[PATCH] RISC-V: Support vi variant for vec_cmp
Juzhe-Zhong
juzhe.zhong@rivai.ai
Thu Jan 18 09:58:57 GMT 2024
While running various benchmarks, I notice we miss vi variant support for integer comparison.
That is, we can vectorize code into vadd.vi but we can't vectorize into vmseq.vi.
Consider this following case:
void
foo (int n, int **__restrict a)
{
int b;
int c;
int d;
for (b = 0; b < n; b++)
for (long e = 8; e > 0; e--)
a[b][e] = a[b][e] == 15;
}
Before this patch:
vsetivli zero,4,e32,m1,ta,ma
vmv.v.i v4,15
vmv.v.i v3,1
vmv.v.i v2,0
.L3:
ld a5,0(a1)
addi a4,a5,4
addi a5,a5,20
vle32.v v1,0(a5)
vle32.v v0,0(a4)
vmseq.vv v0,v0,v4
After this patch:
ld a5,0(a1)
addi a4,a5,4
addi a5,a5,20
vle32.v v1,0(a5)
vle32.v v0,0(a4)
vmseq.vi v0,v0,15
It's the missing feature caused by our some mistakes, support vi variant for vec_cmp like other patterns (add, sub, ..., etc).
Tested with no regression, ok for trunk ?
gcc/ChangeLog:
* config/riscv/autovec.md: Support vi variant.
gcc/testsuite/ChangeLog:
* gcc.target/riscv/rvv/autovec/cmp/cmp_vi-1.c: New test.
* gcc.target/riscv/rvv/autovec/cmp/cmp_vi-2.c: New test.
* gcc.target/riscv/rvv/autovec/cmp/cmp_vi-3.c: New test.
* gcc.target/riscv/rvv/autovec/cmp/cmp_vi-4.c: New test.
* gcc.target/riscv/rvv/autovec/cmp/cmp_vi-5.c: New test.
* gcc.target/riscv/rvv/autovec/cmp/cmp_vi-6.c: New test.
* gcc.target/riscv/rvv/autovec/cmp/cmp_vi-7.c: New test.
* gcc.target/riscv/rvv/autovec/cmp/cmp_vi-8.c: New test.
* gcc.target/riscv/rvv/autovec/cmp/cmp_vi-9.c: New test.
* gcc.target/riscv/rvv/autovec/cmp/macro.h: New test.
---
gcc/config/riscv/autovec.md | 4 +--
.../riscv/rvv/autovec/cmp/cmp_vi-1.c | 16 +++++++++++
.../riscv/rvv/autovec/cmp/cmp_vi-2.c | 16 +++++++++++
.../riscv/rvv/autovec/cmp/cmp_vi-3.c | 28 +++++++++++++++++++
.../riscv/rvv/autovec/cmp/cmp_vi-4.c | 28 +++++++++++++++++++
.../riscv/rvv/autovec/cmp/cmp_vi-5.c | 16 +++++++++++
.../riscv/rvv/autovec/cmp/cmp_vi-6.c | 16 +++++++++++
.../riscv/rvv/autovec/cmp/cmp_vi-7.c | 28 +++++++++++++++++++
.../riscv/rvv/autovec/cmp/cmp_vi-8.c | 28 +++++++++++++++++++
.../riscv/rvv/autovec/cmp/cmp_vi-9.c | 18 ++++++++++++
.../gcc.target/riscv/rvv/autovec/cmp/macro.h | 11 ++++++++
11 files changed, 207 insertions(+), 2 deletions(-)
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-1.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-2.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-3.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-4.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-5.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-6.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-7.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-8.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-9.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/macro.h
diff --git a/gcc/config/riscv/autovec.md b/gcc/config/riscv/autovec.md
index 706cd9717cb..5ec1c59bdd4 100644
--- a/gcc/config/riscv/autovec.md
+++ b/gcc/config/riscv/autovec.md
@@ -664,7 +664,7 @@
[(set (match_operand:<VM> 0 "register_operand")
(match_operator:<VM> 1 "comparison_operator"
[(match_operand:V_VLSI 2 "register_operand")
- (match_operand:V_VLSI 3 "register_operand")]))]
+ (match_operand:V_VLSI 3 "nonmemory_operand")]))]
"TARGET_VECTOR"
{
riscv_vector::expand_vec_cmp (operands[0], GET_CODE (operands[1]),
@@ -677,7 +677,7 @@
[(set (match_operand:<VM> 0 "register_operand")
(match_operator:<VM> 1 "comparison_operator"
[(match_operand:V_VLSI 2 "register_operand")
- (match_operand:V_VLSI 3 "register_operand")]))]
+ (match_operand:V_VLSI 3 "nonmemory_operand")]))]
"TARGET_VECTOR"
{
riscv_vector::expand_vec_cmp (operands[0], GET_CODE (operands[1]),
