===================================================================
@@ -8874,6 +8874,10 @@ The maximum number of conditional stores paires th
if either vectorization (@option{-ftree-vectorize}) or if-conversion
(@option{-ftree-loop-if-convert}) is disabled. The default is 2.
+@item preferred-vector-size
+Preferred vector size in bits for targets that support multiple vector sizes.
+Invalid values are ignored. The default is 128.
+
@end table
@end table
===================================================================
@@ -204,6 +204,8 @@ extern void init_param_values (int *params);
PARAM_VALUE (PARAM_PREFETCH_MIN_INSN_TO_MEM_RATIO)
#define MIN_NONDEBUG_INSN_UID \
PARAM_VALUE (PARAM_MIN_NONDEBUG_INSN_UID)
+#define PREFERRED_VECTOR_SIZE \
+ PARAM_VALUE (PARAM_PREFERRED_VECTOR_SIZE)
#define MAX_STORES_TO_SINK \
PARAM_VALUE (PARAM_MAX_STORES_TO_SINK)
#endif /* ! GCC_PARAMS_H */
===================================================================
@@ -3203,6 +3203,24 @@ proc check_effective_target_vect_strided_wide { }
return $et_vect_strided_wide_saved
}
+# Return 1 if the target supports multiple vector sizes
+
+proc check_effective_target_vect_multiple_sizes { } {
+ global et_vect_multiple_sizes
+
+ if [info exists et_vect_multiple_sizes_saved] {
+ verbose "check_effective_target_vect_multiple_sizes: using cached result" 2
+ } else {
+ set et_vect_multiple_sizes_saved 0
+ if { ([istarget arm*-*-*] && [check_effective_target_arm_neon]) } {
+ set et_vect_multiple_sizes_saved 1
+ }
+ }
+
+ verbose "check_effective_target_vect_multiple_sizes: returning $et_vect_multiple_sizes_saved" 2
+ return $et_vect_multiple_sizes_saved
+}
+
# Return 1 if the target supports section-anchors
proc check_effective_target_section_anchors { } {
@@ -3585,9 +3603,9 @@ proc add_options_for_bind_pic_locally { flags } {
# Add to FLAGS the flags needed to enable 128-bit vectors.
-proc add_options_for_quad_vectors { flags } {
+proc add_options_for_double_vectors { flags } {
if [is-effective-target arm_neon_ok] {
- return "$flags -mvectorize-with-neon-quad"
+ return "$flags --param preferred-vector-size=64"
}
return $flags
===================================================================
@@ -19,6 +19,7 @@ subroutine s111 (ntimes,ld,n,ctime,dtime,a,b,c,d,e
end
! { dg-final { scan-tree-dump-times "vectorized 1 loops" 0 "vect" } }
-! { dg-final { scan-tree-dump-times "complicated access pattern" 1 "vect" } }
+! { dg-final { scan-tree-dump-times "complicated access pattern" 1 "vect" { xfail vect_multiple_sizes } } }
+! { dg-final { scan-tree-dump-times "complicated access pattern" 2 "vect" { target vect_multiple_sizes } } }
! { dg-final { cleanup-tree-dump "vect" } }
===================================================================
@@ -1,38 +0,0 @@
-/* { dg-require-effective-target vect_float } */
-
-#include <stdarg.h>
-#include "tree-vect.h"
-
-#define N 16
-#define DIFF 240
-
-__attribute__ ((noinline))
-int main1 ()
-{
- int i;
- float b[N] = {0,3,6,9,12,15,18,21,24,27,30,33,36,39,42,45};
- float c[N] = {0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15};
- float diff;
-
- diff = 0;
- for (i = 0; i < N; i++) {
- diff += (b[i] - c[i]);
- }
-
- /* check results: */
- if (diff != DIFF)
- abort ();
-
- return 0;
-}
-
-int main (void)
-{
- check_vect ();
-
- return main1 ();
-}
-
-/* Requires fast-math. */
-/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" { xfail *-*-* } } } */
-/* { dg-final { cleanup-tree-dump "vect" } } */
===================================================================
@@ -1,4 +1,5 @@
/* { dg-require-effective-target vect_float } */
+/* { dg-add-options double_vectors } */
#include <stdarg.h>
#include "tree-vect.h"
===================================================================
@@ -44,6 +44,7 @@ int main (void)
