[AArch64] Use SVE MOV /M of scalars

This patch uses MOV /M to optimise selects between a duplicated
scalar variable and a vector.

2019-08-14  Richard Sandiford  <richard.sandiford@arm.com>
	    Kugan Vivekanandarajah  <kugan.vivekanandarajah@linaro.org>

gcc/
	* config/aarch64/aarch64-sve.md (*aarch64_sel_dup<mode>): New pattern.

gcc/testsuite/
	* g++.target/aarch64/sve/dup_sel_1.C: New test.
	* g++.target/aarch64/sve/dup_sel_2.C: Likewise.
	* g++.target/aarch64/sve/dup_sel_3.C: Likewise.
	* g++.target/aarch64/sve/dup_sel_4.C: Likewise.
	* g++.target/aarch64/sve/dup_sel_5.C: Likewise.
	* g++.target/aarch64/sve/dup_sel_6.C: Likewise.

Co-Authored-By: Kugan Vivekanandarajah <kuganv@linaro.org>

From-SVN: r274442

9 files changed, 156 insertions, 0 deletions

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index 777f1c8..73156f1 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,6 +1,11 @@
 2019-08-14  Richard Sandiford  <richard.sandiford@arm.com>
 	    Kugan Vivekanandarajah  <kugan.vivekanandarajah@linaro.org>
 
+	* config/aarch64/aarch64-sve.md (*aarch64_sel_dup<mode>): New pattern.
+
+2019-08-14  Richard Sandiford  <richard.sandiford@arm.com>
+	    Kugan Vivekanandarajah  <kugan.vivekanandarajah@linaro.org>
+
 	* config/aarch64/aarch64.c (aarch64_bit_representation): New function.
 	(aarch64_print_vector_float_operand): Also handle 8-bit floats.
 	(aarch64_print_operand): Add support for %I.
diff --git a/gcc/config/aarch64/aarch64-sve.md b/gcc/config/aarch64/aarch64-sve.md
index 1bdfc47..e12b5d9 100644
--- a/gcc/config/aarch64/aarch64-sve.md
+++ b/gcc/config/aarch64/aarch64-sve.md
@@ -3070,6 +3070,29 @@
   [(set_attr "movprfx" "*,*,*,*,yes,yes,yes")]
 )
 
+;; Optimize selects between a duplicated scalar variable and another vector,
+;; the latter of which can be a zero constant or a variable.  Treat duplicates
+;; of GPRs as being more expensive than duplicates of FPRs, since they
+;; involve a cross-file move.
+(define_insn "*aarch64_sel_dup<mode>"
+  [(set (match_operand:SVE_ALL 0 "register_operand" "=?w, w, ??w, ?&w, ??&w, ?&w")
+	(unspec:SVE_ALL
+	  [(match_operand:<VPRED> 3 "register_operand" "Upa, Upa, Upl, Upl, Upl, Upl")
+	   (vec_duplicate:SVE_ALL
+	     (match_operand:<VEL> 1 "register_operand" "r, w, r, w, r, w"))
+	   (match_operand:SVE_ALL 2 "aarch64_simd_reg_or_zero" "0, 0, Dz, Dz, w, w")]
+	  UNSPEC_SEL))]
+  "TARGET_SVE"
+  "@
+   mov\t%0.<Vetype>, %3/m, %<vwcore>1
+   mov\t%0.<Vetype>, %3/m, %<Vetype>1
+   movprfx\t%0.<Vetype>, %3/z, %0.<Vetype>\;mov\t%0.<Vetype>, %3/m, %<vwcore>1
+   movprfx\t%0.<Vetype>, %3/z, %0.<Vetype>\;mov\t%0.<Vetype>, %3/m, %<Vetype>1
+   movprfx\t%0, %2\;mov\t%0.<Vetype>, %3/m, %<vwcore>1
+   movprfx\t%0, %2\;mov\t%0.<Vetype>, %3/m, %<Vetype>1"
+  [(set_attr "movprfx" "*,*,yes,yes,yes,yes")]
+)
+
 ;; -------------------------------------------------------------------------
 ;; ---- [INT,FP] Compare and select
 ;; -------------------------------------------------------------------------
diff --git a/gcc/testsuite/ChangeLog b/gcc/testsuite/ChangeLog
index a2507f7..de09124 100644
--- a/gcc/testsuite/ChangeLog
+++ b/gcc/testsuite/ChangeLog
@@ -1,6 +1,16 @@
 2019-08-14  Richard Sandiford  <richard.sandiford@arm.com>
 	    Kugan Vivekanandarajah  <kugan.vivekanandarajah@linaro.org>
 