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-1.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-1.c
new file mode 100644
index 00000000000..10c232f77bd
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-1.c
@@ -0,0 +1,16 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gcv -mabi=lp64d -O3" } */
+
+#include "macro.h"
+
+CMP_VI (ne_char, char, n, !=, 15)
+CMP_VI (ne_short, short, n, !=, 15)
+CMP_VI (ne_int, int, n, !=, 15)
+CMP_VI (ne_long, long, n, !=, 15)
+CMP_VI (ne_unsigned_char, unsigned char, n, !=, 15)
+CMP_VI (ne_unsigned_short, unsigned short, n, !=, 15)
+CMP_VI (ne_unsigned_int, unsigned int, n, !=, 15)
+CMP_VI (ne_unsigned_long, unsigned long, n, !=, 15)
+
+/* { dg-final { scan-assembler-times {vmsne\.vi} 16 } } */
+/* { dg-final { scan-assembler-not {vmsne\.vv} } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-2.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-2.c
new file mode 100644
index 00000000000..92bea596cd8
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-2.c
@@ -0,0 +1,16 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gcv -mabi=lp64d -O3" } */
+
+#include "macro.h"
+
+CMP_VI (ne_char, char, n, !=, -16)
+CMP_VI (ne_short, short, n, !=, -16)
+CMP_VI (ne_int, int, n, !=, -16)
+CMP_VI (ne_long, long, n, !=, -16)
+CMP_VI (ne_unsigned_char, unsigned char, n, !=, -16)
+CMP_VI (ne_unsigned_short, unsigned short, n, !=, -16)
+CMP_VI (ne_unsigned_int, unsigned int, n, !=, -16)
+CMP_VI (ne_unsigned_long, unsigned long, n, !=, -16)
+
+/* { dg-final { scan-assembler-times {vmsne\.vi} 13 } } */
+/* { dg-final { scan-assembler-not {vmsne\.vv} } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-3.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-3.c
new file mode 100644
index 00000000000..c9003279b0c
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-3.c
@@ -0,0 +1,28 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gcv -mabi=lp64d -O3 --param=riscv-autovec-lmul=dynamic -fdump-tree-optimized-details" } */
+
+#include "macro.h"
+
+CMP_VI (ne_char, char, 4, !=, 15)
+CMP_VI (ne_short, short, 4, !=, 15)
+CMP_VI (ne_int, int, 4, !=, 15)
+CMP_VI (ne_long, long, 4, !=, 15)
+CMP_VI (ne_unsigned_char, unsigned char, 4, !=, 15)
+CMP_VI (ne_unsigned_short, unsigned short, 4, !=, 15)
+CMP_VI (ne_unsigned_int, unsigned int, 4, !=, 15)
+CMP_VI (ne_unsigned_long, unsigned long, 4, !=, 15)
+
+/* { dg-final { scan-assembler-times {vmsne\.vi} 32 } } */
+/* { dg-final { scan-assembler-not {vmsne\.vv} } } */
+/* { dg-final { scan-tree-dump-not "1,1" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "2,2" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "4,4" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "16,16" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "32,32" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "64,64" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "128,128" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "256,256" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "512,512" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "1024,1024" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "2048,2048" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "4096,4096" "optimized" } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-4.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-4.c
new file mode 100644
index 00000000000..544ff751522
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-4.c
@@ -0,0 +1,28 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gcv -mabi=lp64d -O3 --param=riscv-autovec-lmul=dynamic -fdump-tree-optimized-details" } */
+
+#include "macro.h"
+
+CMP_VI (ne_char, char, 4, !=, -16)
+CMP_VI (ne_short, short, 4, !=, -16)
+CMP_VI (ne_int, int, 4, !=, -16)
