This is the mail archive of the gcc-patches@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]

Re: patch: cleanup gcc.dg/vect/ (take 2)





> Is this a sufficient set of new check_* procs?
>
>   vect_int_fp:   target supports both integer and floating point vectors
>   vect_align:    target handles mis-aligned vectors
>   vect_bitwise:  target provides vector bitwise operations
>   vect_max:      target provides a vector max operation

we probably need one more class of targets to xfail on vectorizing
testcases vect-[46,50,52,58,60,77,77a].c - xfailing on these will allow us
to close PR18403, or rather convert it to "missed optimization". We can
call it:

    vect_64bit_ptrs: 64bit linux targets? (at least powerpc64-suse-linux).


dorit




|---------+---------------------------->
|         |           Janis Johnson    |
|         |           <janis187@us.ibm.|
|         |           com>             |
|         |                            |
|         |           09/11/2004 00:00 |
|---------+---------------------------->
  >-----------------------------------------------------------------------------------------------------------------------------|
  |                                                                                                                             |
  |       To:       Dorit Naishlos/Haifa/IBM@IBMIL                                                                              |
  |       cc:       Janis Johnson <janis187@us.ibm.com>, aldyh@redhat.com, gcc-patches@gcc.gnu.org                              |
  |       Subject:  Re: patch: cleanup gcc.dg/vect/ (take 2)                                                                    |
  >-----------------------------------------------------------------------------------------------------------------------------|




On Sun, Nov 07, 2004 at 05:52:55PM +0200, Dorit Naishlos wrote:
>
> aldy wrote:
> > We want the DejaGnu stuff in the test headers to be clear, and we want
> > it to be easy to update the tests to be used for a new target.
>
> Just a thought -
> might make sense to also automate the dg-final stuff, together with the
> dg-do. My concern is that when the targets and compilation flags are no
> longer explicitly listed in the testcase header, it may become less
obvious
> to know if and which targets should xfail on:
> /* { dg-final { scan-tree-dump-times "vectorized X loops" Y "vect" } }
*/)

Thanks for the suggestion, Dorit, that will make it easier if new
targets use the tests later.

Is this a sufficient set of new check_* procs?

  vect_int_fp:   target supports both integer and floating point vectors
  vect_align:    target handles mis-aligned vectors
  vect_bitwise:  target provides vector bitwise operations
  vect_max:      target provides a vector max operation

New support in http://gcc.gnu.org/ml/gcc-patches/2004-11/msg00417.html
would allow combining check_* procs and targets, so a test could use
[dg-target-list xfail [expr [check_vect_int_fp]] && [check_vect_align]]].

If these classes of targets are only relevant for the gcc.dg/vect tests
then they can be supported within gcc.dg/vect/vect.exp; otherwise we can
support them in lib/target-supports.exp so they'll be available in other
parts of the testsuite.

Janis



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