/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 2 "vect" { xfail { vect_no_int_add || { ! vect_unpack } } } } } */
/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 0 "vect" } } */
-/* { dg-final { scan-tree-dump-times "different interleaving chains in one node" 1 "vect" { target { ! vect_no_int_add } } } } */
+/* { dg-final { scan-tree-dump-times "different interleaving chains in one node" 1 "vect" { xfail { vect_multiple_sizes || vect_no_int_add } } } } */
+/* { dg-final { scan-tree-dump-times "different interleaving chains in one node" 2 "vect" { target { {! vect_no_int_add} && vect_multiple_sizes } } } } */
/* { dg-final { cleanup-tree-dump "vect" } } */
===================================================================
@@ -46,5 +46,6 @@ int main (void)
If/when the aliasing problems are resolved, unalignment may
prevent vectorization on some targets. */
/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" { xfail *-*-* } } } */
-/* { dg-final { scan-tree-dump-times "can't determine dependence between" 1 "vect" } } */
+/* { dg-final { scan-tree-dump-times "can't determine dependence" 1 "vect" { xfail vect_multiple_sizes } } } */
+/* { dg-final { scan-tree-dump-times "can't determine dependence" 2 "vect" { target vect_multiple_sizes } } } */
/* { dg-final { cleanup-tree-dump "vect" } } */
===================================================================
@@ -53,6 +53,7 @@ int main (void)
}
/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 0 "vect" } } */
-/* { dg-final { scan-tree-dump-times "possible dependence between data-refs" 1 "vect" } } */
+/* { dg-final { scan-tree-dump-times "possible dependence between data-refs" 1 "vect" { xfail vect_multiple_sizes } } } */
+/* { dg-final { scan-tree-dump-times "possible dependence between data-refs" 2 "vect" { target vect_multiple_sizes } } } */
/* { dg-final { cleanup-tree-dump "vect" } } */
===================================================================
@@ -1,5 +1,4 @@
/* { dg-require-effective-target vect_float } */
-/* { dg-add-options quad_vectors } */
#include <stdarg.h>
#include <signal.h>
===================================================================
@@ -1,4 +1,5 @@
/* { dg-require-effective-target vect_int } */
+/* { dg-add-options double_vectors } */
#include <stdarg.h>
#include "tree-vect.h"
===================================================================
@@ -1,4 +1,5 @@
/* { dg-require-effective-target vect_float } */
+/* { dg-add-options double_vectors } */
#include <stdarg.h>
#include "tree-vect.h"
===================================================================
@@ -1,5 +1,4 @@
/* { dg-require-effective-target vect_int } */
-/* { dg-add-options quad_vectors } */
#include <stdarg.h>
#include "tree-vect.h"
===================================================================
@@ -1,4 +1,5 @@
/* { dg-require-effective-target vect_float } */
+/* { dg-add-options double_vectors } */
#include <stdarg.h>
#include "tree-vect.h"
===================================================================
@@ -1,4 +1,5 @@
/* { dg-require-effective-target vect_float } */
+/* { dg-add-options double_vectors } */
#include <stdarg.h>
#include "tree-vect.h"
===================================================================
@@ -1,5 +1,4 @@
/* { dg-require-effective-target vect_int } */
-/* { dg-add-options quad_vectors } */
#include <stdarg.h>
#include "tree-vect.h"
===================================================================
@@ -20,5 +20,6 @@ foo (){
}
/* { dg-final { scan-tree-dump-times "OUTER LOOP VECTORIZED" 1 "vect" { xfail *-*-* } } } */
-/* { dg-final { scan-tree-dump-times "strided access in outer loop" 1 "vect" } } */
+/* { dg-final { scan-tree-dump-times "strided access in outer loop" 1 "vect" { xfail vect_multiple_sizes } } } */