+CMP_VI (ne_long, long, 4, !=, -16)
+CMP_VI (ne_unsigned_char, unsigned char, 4, !=, -16)
+CMP_VI (ne_unsigned_short, unsigned short, 4, !=, -16)
+CMP_VI (ne_unsigned_int, unsigned int, 4, !=, -16)
+CMP_VI (ne_unsigned_long, unsigned long, 4, !=, -16)
+
+/* { dg-final { scan-assembler-times {vmsne\.vi} 20 } } */
+/* { dg-final { scan-assembler-not {vmsne\.vv} } } */
+/* { dg-final { scan-tree-dump-not "1,1" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "2,2" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "4,4" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "16,16" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "32,32" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "64,64" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "128,128" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "256,256" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "512,512" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "1024,1024" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "2048,2048" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "4096,4096" "optimized" } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-5.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-5.c
new file mode 100644
index 00000000000..b7a5a4397c1
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-5.c
@@ -0,0 +1,16 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gcv -mabi=lp64d -O3" } */
+
+#include "macro.h"
+
+CMP_VI (eq_char, char, n, ==, 15)
+CMP_VI (eq_short, short, n, ==, 15)
+CMP_VI (eq_int, int, n, ==, 15)
+CMP_VI (eq_long, long, n, ==, 15)
+CMP_VI (eq_unsigned_char, unsigned char, n, ==, 15)
+CMP_VI (eq_unsigned_short, unsigned short, n, ==, 15)
+CMP_VI (eq_unsigned_int, unsigned int, n, ==, 15)
+CMP_VI (eq_unsigned_long, unsigned long, n, ==, 15)
+
+/* { dg-final { scan-assembler-times {vmseq\.vi} 16 } } */
+/* { dg-final { scan-assembler-not {vmseq\.vv} } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-6.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-6.c
new file mode 100644
index 00000000000..f297ac80bbd
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-6.c
@@ -0,0 +1,16 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gcv -mabi=lp64d -O3" } */
+
+#include "macro.h"
+
+CMP_VI (eq_char, char, n, ==, -16)
+CMP_VI (eq_short, short, n, ==, -16)
+CMP_VI (eq_int, int, n, ==, -16)
+CMP_VI (eq_long, long, n, ==, -16)
+CMP_VI (eq_unsigned_char, unsigned char, n, ==, -16)
+CMP_VI (eq_unsigned_short, unsigned short, n, ==, -16)
+CMP_VI (eq_unsigned_int, unsigned int, n, ==, -16)
+CMP_VI (eq_unsigned_long, unsigned long, n, ==, -16)
+
+/* { dg-final { scan-assembler-times {vmseq\.vi} 13 } } */
+/* { dg-final { scan-assembler-not {vmseq\.vv} } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-7.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-7.c
new file mode 100644
index 00000000000..63ded00947d
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-7.c
@@ -0,0 +1,28 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gcv -mabi=lp64d -O3 --param=riscv-autovec-lmul=dynamic -fdump-tree-optimized-details" } */
+
+#include "macro.h"
+
+CMP_VI (eq_char, char, 4, ==, 15)
+CMP_VI (eq_short, short, 4, ==, 15)
+CMP_VI (eq_int, int, 4, ==, 15)
+CMP_VI (eq_long, long, 4, ==, 15)
+CMP_VI (eq_unsigned_char, unsigned char, 4, ==, 15)
+CMP_VI (eq_unsigned_short, unsigned short, 4, ==, 15)
+CMP_VI (eq_unsigned_int, unsigned int, 4, ==, 15)
+CMP_VI (eq_unsigned_long, unsigned long, 4, ==, 15)
+
+/* { dg-final { scan-assembler-times {vmseq\.vi} 32 } } */
+/* { dg-final { scan-assembler-not {vmseq\.vv} } } */
