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: [autovect] merge from mainline





> I enabled it in tree-vectorizer.c and bootstrap (with and without
> vectorizer) succeeded. May be Sebastian's patch fixed it? What error
> did you see ?

maybe; I tried to fix the bootstrap failure by switching to pass
loop->depth instead of loop->num to the distance utility (to comply with
mainline changes), but bootstrap still failed for me after that, so either
sebastian fixed it, or maybe I just should have tried a clean bootstrap all
over again after I put in that fix. Anyhow, good to know that it works!

thanks,

dorit



gcc-patches-owner@gcc.gnu.org wrote on 31/01/2005 23:54:12:

>
> On Jan 30, 2005, at 11:55 AM, Dorit Naishlos wrote:
>
> > Merged mainline into autovect.
> > I temporarily disabled the dependence-distance stuff in autovect
> > because it
> > broke bootstrap after the merge, so until this functionality is
> > restored we
> > have the following test failures (on powerpc-darwin):
>
>
> I enabled it in tree-vectorizer.c and bootstrap (with and without
> vectorizer) succeeded. May be Sebastian's patch fixed it? What error
> did you see ?
>
> > FAIL: gcc.dg/vect/vect-ifcvt-1.c scan-tree-dump-times vectorized 2
> > loops 1
> > FAIL: gcc.dg/vect/vect-ifcvt-2.c scan-tree-dump-times vectorized 1
> > loops 1
> > FAIL: gcc.dg/vect/vect-ifcvt-3.c scan-tree-dump-times vectorized 1
> > loops 1
> > FAIL: gcc.dg/vect/vect-ifcvt-4.c scan-tree-dump-times vectorized 1
> > loops 1
> > FAIL: gcc.dg/vect/vect-ifcvt-5.c scan-tree-dump-times vectorized 1
> > loops 1
> > FAIL: gcc.dg/vect/vect-ifcvt-6.c scan-tree-dump-times vectorized 1
> > loops 1
> > FAIL: gcc.dg/vect/vect-ifcvt-7.c scan-tree-dump-times vectorized 1
> > loops 1
> > FAIL: gcc.dg/vect/vect-ifcvt-8.c scan-tree-dump-times vectorized 1
> > loops 1
> > FAIL: gcc.dg/vect/vect-ifcvt-9.c scan-tree-dump-times vectorized 1
> > loops 2
> > FAIL: gcc.dg/vect/vect-subsat.c scan-tree-dump-times vectorized 1
> > loops 1
> > Tested on powerpc-darwin:
> > passed bootstrap with vectorization enabled,
> > passed 'make check' with no regressions (except the ones mentioned
> > above),
> > and passed SPEC with vectorization enabled with no regressions.
> > We do have one SPEC failure with autovect, from sometime before the
> > merge -
> > an ICE in vortex during ivopts:
> > gcc -c -o query.o -O1 query.c
> > query.c: In function 'Query_EndBuild':
> > query.c:787: error: Definition in block 11 does not dominate use in
> > block
> > 12
> > for SSA_NAME: ivtmp.191_79in statement:
> > ivtmp.191_80 = PHI <ivtmp.191_79(12), 0(4)>;
> > PHI argument
> > ivtmp.191_79
> > for PHI node
> > ivtmp.191_80 = PHI <ivtmp.191_79(12), 0(4)>;
> > query.c:787: internal compiler error: verify_ssa failed.
> > (To do:
> > - restore the dependence-distance functionality in the vectorizer
> > - switch to use the new target-keywords (dg-require-effective-target
> > etc)
> > in the vect-ifcvt*.c tests
> > - fix the SPEC error)
> > dorit
>
> -
> Devang
>


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