44284 – vectorization does not work for short variable

Bug 44284 - vectorization does not work for short variable

Summary: vectorization does not work for short variable

Status:	RESOLVED FIXED

Alias:	None

Product:	gcc
Classification:	Unclassified
Component:	tree-optimization (show other bugs)
Version:	4.4.3

Importance:	P3 enhancement
Target Milestone:	4.5.1
Assignee:	Richard Biener

URL:
Keywords:	missed-optimization

Depends on:
Blocks:

Reported:	2010-05-26 15:36 UTC by Roy Rosen
Modified:	2010-07-08 13:12 UTC (History)
CC List:	2 users (show)

See Also:
Host:
Target:
Build:
Known to work:
Known to fail:
Last reconfirmed:	2010-05-26 15:40:22

Attachments
preprocessed file (142 bytes, text/plain) 2010-05-26 15:38 UTC, Roy Rosen	Details
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.

Description Roy Rosen 2010-05-26 15:36:51 UTC

Hi,

I have tried vectorization and encountered a problem which I can see
is common to some ports (I tried ia64 and bfin).

For this function:

#define ts unsigned short
void f(ts* __restrict__ a, ts* __restrict__ b, ts* __restrict__ x)
{
   int i;
   for (i=0;i<1024;i++)
       x[i] = a[i] + b[i];
}

the loop is vectorized.
But if I define ts as follows: #define ts short
then the loop is not vectorized. The message I get is:

./a.c:21: note: no optab.
./a.c:21: note: not vectorized: relevant stmt not supported: D.1279_12
= (short unsigned int) D.1278_11;

I have tried to look a bit in the vectorizer code and it seems that
for this stmt I get to vectorizable_operation with code==NOP_EXPR
which is not handled.

Does anyone knows anything about this problem?

Thanks, Roy.

./xgcc -v -save-temps -O3 -S ./a.c -ftree-vectorizer-verbose=2
Using built-in specs.
Target: ia64-*-linux
Configured with: ../gcc-4.4.3/configure --target='ia64-*-linux' --enable-languages=c : (reconfigured)
Thread model: posix
gcc version 4.4.3 (GCC)
COLLECT_GCC_OPTIONS='-v' '-save-temps' '-O3' '-S' '-ftree-vectorizer-verbose=2'
 cc1 -E -quiet -v -iprefix /home/swproj/sw/users/eyalhar/gcc-ia64/gcc/../lib/gcc/ia64-*-linux/4.4.3/ ./a.c -ftree-vectorizer-verbose=2 -O3 -fpch-preprocess -o a.i
ignoring nonexistent directory "/home/swproj/sw/users/eyalhar/gcc-ia64/gcc/../lib/gcc/ia64-*-linux/4.4.3/include"
ignoring nonexistent directory "/home/swproj/sw/users/eyalhar/gcc-ia64/gcc/../lib/gcc/ia64-*-linux/4.4.3/include-fixed"
ignoring nonexistent directory "/home/swproj/sw/users/eyalhar/gcc-ia64/gcc/../lib/gcc/ia64-*-linux/4.4.3/../../../../ia64-*-linux/sys-include"
ignoring nonexistent directory "/home/swproj/sw/users/eyalhar/gcc-ia64/gcc/../lib/gcc/ia64-*-linux/4.4.3/../../../../ia64-*-linux/include"
ignoring nonexistent directory "/home/swproj/sw/users/eyalhar/gcc-ia64/gcc/../lib/gcc/../../lib/gcc/ia64-*-linux/4.4.3/include"
ignoring nonexistent directory "/home/swproj/sw/users/eyalhar/gcc-ia64/gcc/../lib/gcc/../../lib/gcc/ia64-*-linux/4.4.3/include-fixed"
ignoring nonexistent directory "/home/swproj/sw/users/eyalhar/gcc-ia64/gcc/../lib/gcc/../../lib/gcc/ia64-*-linux/4.4.3/../../../../ia64-*-linux/sys-include"
ignoring nonexistent directory "/home/swproj/sw/users/eyalhar/gcc-ia64/gcc/../lib/gcc/../../lib/gcc/ia64-*-linux/4.4.3/../../../../ia64-*-linux/include"
#include "..." search starts here:
#include <...> search starts here:
End of search list.
COLLECT_GCC_OPTIONS='-v' '-save-temps' '-O3' '-S' '-ftree-vectorizer-verbose=2'
 cc1 -fpreprocessed a.i -quiet -dumpbase a.c -auxbase a -O3 -version -ftree-vectorizer-verbose=2 -o a.s
GNU C (GCC) version 4.4.3 (ia64-*-linux)
        compiled by GNU C version 4.1.2 20080704 (Red Hat 4.1.2-44), GMP version 4.3.2, MPFR version 2.4.2.
GGC heuristics: --param ggc-min-expand=100 --param ggc-min-heapsize=131072
Compiler executable checksum: cc15465375f06438c45a5c1ccdf35a17