+/* { dg-final { scan-tree-dump-not "1,1" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "2,2" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "4,4" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "16,16" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "32,32" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "64,64" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "128,128" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "256,256" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "512,512" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "1024,1024" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "2048,2048" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "4096,4096" "optimized" } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-8.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-8.c
new file mode 100644
index 00000000000..f29b5f12c51
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-8.c
@@ -0,0 +1,28 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gcv -mabi=lp64d -O3 --param=riscv-autovec-lmul=dynamic -fdump-tree-optimized-details" } */
+
+#include "macro.h"
+
+CMP_VI (eq_char, char, 4, ==, -16)
+CMP_VI (eq_short, short, 4, ==, -16)
+CMP_VI (eq_int, int, 4, ==, -16)
+CMP_VI (eq_long, long, 4, ==, -16)
+CMP_VI (eq_unsigned_char, unsigned char, 4, ==, -16)
+CMP_VI (eq_unsigned_short, unsigned short, 4, ==, -16)
+CMP_VI (eq_unsigned_int, unsigned int, 4, ==, -16)
+CMP_VI (eq_unsigned_long, unsigned long, 4, ==, -16)
+
+/* { dg-final { scan-assembler-times {vmseq\.vi} 20 } } */
+/* { dg-final { scan-assembler-not {vmseq\.vv} } } */
+/* { dg-final { scan-tree-dump-not "1,1" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "2,2" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "4,4" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "16,16" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "32,32" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "64,64" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "128,128" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "256,256" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "512,512" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "1024,1024" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "2048,2048" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "4096,4096" "optimized" } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-9.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-9.c
new file mode 100644
index 00000000000..bfc1a68a1e4
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/cmp_vi-9.c
@@ -0,0 +1,18 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gcv -mabi=lp64d -O3" } */
+
+#include "macro.h"
+
+CMP_VI (le_char, char, n, <=, 15)
+CMP_VI (le_short, short, n, <=, 15)
+CMP_VI (le_int, int, n, <=, 15)
+CMP_VI (le_long, long, n, <=, 15)
+CMP_VI (le_unsigned_char, unsigned char, n, <=, 15)
+CMP_VI (le_unsigned_short, unsigned short, n, <=, 15)
+CMP_VI (le_unsigned_int, unsigned int, n, <=, 15)
+CMP_VI (le_unsigned_long, unsigned long, n, <=, 15)
+
+/* { dg-final { scan-assembler-times {vmsle\.vi} 7 } } */
+/* { dg-final { scan-assembler-times {vmsleu\.vi} 9 } } */
+/* { dg-final { scan-assembler-not {vmsle\.vv} } } */
+/* { dg-final { scan-assembler-not {vmsleu\.vv} } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/macro.h b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/macro.h
new file mode 100644
index 00000000000..3fe6ee89a99
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/cmp/macro.h
@@ -0,0 +1,11 @@
+#define CMP_VI(NAME, TYPE, NITERS, OP, IMM) \
+ void NAME (int n, TYPE **__restrict a) \
+ { \
+ int b; \
+ int c; \
+ int d; \
+ for (b = 0; b < NITERS; b++) \
+ for (int e = 8; e > 0; e--) \
+ a[b][e] = a[b][e] OP IMM; \
+ }
+
--
2.36.3
More information about the Gcc-patches
mailing list