aarch64: Improve scalar mode popcount expansion by using SVE [PR113860]

This is similar to the recent improvements to the Advanced SIMD popcount
expansion by using SVE. We can utilize SVE to generate more efficient code for
scalar mode popcount too.

Changes since v1:
* v2: Add a new VNx1BI mode and a new test case for V1DI.
* v3: Abandon VNx1BI changes and add a new variant of aarch64_ptrue_reg.

	PR target/113860

gcc/ChangeLog:

	* config/aarch64/aarch64-protos.h (aarch64_ptrue_reg): New function.
	* config/aarch64/aarch64-simd.md (popcount<mode>2): Update pattern to
	also support V1DI mode.
	* config/aarch64/aarch64.cc (aarch64_ptrue_reg): New function.
	* config/aarch64/aarch64.md (popcount<mode>2): Add TARGET_SVE support.
	* config/aarch64/iterators.md (VDQHSD_V1DI): New mode iterator.
	(SVE_VDQ_I): Add V1DI.
	(bitsize): Likewise.
	(VPRED): Likewise.
	(VEC_POP_MODE): New mode attribute.
	(vec_pop_mode): Likewise.

gcc/testsuite/ChangeLog:

	* gcc.target/aarch64/popcnt-sve.c: Update test.
	* gcc.target/aarch64/popcnt11.c: New test.
	* gcc.target/aarch64/popcnt12.c: New test.

Signed-off-by: Pengxuan Zheng <quic_pzheng@quicinc.com>

gcc/config/aarch64/aarch64-protos.h

@@ -917,6 +917,7 @@ rtx aarch64_expand_sve_dupq (rtx, machine_mode, rtx);
 void aarch64_expand_mov_immediate (rtx, rtx);
 rtx aarch64_stack_protect_canary_mem (machine_mode, rtx, aarch64_salt_type);
 rtx aarch64_ptrue_reg (machine_mode);
+rtx aarch64_ptrue_reg (machine_mode, unsigned int);
 rtx aarch64_pfalse_reg (machine_mode);
 bool aarch64_sve_same_pred_for_ptest_p (rtx *, rtx *);
 void aarch64_emit_sve_pred_move (rtx, rtx, rtx);

gcc/config/aarch64/aarch64-simd.md

@@ -3516,19 +3516,28 @@
 )
 
 (define_expand "popcount<mode>2"
-  [(set (match_operand:VDQHSD 0 "register_operand")
-	(popcount:VDQHSD (match_operand:VDQHSD 1 "register_operand")))]
+  [(set (match_operand:VDQHSD_V1DI 0 "register_operand")
+	(popcount:VDQHSD_V1DI
+	  (match_operand:VDQHSD_V1DI 1 "register_operand")))]
   "TARGET_SIMD"
   {
     if (TARGET_SVE)
       {
-	rtx p = aarch64_ptrue_reg (<VPRED>mode);
+	rtx p = aarch64_ptrue_reg (<VPRED>mode, <bitsize> == 64 ? 8 : 16);
 	emit_insn (gen_aarch64_pred_popcount<mode> (operands[0],
 						    p,
 						    operands[1]));
 	DONE;
       }
 
+    if (<MODE>mode == V1DImode)
+      {
+	rtx out = gen_reg_rtx (DImode);
+	emit_insn (gen_popcountdi2 (out, gen_lowpart (DImode, operands[1])));
+	emit_move_insn (operands[0], gen_lowpart (<MODE>mode, out));
+	DONE;
+      }
+
     /* Generate a byte popcount.  */
     machine_mode mode = <bitsize> == 64 ? V8QImode : V16QImode;
     machine_mode mode2 = <bitsize> == 64 ? V2SImode : V4SImode;

gcc/config/aarch64/aarch64.cc

@@ -3630,6 +3630,27 @@ aarch64_ptrue_reg (machine_mode mode)
   return gen_lowpart (mode, reg);
 }
 