./a.c:21: note: not vectorized: relevant stmt not supported: D.1236_12 = (short unsigned int) D.1235_11;

./a.c:18: note: vectorized 0 loops in function.
COMPILER_PATH=
LIBRARY_PATH=
COLLECT_GCC_OPTIONS='-v' '-save-temps' '-O3' '-S' '-ftree-vectorizer-verbose=2'

Comment 1 Roy Rosen 2010-05-26 15:38:21 UTC

Created attachment 20752 [details]
preprocessed file

Comment 2 Richard Biener 2010-05-26 15:40:22 UTC

Confirmed.  Mine.

Comment 3 Richard Biener 2010-05-27 13:24:15 UTC

Subject: Bug 44284

Author: rguenth
Date: Thu May 27 13:23:45 2010
New Revision: 159920

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=159920
Log:
2010-05-27  Richard Guenther  <rguenther@suse.de>

	PR tree-optimization/44284
	* tree-vect-stmts.c (vectorizable_assignment): Handle
	sign-changing conversions as simple copy.

	* gcc.dg/vect/vect-118.c: New testcase.
	* gcc.dg/vect/bb-slp-20.c: Adjust.
	* gcc.dg/vect/no-section-anchors-vect-36.c: Likewise.
	* gcc.dg/vect/slp-9.c: Likewise.
	* gcc.dg/vect/slp-reduc-4.c: Likewise.
	* gcc.dg/vect/vect-10.c: Likewise.
	* gcc.dg/vect/vect-109.c: Likewise.
	* gcc.dg/vect/vect-12.c: Likewise.
	* gcc.dg/vect/vect-36.c: Likewise.
	* gcc.dg/vect/vect-7.c: Likewise.
	* gcc.dg/vect/vect-iv-8.c: Likewise.
	* gcc.dg/vect/vect-multitypes-10.c: Likewise.
	* gcc.dg/vect/vect-multitypes-13.c: Likewise.
	* gcc.dg/vect/vect-multitypes-14.c: Likewise.
	* gcc.dg/vect/vect-multitypes-15.c: Likewise.
	* gcc.dg/vect/vect-multitypes-7.c: Likewise.
	* gcc.dg/vect/vect-multitypes-8.c: Likewise.
	* gcc.dg/vect/vect-multitypes-9.c: Likewise.
	* gcc.dg/vect/vect-reduc-dot-s16b.c: Likewise.
	* gcc.dg/vect/vect-reduc-dot-s8a.c: Likewise.
	* gcc.dg/vect/vect-reduc-dot-s8b.c: Likewise.
	* gcc.dg/vect/vect-reduc-dot-u16b.c: Likewise.
	* gcc.dg/vect/vect-strided-a-u32-mult.c: Likewise.
	* gcc.dg/vect/vect-strided-u32-mult.c: Likewise.
	* gcc.dg/vect/vect-widen-mult-s16.c: Likewise.
	* gcc.dg/vect/vect-widen-mult-s8.c: Likewise.
	* gcc.dg/vect/vect-widen-mult-sum.c: Likewise.
	* gcc.dg/vect/vect-widen-mult-u16.c: Likewise.

Added:
    trunk/gcc/testsuite/gcc.dg/vect/vect-118.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/testsuite/gcc.dg/vect/bb-slp-20.c
    trunk/gcc/testsuite/gcc.dg/vect/no-section-anchors-vect-36.c
    trunk/gcc/testsuite/gcc.dg/vect/slp-9.c
    trunk/gcc/testsuite/gcc.dg/vect/slp-reduc-4.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-10.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-109.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-12.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-36.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-7.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-iv-8.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-multitypes-10.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-multitypes-13.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-multitypes-14.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-multitypes-15.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-multitypes-7.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-multitypes-8.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-multitypes-9.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-reduc-dot-s16a.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-reduc-dot-s16b.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-reduc-dot-s8a.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-reduc-dot-s8b.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-reduc-dot-s8c.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-reduc-dot-u16b.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-strided-a-u32-mult.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-strided-u32-mult.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-widen-mult-s16.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-widen-mult-s8.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-widen-mult-sum.c
    trunk/gcc/testsuite/gcc.dg/vect/vect-widen-mult-u16.c
    trunk/gcc/tree-vect-stmts.c

Comment 4 Richard Biener 2010-05-27 13:24:28 UTC

Fixed.