+	* g++.target/aarch64/sve/dup_sel_1.C: New test.
+	* g++.target/aarch64/sve/dup_sel_2.C: Likewise.
+	* g++.target/aarch64/sve/dup_sel_3.C: Likewise.
+	* g++.target/aarch64/sve/dup_sel_4.C: Likewise.
+	* g++.target/aarch64/sve/dup_sel_5.C: Likewise.
+	* g++.target/aarch64/sve/dup_sel_6.C: Likewise.
+
+2019-08-14  Richard Sandiford  <richard.sandiford@arm.com>
+	    Kugan Vivekanandarajah  <kugan.vivekanandarajah@linaro.org>
+
 	* gcc.target/aarch64/sve/vcond_18.c: New test.
 	* gcc.target/aarch64/sve/vcond_18_run.c: Likewise.
 	* gcc.target/aarch64/sve/vcond_19.c: Likewise.
diff --git a/gcc/testsuite/g++.target/aarch64/sve/dup_sel_1.C b/gcc/testsuite/g++.target/aarch64/sve/dup_sel_1.C
new file mode 100644
index 0000000..a59862c
--- /dev/null
+++ b/gcc/testsuite/g++.target/aarch64/sve/dup_sel_1.C
@@ -0,0 +1,21 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -msve-vector-bits=256" } */
+
+#include <stdint.h>
+
+typedef int32_t vnx4si __attribute__((vector_size(32)));
+
+void
+foo (int32_t val)
+{
+  register vnx4si x asm ("z0");
+  register vnx4si y asm ("z0");
+  asm volatile ("" : "=w" (y));
+  val += 1;
+  vnx4si z = { val, val, val, val, val, val, val, val };
+  x = (vnx4si) { -1, 0, 0, -1, 0, -1, 0, -1 } ? z : y;
+  asm volatile ("" :: "w" (x));
+}
+
+/* { dg-final { scan-assembler {\tmov\tz0\.s, p[0-7]/m, w[0-9]+\n} } } */
+/* { dg-final { scan-assembler-not {\tmovprfx\t} } } */
diff --git a/gcc/testsuite/g++.target/aarch64/sve/dup_sel_2.C b/gcc/testsuite/g++.target/aarch64/sve/dup_sel_2.C
new file mode 100644
index 0000000..47aad2d
--- /dev/null
+++ b/gcc/testsuite/g++.target/aarch64/sve/dup_sel_2.C
@@ -0,0 +1,20 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -msve-vector-bits=256" } */
+
+#include <stdint.h>
+
+typedef int32_t vnx4si __attribute__((vector_size(32)));
+
+void
+foo (int32_t val)
+{
+  register vnx4si x asm ("z0");
+  register vnx4si y asm ("z1");
+  asm volatile ("" : "=w" (y));
+  val += 1;
+  vnx4si z = { val, val, val, val, val, val, val, val };
+  x = (vnx4si) { -1, 0, 0, -1, 0, -1, 0, -1 } ? z : y;
+  asm volatile ("" :: "w" (x));
+}
+
+/* { dg-final { scan-assembler {\tmovprfx\tz0, z1\n\tmov\tz0\.s, p[0-7]/m, w[0-9]+\n} } } */
diff --git a/gcc/testsuite/g++.target/aarch64/sve/dup_sel_3.C b/gcc/testsuite/g++.target/aarch64/sve/dup_sel_3.C
new file mode 100644
index 0000000..e8ec6f8
--- /dev/null
+++ b/gcc/testsuite/g++.target/aarch64/sve/dup_sel_3.C
@@ -0,0 +1,21 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -msve-vector-bits=256" } */
+
+#include <stdint.h>
+
+typedef int32_t vnx4si __attribute__((vector_size(32)));
+typedef float vnx4sf __attribute__((vector_size(32)));
+
+void
+foo (float val)
+{
+  register vnx4sf x asm ("z0");