+/* { dg-final { scan-tree-dump-times "strided access in outer loop" 2 "vect" { target vect_multiple_sizes } } } */
/* { dg-final { cleanup-tree-dump "vect" } } */
===================================================================
@@ -1,5 +1,4 @@
/* { dg-require-effective-target vect_int } */
-/* { dg-add-options quad_vectors } */
#include <stdarg.h>
#include "tree-vect.h"
===================================================================
@@ -1,4 +1,5 @@
/* { dg-require-effective-target vect_float } */
+/* { dg-add-options double_vectors } */
#include <stdarg.h>
#include "tree-vect.h"
===================================================================
@@ -22,5 +22,6 @@ foo (){
}
/* { dg-final { scan-tree-dump-times "OUTER LOOP VECTORIZED" 1 "vect" { xfail *-*-* } } } */
-/* { dg-final { scan-tree-dump-times "strided access in outer loop" 1 "vect" } } */
+/* { dg-final { scan-tree-dump-times "strided access in outer loop" 1 "vect" { xfail vect_multiple_sizes } } } */
+/* { dg-final { scan-tree-dump-times "strided access in outer loop" 2 "vect" { target vect_multiple_sizes } } } */
/* { dg-final { cleanup-tree-dump "vect" } } */
===================================================================
@@ -37,5 +37,6 @@ int main (void)
return 0;
}
-/* { dg-final { scan-tree-dump-times "strided access in outer loop." 1 "vect" } } */
+/* { dg-final { scan-tree-dump-times "strided access in outer loop" 1 "vect" { xfail vect_multiple_sizes } } } */
+/* { dg-final { scan-tree-dump-times "strided access in outer loop" 2 "vect" { target vect_multiple_sizes } } } */
/* { dg-final { cleanup-tree-dump "vect" } } */
===================================================================
@@ -49,5 +49,6 @@ int main (void)
}
/* { dg-final { scan-tree-dump-times "OUTER LOOP VECTORIZED" 1 "vect" { xfail vect_no_align } } } */
-/* { dg-final { scan-tree-dump-times "step doesn't divide the vector-size" 2 "vect" } } */
+/* { dg-final { scan-tree-dump-times "step doesn't divide the vector-size" 2 "vect" { xfail vect_multiple_sizes } } } */
+/* { dg-final { scan-tree-dump-times "step doesn't divide the vector-size" 4 "vect" { target vect_multiple_sizes } } } */
/* { dg-final { cleanup-tree-dump "vect" } } */
===================================================================
@@ -58,5 +58,6 @@ int main (void)
If/when the aliasing problems are resolved, unalignment may
prevent vectorization on some targets. */
/* { dg-final { scan-tree-dump-times "vectorized 2 loops" 1 "vect" { xfail *-*-* } } } */
-/* { dg-final { scan-tree-dump-times "can't determine dependence between" 2 "vect" } } */
+/* { dg-final { scan-tree-dump-times "can't determine dependence" 2 "vect" { xfail vect_multiple_sizes } } } */
+/* { dg-final { scan-tree-dump-times "can't determine dependence" 4 "vect" { target vect_multiple_sizes } } } */
/* { dg-final { cleanup-tree-dump "vect" } } */
===================================================================
@@ -1,5 +1,4 @@
/* { dg-require-effective-target vect_int } */
-/* { dg-add-options quad_vectors } */
#include <stdarg.h>
#include "tree-vect.h"
===================================================================
@@ -49,5 +49,6 @@ int main (void)
}
/* { dg-final { scan-tree-dump-times "OUTER LOOP VECTORIZED" 1 "vect" { xfail *-*-* } } } */
-/* { dg-final { scan-tree-dump-times "strided access in outer loop" 2 "vect" } } */
+/* { dg-final { scan-tree-dump-times "strided access in outer loop" 2 "vect" { xfail vect_multiple_sizes } } } */
+/* { dg-final { scan-tree-dump-times "strided access in outer loop" 4 "vect" { target vect_multiple_sizes } } } */
/* { dg-final { cleanup-tree-dump "vect" } } */
===================================================================
@@ -0,0 +1,38 @@
+/* { dg-require-effective-target vect_float } */