Comment 5 Richard Biener 2010-07-08 11:56:42 UTC

Subject: Bug 44284

Author: rguenth
Date: Thu Jul  8 11:56:08 2010
New Revision: 161951

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=161951
Log:
2010-07-08  Richard Guenther  <rguenther@suse.de>

	Backport from mainline
	2010-05-27  Richard Guenther  <rguenther@suse.de>

	PR tree-optimization/44284
	* tree-vect-stmts.c (vectorizable_assignment): Handle
	sign-changing conversions as simple copy.

	* gcc.dg/vect/vect-118.c: New testcase.
	* gcc.dg/vect/bb-slp-20.c: Adjust.
	* gcc.dg/vect/no-section-anchors-vect-36.c: Likewise.
	* gcc.dg/vect/slp-9.c: Likewise.
	* gcc.dg/vect/slp-reduc-4.c: Likewise.
	* gcc.dg/vect/vect-10.c: Likewise.
	* gcc.dg/vect/vect-109.c: Likewise.
	* gcc.dg/vect/vect-12.c: Likewise.
	* gcc.dg/vect/vect-36.c: Likewise.
	* gcc.dg/vect/vect-7.c: Likewise.
	* gcc.dg/vect/vect-iv-8.c: Likewise.
	* gcc.dg/vect/vect-multitypes-10.c: Likewise.
	* gcc.dg/vect/vect-multitypes-13.c: Likewise.
	* gcc.dg/vect/vect-multitypes-14.c: Likewise.
	* gcc.dg/vect/vect-multitypes-15.c: Likewise.
	* gcc.dg/vect/vect-multitypes-7.c: Likewise.
	* gcc.dg/vect/vect-multitypes-8.c: Likewise.
	* gcc.dg/vect/vect-multitypes-9.c: Likewise.
	* gcc.dg/vect/vect-reduc-dot-s16b.c: Likewise.
	* gcc.dg/vect/vect-reduc-dot-s8a.c: Likewise.
	* gcc.dg/vect/vect-reduc-dot-s8b.c: Likewise.
	* gcc.dg/vect/vect-reduc-dot-u16b.c: Likewise.
	* gcc.dg/vect/vect-strided-a-u32-mult.c: Likewise.
	* gcc.dg/vect/vect-strided-u32-mult.c: Likewise.
	* gcc.dg/vect/vect-widen-mult-s16.c: Likewise.
	* gcc.dg/vect/vect-widen-mult-s8.c: Likewise.
	* gcc.dg/vect/vect-widen-mult-sum.c: Likewise.
	* gcc.dg/vect/vect-widen-mult-u16.c: Likewise.

	2010-07-06  Richard Guenther  <rguenther@suse.de>

	PR middle-end/44828
	* convert.c (convert_to_integer): Watch out for overflowing
	MULT_EXPR as well.

	* gcc.c-torture/execute/pr44828.c: New testcase.

Added:
    branches/gcc-4_5-branch/gcc/testsuite/gcc.c-torture/execute/pr44828.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-118.c
Modified:
    branches/gcc-4_5-branch/gcc/ChangeLog
    branches/gcc-4_5-branch/gcc/convert.c
    branches/gcc-4_5-branch/gcc/testsuite/ChangeLog
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/bb-slp-20.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/no-section-anchors-vect-36.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/slp-9.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-10.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-109.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-12.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-36.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-7.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-iv-8.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-multitypes-10.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-multitypes-13.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-multitypes-14.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-multitypes-15.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-multitypes-7.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-multitypes-8.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-multitypes-9.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-reduc-dot-s16a.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-reduc-dot-s16b.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-reduc-dot-s8a.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-reduc-dot-s8b.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-reduc-dot-s8c.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-reduc-dot-u16b.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-strided-a-u32-mult.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-strided-u32-mult.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-widen-mult-s16.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-widen-mult-s8.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-widen-mult-sum.c
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/vect/vect-widen-mult-u16.c
    branches/gcc-4_5-branch/gcc/tree-vect-stmts.c