On Tue, 27 Oct 2020, sunil.k.pandey wrote:
> On Linux/x86_64,
>
> 5af1e827bbb624eb28f80d2c5e0da46185af3708 is the first bad commit
> commit 5af1e827bbb624eb28f80d2c5e0da46185af3708
> Author: Richard Biener <rguent...@suse.de>
> Date: Tue Oct 27 11:03:27 2020 +0100
>
> Avoid uniform lane BB vectorization
>
> caused
>
> FAIL: gfortran.dg/vect/pr83232.f90 -O scan-tree-dump-times slp1
> "vectorizing stmts using SLP" 3
>
> with GCC configured with
>
> ../../gcc/configure
> --prefix=/local/skpandey/gccwork/toolwork/gcc-bisect-master/master/r11-4427/usr
> --enable-clocale=gnu --with-system-zlib --with-demangler-in-ld
> --with-fpmath=sse --enable-languages=c,c++,fortran --enable-cet --without-isl
> --enable-libmpx x86_64-linux --disable-bootstrap
>
> To reproduce:
>
> $ cd {build_dir}/gcc && make check
> RUNTESTFLAGS="vect.exp=gfortran.dg/vect/pr83232.f90
> --target_board='unix{-m32\ -march=cascadelake}'"
> $ cd {build_dir}/gcc && make check
> RUNTESTFLAGS="vect.exp=gfortran.dg/vect/pr83232.f90
> --target_board='unix{-m64\ -march=cascadelake}'"
>
> (Please do not reply to this email, for question about this report, contact
> me at skpgkp2 at gmail dot com)
So the testcase is now improved with AVX but the dump scanning doesn't
match the need to split the vectorization at (1) into a AVX and SSE
part. There's no suitable target selector yet, the vect_sizes_32B_16B
one is an exact match and doesn't fit. I filed PR97611 for the
testsuite harness issue.