+  register vnx4sf y asm ("z0");
+  asm volatile ("" : "=w" (y));
+  vnx4sf z = { val, val, val, val, val, val, val, val };
+  x = (vnx4si) { -1, 0, 0, -1, 0, -1, 0, -1 } ? z : y;
+  asm volatile ("" :: "w" (x));
+}
+
+/* { dg-final { scan-assembler {\tmov\tz0\.s, p[0-7]/m, s[0-9]+\n} } } */
+/* { dg-final { scan-assembler-not {\tmovprfx\t} } } */
diff --git a/gcc/testsuite/g++.target/aarch64/sve/dup_sel_4.C b/gcc/testsuite/g++.target/aarch64/sve/dup_sel_4.C
new file mode 100644
index 0000000..32ca594
--- /dev/null
+++ b/gcc/testsuite/g++.target/aarch64/sve/dup_sel_4.C
@@ -0,0 +1,20 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -msve-vector-bits=256" } */
+
+#include <stdint.h>
+
+typedef int32_t vnx4si __attribute__((vector_size(32)));
+typedef float vnx4sf __attribute__((vector_size(32)));
+
+void
+foo (float val)
+{
+  register vnx4sf x asm ("z0");
+  register vnx4sf y asm ("z1");
+  asm volatile ("" : "=w" (y));
+  vnx4sf z = { val, val, val, val, val, val, val, val };
+  x = (vnx4si) { -1, 0, 0, -1, 0, -1, 0, -1 } ? z : y;
+  asm volatile ("" :: "w" (x));
+}
+
+/* { dg-final { scan-assembler {\tmovprfx\tz0, z1\n\tmov\tz0\.s, p[0-7]/m, s[0-9]+\n} } } */
diff --git a/gcc/testsuite/g++.target/aarch64/sve/dup_sel_5.C b/gcc/testsuite/g++.target/aarch64/sve/dup_sel_5.C
new file mode 100644
index 0000000..2fb903a
--- /dev/null
+++ b/gcc/testsuite/g++.target/aarch64/sve/dup_sel_5.C
@@ -0,0 +1,18 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -msve-vector-bits=256" } */
+
+#include <stdint.h>
+
+typedef int32_t vnx4si __attribute__((vector_size(32)));
+
+void
+foo (int32_t val)
+{
+  register vnx4si x asm ("z0");
+  val += 1;
+  vnx4si y = { val, val, val, val, val, val, val, val };
+  x = (vnx4si) { -1, 0, 0, -1, 0, -1, 0, -1 } ? y : (vnx4si) { 0 };
+  asm volatile ("" :: "w" (x));
+}
+
+/* { dg-final { scan-assembler {\tmovprfx\tz0\.s, p[0-7]/z, z0\.s\n\tmov\tz0\.s, p[0-7]/m, w[0-9]+\n} } } */
diff --git a/gcc/testsuite/g++.target/aarch64/sve/dup_sel_6.C b/gcc/testsuite/g++.target/aarch64/sve/dup_sel_6.C
new file mode 100644
index 0000000..f2b0181
--- /dev/null
+++ b/gcc/testsuite/g++.target/aarch64/sve/dup_sel_6.C
@@ -0,0 +1,18 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -msve-vector-bits=256" } */
+
+#include <stdint.h>
+
+typedef int32_t vnx4si __attribute__((vector_size(32)));
+typedef float vnx4sf __attribute__((vector_size(32)));
+
+void
+foo (float val)
+{
+  register vnx4sf x asm ("z0");
+  vnx4sf y = { val, val, val, val, val, val, val, val };
+  x = (vnx4si) { -1, 0, 0, -1, 0, -1, 0, -1 } ? y : (vnx4sf) { 0 };
+  asm volatile ("" :: "w" (x));
+}
+
+/* { dg-final { scan-assembler {\tmovprfx\tz0\.s, p[0-7]/z, z0\.s\n\tmov\tz0\.s, p[0-7]/m, s[0-9]+\n} } } */