[LLVMbugs] [Bug 11439] New: Operations on double3 mis-optimised to double2

[bugzilla-daemon at llvm.org]

http://llvm.org/bugs/show_bug.cgi?id=11439

             Bug #: 11439
           Summary: Operations on double3 mis-optimised to double2
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P
         Component: Common Code Generator Code
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: schnetter at gmail.com
                CC: llvmbugs at cs.uiuc.edu
    Classification: Unclassified


I am using the release candidate llvm-3.0, checked out from
http://llvm.org/svn/llvm-project/llvm/branches/release_30 and
http://llvm.org/svn/llvm-project/cfe/branches/release_30.

This code tests double3, i.e. a vector of 3 doubles:

typedef double double3 __attribute__((__ext_vector_type__(3)));
double3 add(double3 x) { return x+1.0; }

When compiled without optimisation, the generated code is correct. When
compiled with optimisation, the generated code handles only 2 vector elements.
That is, there is only a single addpd instruction, but there should be two (or
one addpd and one addps).


clang -S -O double3.c 


    .section    __TEXT,__text,regular,pure_instructions
    .section    __TEXT,__const
    .align    4
LCPI0_0:
    .quad    4607182418800017408     ## double 1.000000e+00
    .quad    4607182418800017408     ## double 1.000000e+00
    .section    __TEXT,__text,regular,pure_instructions
    .globl    _add
    .align    4, 0x90
_add:                                   ## @add
Ltmp2:
    .cfi_startproc
## BB#0:
    pushq    %rbp
Ltmp3:
    .cfi_def_cfa_offset 16
Ltmp4:
    .cfi_offset %rbp, -16
    movq    %rsp, %rbp
Ltmp5:
    .cfi_def_cfa_register %rbp
    addpd    LCPI0_0(%rip), %xmm0
    popq    %rbp
    ret
Ltmp6:
    .cfi_endproc
Leh_func_end0:


.subsections_via_symbols

-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.