+/* Return an all-true (restricted to the leading VL bits) predicate register of
+   mode MODE.  */
+
+rtx
+aarch64_ptrue_reg (machine_mode mode, unsigned int vl)
+{
+  gcc_assert (aarch64_sve_pred_mode_p (mode));
+
+  rtx_vector_builder builder (VNx16BImode, vl, 2);
+
+  for (int i = 0; i < vl; i++)
+    builder.quick_push (CONST1_RTX (BImode));
+
+  for (int i = 0; i < vl; i++)
+    builder.quick_push (CONST0_RTX (BImode));
+
+  rtx const_vec = builder.build ();
+  rtx reg = force_reg (VNx16BImode, const_vec);
+  return gen_lowpart (mode, reg);
+}
+
 /* Return an all-false predicate register of mode MODE.  */
 
 rtx

gcc/config/aarch64/aarch64.md

@@ -5345,6 +5345,15 @@
 	(popcount:ALLI (match_operand:ALLI 1 "register_operand")))]
   "TARGET_CSSC ? GET_MODE_BITSIZE (<MODE>mode) >= 32 : TARGET_SIMD"
 {
+  if (!TARGET_CSSC && TARGET_SVE && <MODE>mode != QImode)
+    {
+      rtx tmp = gen_reg_rtx (<VEC_POP_MODE>mode);
+      rtx op1 = gen_lowpart (<VEC_POP_MODE>mode, operands[1]);
+      emit_insn (gen_popcount<vec_pop_mode>2 (tmp, op1));
+      emit_move_insn (operands[0], gen_lowpart (<MODE>mode, tmp));
+      DONE;
+    }
+
   if (!TARGET_CSSC)
     {
       rtx v = gen_reg_rtx (V8QImode);

gcc/config/aarch64/iterators.md

@@ -290,6 +290,8 @@
 ;; Advanced SIMD modes for H, S and D types.
 (define_mode_iterator VDQHSD [V4HI V8HI V2SI V4SI V2DI])
 
+(define_mode_iterator VDQHSD_V1DI [VDQHSD V1DI])
+
 ;; Advanced SIMD and scalar integer modes for H and S.
 (define_mode_iterator VSDQ_HSI [V4HI V8HI V2SI V4SI HI SI])
 
@@ -559,7 +561,7 @@
 (define_mode_iterator SVE_I_SIMD_DI [SVE_I V2DI])
 
 ;; All SVE and Advanced SIMD integer vector modes.
-(define_mode_iterator SVE_VDQ_I [SVE_I VDQ_I])
+(define_mode_iterator SVE_VDQ_I [SVE_I VDQ_I V1DI])
 
 ;; SVE integer vector modes whose elements are 16 bits or wider.
 (define_mode_iterator SVE_HSDI [VNx8HI VNx4HI VNx2HI
@@ -1235,7 +1237,7 @@
 (define_mode_attr bitsize [(V8QI "64") (V16QI "128")
 			   (V4HI "64") (V8HI "128")
 			   (V2SI "64") (V4SI "128")
-				       (V2DI "128")])
+			   (V1DI "64") (V2DI "128")])
 
 ;; Map a floating point or integer mode to the appropriate register name prefix
 (define_mode_attr s [(HF "h") (SF "s") (DF "d") (SI "s") (DI "d")])
@@ -2297,7 +2299,7 @@
 			 (VNx8DI "VNx2BI") (VNx8DF "VNx2BI")
 			 (V8QI "VNx8BI") (V16QI "VNx16BI")
 			 (V4HI "VNx4BI") (V8HI "VNx8BI") (V2SI "VNx2BI")
-			 (V4SI "VNx4BI") (V2DI "VNx2BI")])
+			 (V4SI "VNx4BI") (V2DI "VNx2BI") (V1DI "VNx2BI")])
 
 ;; ...and again in lower case.
 (define_mode_attr vpred [(VNx16QI "vnx16bi") (VNx8QI "vnx8bi")
@@ -2331,6 +2333,14 @@
 			   (VNx4SI "VNx8SI") (VNx4SF "VNx8SF")
 			   (VNx2DI "VNx4DI") (VNx2DF "VNx4DF")])
 
+;; The Advanced SIMD modes of popcount corresponding to scalar modes.
+(define_mode_attr VEC_POP_MODE [(QI "V8QI") (HI "V4HI")
+				(SI "V2SI") (DI "V1DI")])
+
+;; ...and again in lower case.
+(define_mode_attr vec_pop_mode [(QI "v8qi") (HI "v4hi")
+				(SI "v2si") (DI "v1di")])
+
 ;; On AArch64 the By element instruction doesn't have a 2S variant.
 ;; However because the instruction always selects a pair of values
 ;; The normal 3SAME instruction can be used here instead.