+
+#include <stdarg.h>
+#include "tree-vect.h"
+
+#define N 16
+#define DIFF 240
+
+__attribute__ ((noinline))
+int main1 ()
+{
+ int i;
+ float b[N] = {0,3,6,9,12,15,18,21,24,27,30,33,36,39,42,45};
+ float c[N] = {0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15};
+ float diff;
+
+ diff = 0;
+ for (i = 0; i < N; i++) {
+ diff += (b[i] - c[i]);
+ }
+
+ /* check results: */
+ if (diff != DIFF)
+ abort ();
+
+ return 0;
+}
+
+int main (void)
+{
+ check_vect ();
+
+ return main1 ();
+}
+
+/* Requires fast-math. */
+/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" { xfail *-*-* } } } */
+/* { dg-final { cleanup-tree-dump "vect" } } */
===================================================================
@@ -45,6 +45,7 @@ int main (void)
}
/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 0 "vect" } } */
-/* { dg-final { scan-tree-dump-times "can't determine dependence" 1 "vect" } } */
+/* { dg-final { scan-tree-dump-times "can't determine dependence" 1 "vect" { xfail vect_multiple_sizes } } } */
+/* { dg-final { scan-tree-dump-times "can't determine dependence" 2 "vect" { target vect_multiple_sizes } } } */
/* { dg-final { cleanup-tree-dump "vect" } } */
===================================================================
@@ -53,6 +53,7 @@ int main (void)
}
/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 0 "vect" } } */
-/* { dg-final { scan-tree-dump-times "possible dependence between data-refs" 1 "vect" } } */
+/* { dg-final { scan-tree-dump-times "possible dependence between data-refs" 1 "vect" { xfail vect_multiple_sizes } } } */
+/* { dg-final { scan-tree-dump-times "possible dependence between data-refs" 2 "vect" { target vect_multiple_sizes } } } */
/* { dg-final { cleanup-tree-dump "vect" } } */
===================================================================
@@ -1,4 +1,5 @@
/* { dg-require-effective-target vect_int } */
+/* { dg-add-options double_vectors } */
#include <stdarg.h>
#include "tree-vect.h"
===================================================================
@@ -1,5 +1,4 @@
/* { dg-require-effective-target vect_int } */
-/* { dg-add-options quad_vectors } */
#include <stdarg.h>
#include "tree-vect.h"
===================================================================
@@ -58,7 +58,8 @@ int main (void)
}
/* { dg-final { scan-tree-dump-times "vect_recog_dot_prod_pattern: detected" 1 "vect" { xfail *-*-* } } } */
-/* { dg-final { scan-tree-dump-times "vect_recog_widen_mult_pattern: detected" 1 "vect" } } */
+/* { dg-final { scan-tree-dump-times "vect_recog_widen_mult_pattern: detected" 1 "vect" { xfail vect_multiple_sizes } } } */
+/* { dg-final { scan-tree-dump-times "vect_recog_widen_mult_pattern: detected" 2 "vect" { target vect_multiple_sizes } } } */
/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" { xfail *-*-* } } } */
===================================================================
@@ -22,5 +22,6 @@ foo (){
}
/* { dg-final { scan-tree-dump-times "OUTER LOOP VECTORIZED" 1 "vect" { xfail *-*-* } } } */
-/* { dg-final { scan-tree-dump-times "strided access in outer loop" 1 "vect" } } */
+/* { dg-final { scan-tree-dump-times "strided access in outer loop" 1 "vect" { xfail vect_multiple_sizes } } } */
+/* { dg-final { scan-tree-dump-times "strided access in outer loop" 2 "vect" { target vect_multiple_sizes } } } */
/* { dg-final { cleanup-tree-dump "vect" } } */
===================================================================
@@ -1,5 +1,4 @@
/* { dg-require-effective-target vect_int } */
-/* { dg-add-options quad_vectors } */
#include <stdarg.h>
#include "tree-vect.h"
@@ -79,7 +78,7 @@ main1 ()
}
/* check results: */
- for (i = 0; i < N/2; i++)
+ for (i = 0; i < N/4; i++)
{
if (out[i*16] != in[i*16]
|| out[i*16 + 1] != in[i*16 + 1]
