From 21e6dd8f0b9b7b183ca20daa3488a68cb43da04a Mon Sep 17 00:00:00 2001 From: Victor Kaplansky Date: Sun, 27 Jul 2008 21:44:25 +0000 Subject: [PATCH] re PR tree-optimization/35252 (No vectorization for complex arrays) 2008-07-27 Victor Kaplansky PR tree-optimization/35252 * tree-vect-analyze.c (vect_build_slp_tree): Make IMAGPART_EXPR and REALPART_EXPR to be considered as same load operation. testsuite PR tree-optimization/35252 * gcc.dg/vect/vect-complex-1.c, gcc.dg/vect/vect-complex-2.c, gcc.dg/vect/fast-math-vect-complex-3.c, gcc.dg/vect/vect-complex-4.c: New tests. From-SVN: r138198 --- gcc/ChangeLog | 6 ++ gcc/testsuite/ChangeLog | 7 ++ .../gcc.dg/vect/fast-math-vect-complex-3.c | 61 ++++++++++++ gcc/testsuite/gcc.dg/vect/vect-complex-1.c | 56 +++++++++++ gcc/testsuite/gcc.dg/vect/vect-complex-2.c | 56 +++++++++++ gcc/testsuite/gcc.dg/vect/vect-complex-4.c | 109 +++++++++++++++++++++ gcc/tree-vect-analyze.c | 4 +- 7 files changed, 298 insertions(+), 1 deletion(-) create mode 100644 gcc/testsuite/gcc.dg/vect/fast-math-vect-complex-3.c create mode 100644 gcc/testsuite/gcc.dg/vect/vect-complex-1.c create mode 100644 gcc/testsuite/gcc.dg/vect/vect-complex-2.c create mode 100644 gcc/testsuite/gcc.dg/vect/vect-complex-4.c diff --git a/gcc/ChangeLog b/gcc/ChangeLog index c69e60e..0c271d28 100644 --- a/gcc/ChangeLog +++ b/gcc/ChangeLog @@ -1,3 +1,9 @@ +2008-07-27 Victor Kaplansky + + PR tree-optimization/35252 + * tree-vect-analyze.c (vect_build_slp_tree): Make IMAGPART_EXPR and + REALPART_EXPR to be considered as same load operation. + 2008-07-27 Eric Botcazou PR tree-optimization/36830 diff --git a/gcc/testsuite/ChangeLog b/gcc/testsuite/ChangeLog index c32bc3d..b5ee93b 100644 --- a/gcc/testsuite/ChangeLog +++ b/gcc/testsuite/ChangeLog @@ -1,3 +1,10 @@ +2008-07-27 Victor Kaplansky + + PR tree-optimization/35252 + * gcc.dg/vect/vect-complex-1.c, gcc.dg/vect/vect-complex-2.c, + gcc.dg/vect/fast-math-vect-complex-3.c, + gcc.dg/vect/vect-complex-4.c: New tests. + 2008-07-27 H.J. Lu PR c++/36944 diff --git a/gcc/testsuite/gcc.dg/vect/fast-math-vect-complex-3.c b/gcc/testsuite/gcc.dg/vect/fast-math-vect-complex-3.c new file mode 100644 index 0000000..1dff116 --- /dev/null +++ b/gcc/testsuite/gcc.dg/vect/fast-math-vect-complex-3.c @@ -0,0 +1,61 @@ +/* { dg-require-effective-target vect_float } */ + +#include +#include +#include "tree-vect.h" + +#define N 16 + +_Complex float a[N] = + { 10.0F + 20.0iF, 11.0F + 21.0iF, 12.0F + 22.0iF, 13.0F + 23.0iF, + 14.0F + 24.0iF, 15.0F + 25.0iF, 16.0F + 26.0iF, 17.0F + 27.0iF, + 18.0F + 28.0iF, 19.0F + 29.0iF, 20.0F + 30.0iF, 21.0F + 31.0iF, + 22.0F + 32.0iF, 23.0F + 33.0iF, 24.0F + 34.0iF, 25.0F + 35.0iF }; +_Complex float b[N] = + { 30.0F + 40.0iF, 31.0F + 41.0iF, 32.0F + 42.0iF, 33.0F + 43.0iF, + 34.0F + 44.0iF, 35.0F + 45.0iF, 36.0F + 46.0iF, 37.0F + 47.0iF, + 38.0F + 48.0iF, 39.0F + 49.0iF, 40.0F + 50.0iF, 41.0F + 51.0iF, + 42.0F + 52.0iF, 43.0F + 53.0iF, 44.0F + 54.0iF, 45.0F + 55.0iF }; + +_Complex float c[N]; +_Complex float res[N] = + { -500.0F + 1000.0iF, -520.0F + 1102.0iF, + -540.0F + 1208.0iF, -560.0F + 1318.0iF, + -580.0F + 1432.0iF, -600.0F + 1550.0iF, + -620.0F + 1672.0iF, -640.0F + 1798.0iF, + -660.0F + 1928.0iF, -680.0F + 2062.0iF, + -700.0F + 2200.0iF, -720.0F + 2342.0iF, + -740.0F + 2488.0iF, -760.0F + 2638.0iF, + -780.0F + 2792.0iF, -800.0F + 2950.0iF }; + + +__attribute__ ((noinline)) void +foo (void) +{ + int i; + + for (i = 0; i < N; i++) + c[i] = a[i] * b[i]; + +} + +int +main (void) +{ + int i; + check_vect (); + + foo (); + + /* check results: */ + for (i = 0; i < N; i++) + { + if (c[i] != res[i]) + abort (); + } + + return 0; +} + +/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" { target { vect_interleave && vect_extract_even_odd } } } } */ +/* { dg-final { cleanup-tree-dump "vect" } } */ diff --git a/gcc/testsuite/gcc.dg/vect/vect-complex-1.c b/gcc/testsuite/gcc.dg/vect/vect-complex-1.c new file mode 100644 index 0000000..9be8864 --- /dev/null +++ b/gcc/testsuite/gcc.dg/vect/vect-complex-1.c @@ -0,0 +1,56 @@ +/* { dg-require-effective-target vect_float } */ + +#include +#include +#include "tree-vect.h" + +#define N 16 + +_Complex float a[N] = + { 10.0F + 20.0iF, 11.0F + 21.0iF, 12.0F + 22.0iF, 13.0F + 23.0iF, + 14.0F + 24.0iF, 15.0F + 25.0iF, 16.0F + 26.0iF, 17.0F + 27.0iF, + 18.0F + 28.0iF, 19.0F + 29.0iF, 20.0F + 30.0iF, 21.0F + 31.0iF, + 22.0F + 32.0iF, 23.0F + 33.0iF, 24.0F + 34.0iF, 25.0F + 35.0iF }; +_Complex float b[N] = + { 30.0F + 40.0iF, 31.0F + 41.0iF, 32.0F + 42.0iF, 33.0F + 43.0iF, + 34.0F + 44.0iF, 35.0F + 45.0iF, 36.0F + 46.0iF, 37.0F + 47.0iF, + 38.0F + 48.0iF, 39.0F + 49.0iF, 40.0F + 50.0iF, 41.0F + 51.0iF, + 42.0F + 52.0iF, 43.0F + 53.0iF, 44.0F + 54.0iF, 45.0F + 55.0iF }; + +_Complex float c[N]; +_Complex float res[N] = + { 40.0F + 60.0iF, 42.0F + 62.0iF, 44.0F + 64.0iF, 46.0F + 66.0iF, + 48.0F + 68.0iF, 50.0F + 70.0iF, 52.0F + 72.0iF, 54.0F + 74.0iF, + 56.0F + 76.0iF, 58.0F + 78.0iF, 60.0F + 80.0iF, 62.0F + 82.0iF, + 64.0F + 84.0iF, 66.0F + 86.0iF, 68.0F + 88.0iF, 70.0F + 90.0iF }; + + +__attribute__ ((noinline)) void +foo (void) +{ + int i; + + for (i = 0; i < N; i++) + c[i] = a[i] + b[i]; + +} + +int +main (void) +{ + int i; + check_vect (); + + foo (); + + /* check results: */ + for (i = 0; i < N; i++) + if (c[i] != res[i]) + abort (); + + return 0; +} + +/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" } } */ +/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 1 "vect" } } */ +/* { dg-final { cleanup-tree-dump "vect" } } */ diff --git a/gcc/testsuite/gcc.dg/vect/vect-complex-2.c b/gcc/testsuite/gcc.dg/vect/vect-complex-2.c new file mode 100644 index 0000000..a034082 --- /dev/null +++ b/gcc/testsuite/gcc.dg/vect/vect-complex-2.c @@ -0,0 +1,56 @@ +/* { dg-require-effective-target vect_double } */ + +#include +#include +#include "tree-vect.h" + +#define N 16 + +_Complex double a[N] = + { 10.0F + 20.0iF, 11.0F + 21.0iF, 12.0F + 22.0iF, 13.0F + 23.0iF, + 14.0F + 24.0iF, 15.0F + 25.0iF, 16.0F + 26.0iF, 17.0F + 27.0iF, + 18.0F + 28.0iF, 19.0F + 29.0iF, 20.0F + 30.0iF, 21.0F + 31.0iF, + 22.0F + 32.0iF, 23.0F + 33.0iF, 24.0F + 34.0iF, 25.0F + 35.0iF }; +_Complex double b[N] = + { 30.0F + 40.0iF, 31.0F + 41.0iF, 32.0F + 42.0iF, 33.0F + 43.0iF, + 34.0F + 44.0iF, 35.0F + 45.0iF, 36.0F + 46.0iF, 37.0F + 47.0iF, + 38.0F + 48.0iF, 39.0F + 49.0iF, 40.0F + 50.0iF, 41.0F + 51.0iF, + 42.0F + 52.0iF, 43.0F + 53.0iF, 44.0F + 54.0iF, 45.0F + 55.0iF }; + +_Complex double c[N]; +_Complex double res[N] = + { 40.0F + 60.0iF, 42.0F + 62.0iF, 44.0F + 64.0iF, 46.0F + 66.0iF, + 48.0F + 68.0iF, 50.0F + 70.0iF, 52.0F + 72.0iF, 54.0F + 74.0iF, + 56.0F + 76.0iF, 58.0F + 78.0iF, 60.0F + 80.0iF, 62.0F + 82.0iF, + 64.0F + 84.0iF, 66.0F + 86.0iF, 68.0F + 88.0iF, 70.0F + 90.0iF }; + + +__attribute__ ((noinline)) void +foo (void) +{ + int i; + + for (i = 0; i < N; i++) + c[i] = a[i] + b[i]; + +} + +int +main (void) +{ + int i; + check_vect (); + + foo (); + + /* check results: */ + for (i = 0; i < N; i++) + if (c[i] != res[i]) + abort (); + + return 0; +} + +/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" } } */ +/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 1 "vect" } } */ +/* { dg-final { cleanup-tree-dump "vect" } } */ diff --git a/gcc/testsuite/gcc.dg/vect/vect-complex-4.c b/gcc/testsuite/gcc.dg/vect/vect-complex-4.c new file mode 100644 index 0000000..85c4165 --- /dev/null +++ b/gcc/testsuite/gcc.dg/vect/vect-complex-4.c @@ -0,0 +1,109 @@ +/* { dg-require-effective-target vect_int } */ + +#include +#include +#include "tree-vect.h" + +#define N 16 + +struct foostr { + _Complex short f1; + _Complex short f2; +}; + +struct foostr a[16] __attribute__ ((__aligned__(16))) = + { + 11 + 23i, 24 + 22i, + 11 + 26i, 24 + 35i, + 19 + 20i, 29 + 14i, + 23 + 31i, 26 + 30i, + 29 + 39i, 24 + 18i, + 20 + 32i, 16 + 23i, + 13 + 26i, 37 + 34i, + 12 + 23i, 26 + 14i, + 36 + 14i, 31 + 17i, + 35 + 17i, 17 + 36i, + 13 + 34i, 19 + 12i, + 27 + 34i, 36 + 19i, + 21 + 39i, 16 + 33i, + 28 + 18i, 39 + 26i, + 32 + 27i, 13 + 38i, + 35 + 36i, 34 + 28i, + }; + +struct foostr b[16] __attribute__ ((__aligned__(16))) = + { + 37 + 12i, 23 + 15i, + 14 + 11i, 13 + 25i, + 35 + 29i, 22 + 34i, + 24 + 34i, 16 + 39i, + 34 + 32i, 26 + 21i, + 34 + 36i, 11 + 37i, + 25 + 21i, 10 + 39i, + 10 + 36i, 35 + 22i, + 39 + 29i, 23 + 21i, + 34 + 33i, 39 + 14i, + 16 + 31i, 32 + 33i, + 20 + 14i, 35 + 30i, + 26 + 24i, 36 + 37i, + 31 + 20i, 32 + 28i, + 25 + 27i, 15 + 30i, + 10 + 31i, 37 + 37i, + }; +struct foostr c[16] __attribute__ ((__aligned__(16))); +struct foostr res[N] = + { + 48 + 35i, 47 + 37i, + 25 + 37i, 37 + 60i, + 54 + 49i, 51 + 48i, + 47 + 65i, 42 + 69i, + 63 + 71i, 50 + 39i, + 54 + 68i, 27 + 60i, + 38 + 47i, 47 + 73i, + 22 + 59i, 61 + 36i, + 75 + 43i, 54 + 38i, + 69 + 50i, 56 + 50i, + 29 + 65i, 51 + 45i, + 47 + 48i, 71 + 49i, + 47 + 63i, 52 + 70i, + 59 + 38i, 71 + 54i, + 57 + 54i, 28 + 68i, + 45 + 67i, 71 + 65i, + }; + +__attribute__ ((noinline)) void +foo (void) +{ + int i; + + for (i = 0; i < N; i++) + { + c[i].f1 = a[i].f1 + b[i].f1; + c[i].f2 = a[i].f2 + b[i].f2; + } + +} + +int +main (void) +{ + int i; + check_vect (); + + foo (); + + /* check results: */ + for (i = 0; i < N; i++) + { + if (c[i].f1 != res[i].f1) + abort (); + if (c[i].f2 != res[i].f2) + abort (); + } + + return 0; +} + +/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" } } */ +/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 1 "vect" } } */ +/* { dg-final { cleanup-tree-dump "vect" } } */ diff --git a/gcc/tree-vect-analyze.c b/gcc/tree-vect-analyze.c index 9f2640d..818ae7c 100644 --- a/gcc/tree-vect-analyze.c +++ b/gcc/tree-vect-analyze.c @@ -2751,7 +2751,9 @@ vect_build_slp_tree (loop_vec_info loop_vinfo, slp_tree *node, } else { - if (first_stmt_code != TREE_CODE (rhs)) + if ((first_stmt_code != TREE_CODE (rhs)) + && ((first_stmt_code != IMAGPART_EXPR) || (TREE_CODE (rhs) != REALPART_EXPR)) + && ((first_stmt_code != REALPART_EXPR) || (TREE_CODE (rhs) != IMAGPART_EXPR))) { if (vect_print_dump_info (REPORT_SLP)) { -- 2.7.4