gcc/testsuite/gcc.target/aarch64/popcnt-sve.c

@@ -4,7 +4,7 @@
 
 /*
 ** f_v4hi:
-**	ptrue	(p[0-7]).b, all
+**	ptrue	(p[0-7]).b, vl8
 **	ldr	d([0-9]+), \[x0\]
 **	cnt	z\2.h, \1/m, z\2.h
 **	str	d\2, \[x1\]
@@ -21,7 +21,7 @@ f_v4hi (unsigned short *__restrict b, unsigned short *__restrict d)
 
 /*
 ** f_v8hi:
-**	ptrue	(p[0-7]).b, all
+**	ptrue	(p[0-7]).b, vl16
 **	ldr	q([0-9]+), \[x0\]
 **	cnt	z\2.h, \1/m, z\2.h
 **	str	q\2, \[x1\]
@@ -42,7 +42,7 @@ f_v8hi (unsigned short *__restrict b, unsigned short *__restrict d)
 
 /*
 ** f_v2si:
-**	ptrue	(p[0-7]).b, all
+**	ptrue	(p[0-7]).b, vl8
 **	ldr	d([0-9]+), \[x0\]
 **	cnt	z\2.s, \1/m, z\2.s
 **	str	d\2, \[x1\]
@@ -57,7 +57,7 @@ f_v2si (unsigned int *__restrict b, unsigned int *__restrict d)
 
 /*
 ** f_v4si:
-**	ptrue	(p[0-7]).b, all
+**	ptrue	(p[0-7]).b, vl16
 **	ldr	q([0-9]+), \[x0\]
 **	cnt	z\2.s, \1/m, z\2.s
 **	str	q\2, \[x1\]
@@ -74,7 +74,7 @@ f_v4si (unsigned int *__restrict b, unsigned int *__restrict d)
 
 /*
 ** f_v2di:
-**	ptrue	(p[0-7]).b, all
+**	ptrue	(p[0-7]).b, vl16
 **	ldr	q([0-9]+), \[x0\]
 **	cnt	z\2.d, \1/m, z\2.d
 **	str	q\2, \[x1\]

gcc/testsuite/gcc.target/aarch64/popcnt11.c

@@ -0,0 +1,58 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -march=armv8.2-a+sve" } */
+/* { dg-final { check-function-bodies "**" "" "" } } */
+
+/*
+** f_qi:
+**	ldr	b([0-9]+), \[x0\]
+**	cnt	v\1.8b, v\1.8b
+**	smov	w0, v\1.b\[0\]
+**	ret
+*/
+unsigned
+f_qi (unsigned char *a)
+{
+  return __builtin_popcountg (a[0]);
+}
+
+/*
+** f_hi:
+**	ldr	h([0-9]+), \[x0\]
+**	ptrue	(p[0-7]).b, vl8
+**	cnt	z\1.h, \2/m, z\1.h
+**	smov	w0, v\1.h\[0\]
+**	ret
+*/
+unsigned
+f_hi (unsigned short *a)
+{
+  return __builtin_popcountg (a[0]);
+}
+
+/*
+** f_si:
+**	ldr	s([0-9]+), \[x0\]
+**	ptrue	(p[0-7]).b, vl8
+**	cnt	z\1.s, \2/m, z\1.s
+**	umov	x0, v\1.d\[0\]
+**	ret
+*/
+unsigned
+f_si (unsigned int *a)
+{
+  return __builtin_popcountg (a[0]);
+}
+
+/*
+** f_di:
+**	ldr	d([0-9]+), \[x0\]
+**	ptrue	(p[0-7])\.b, vl8
+**	cnt	z\1\.d, \2/m, z\1\.d
+**	fmov	x0, d\1
+**	ret
+*/
+unsigned
+f_di (unsigned long *a)
+{
+  return __builtin_popcountg (a[0]);
+}

gcc/testsuite/gcc.target/aarch64/popcnt12.c

@@ -0,0 +1,20 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -fgimple" } */
+/* { dg-final { check-function-bodies "**" "" "" } } */
+
+#pragma GCC target "+nosve"
+
+/*
+** foo:
+**	cnt	(v[0-9]+\.8b), v0\.8b
+**	addv	b0, \1
+**	ret
+*/
+__Uint64x1_t __GIMPLE
+foo (__Uint64x1_t x)
+{
+  __Uint64x1_t z;
+
+  z = .POPCOUNT (x);
+  return z;
+}