>Description:
compile the follow source with gcc testaltivect1.c -include altivec.h -maltivec -O3 and see that there is `li r2, 0` which is useless.
int f(int a, int *c, vector signed int v1) {
    vec_ste(v1, 0, c);
    return a;
}
comes out:
_f:
        li r5,0
        stvewx v2,r5,r4
        blr

when it is faster and smaller to do this:
_f:
        stvewx v2,0,r4
        blr
>How-To-Repeat:
See Description
>Fix:
add or fix the altivec patterns for load/store index so that they allow one register as memory.
>Release-Note:
>Audit-Trail:
>Unformatted:


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]