===================================================================
@@ -1,5 +1,4 @@
/* { dg-require-effective-target vect_int } */
-/* { dg-add-options quad_vectors } */
#include <stdarg.h>
#include "tree-vect.h"
===================================================================
@@ -102,6 +102,12 @@ lappend DEFAULT_VECTCFLAGS "-ffast-math"
dg-runtest [lsort [glob -nocomplain $srcdir/$subdir/fast-math-*.\[cS\]]] \
"" $DEFAULT_VECTCFLAGS
+# -fno-fast-math tests
+set DEFAULT_VECTCFLAGS $SAVED_DEFAULT_VECTCFLAGS
+lappend DEFAULT_VECTCFLAGS "-fno-fast-math"
+dg-runtest [lsort [glob -nocomplain $srcdir/$subdir/no-fast-math-*.\[cS\]]] \
+ "" $DEFAULT_VECTCFLAGS
+
# -fno-math-errno tests
set DEFAULT_VECTCFLAGS $SAVED_DEFAULT_VECTCFLAGS
lappend DEFAULT_VECTCFLAGS "-fno-math-errno"
===================================================================
@@ -1,4 +1,5 @@
/* { dg-require-effective-target vect_int } */
+/* { dg-add-options double_vectors } */
#include <stdarg.h>
#include "tree-vect.h"
===================================================================
@@ -1,4 +1,5 @@
/* { dg-require-effective-target vect_float } */
+/* { dg-add-options double_vectors } */
#include <stdarg.h>
#include "tree-vect.h"
===================================================================
@@ -64,6 +64,7 @@ int main (void)
}
/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 0 "vect" } } */
-/* { dg-final { scan-tree-dump-times "possible dependence between data-refs" 1 "vect" } } */
+/* { dg-final { scan-tree-dump-times "possible dependence between data-refs" 1 "vect" { xfail vect_multiple_sizes } } } */
+/* { dg-final { scan-tree-dump-times "possible dependence between data-refs" 2 "vect" { target vect_multiple_sizes } } } */
/* { dg-final { cleanup-tree-dump "vect" } } */
===================================================================
@@ -1,5 +1,4 @@
/* { dg-require-effective-target vect_int } */
-/* { dg-add-options quad_vectors } */
#include <stdarg.h>
#include "tree-vect.h"
===================================================================
@@ -22297,17 +22297,16 @@ arm_preferred_simd_mode (enum machine_mode mode)
switch (mode)
{
case SFmode:
- return TARGET_NEON_VECTORIZE_QUAD ? V4SFmode : V2SFmode;
+ return (PREFERRED_VECTOR_SIZE == 64) ? V2SFmode : V4SFmode;
case SImode:
- return TARGET_NEON_VECTORIZE_QUAD ? V4SImode : V2SImode;
+ return (PREFERRED_VECTOR_SIZE == 64) ? V2SImode : V4SImode;
case HImode:
- return TARGET_NEON_VECTORIZE_QUAD ? V8HImode : V4HImode;
+ return (PREFERRED_VECTOR_SIZE == 64) ? V4HImode : V8HImode;
case QImode:
- return TARGET_NEON_VECTORIZE_QUAD ? V16QImode : V8QImode;
+ return (PREFERRED_VECTOR_SIZE == 64)? V8QImode : V16QImode;
case DImode:
- if (TARGET_NEON_VECTORIZE_QUAD)
- return V2DImode;
- break;
+ if (PREFERRED_VECTOR_SIZE != 64)
+ return V2DImode;
default:;
}
@@ -23535,7 +23534,7 @@ arm_expand_sync (enum machine_mode mode,
static unsigned int
arm_autovectorize_vector_sizes (void)
{
- return TARGET_NEON_VECTORIZE_QUAD ? 16 | 8 : 0;
+ return (PREFERRED_VECTOR_SIZE != 64) ? 16 | 8 : 0;
}
static bool
===================================================================
@@ -158,7 +158,7 @@ Target Report RejectNegative Mask(LITTLE_WORDS)
Assume big endian bytes, little endian words
mvectorize-with-neon-quad
-Target Report Mask(NEON_VECTORIZE_QUAD)
+Target Report RejectNegative Mask(NEON_VECTORIZE_QUAD)
Use Neon quad-word (rather than double-word) registers for vectorization
mword-relocations
===================================================================
@@ -880,6 +880,12 @@ DEFPARAM (PARAM_MAX_STORES_TO_SINK,
2, 0, 0)
+/* Preferred vector size in bits. Invalid values are ignored. */
+DEFPARAM (PARAM_PREFERRED_VECTOR_SIZE,
+ "preferred-vector-size",
+ "Preferred vector size in bits",
+ 128, 0, 0)
+
/*
Local variables:
mode:c