* [PATCH] RISC-V: Add testcases for VSETVL PASS
@ 2022-12-14 8:09 juzhe.zhong
2022-12-16 20:07 ` Jeff Law
2022-12-26 9:20 ` Andreas Schwab
0 siblings, 2 replies; 9+ messages in thread
From: juzhe.zhong @ 2022-12-14 8:09 UTC (permalink / raw)
To: gcc-patches; +Cc: kito.cheng, palmer, Ju-Zhe Zhong
From: Ju-Zhe Zhong <juzhe.zhong@rivai.ai>
gcc/testsuite/ChangeLog:
* gcc.target/riscv/rvv/rvv.exp: Adjust to enable tests for VSETVL PASS.
* gcc.target/riscv/rvv/vsetvl/dump-1.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-1.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-10.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-11.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-12.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-13.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-14.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-15.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-16.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-17.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-18.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-19.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-2.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-3.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-4.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-5.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-6.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-7.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-8.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_block-9.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-1.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-2.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-3.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-4.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-5.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-6.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-7.c: New test.
* gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-8.c: New test.
---
gcc/testsuite/gcc.target/riscv/rvv/rvv.exp | 2 +
.../gcc.target/riscv/rvv/vsetvl/dump-1.c | 33 ++++
.../riscv/rvv/vsetvl/vlmax_single_block-1.c | 154 ++++++++++++++++++
.../riscv/rvv/vsetvl/vlmax_single_block-10.c | 143 ++++++++++++++++
.../riscv/rvv/vsetvl/vlmax_single_block-11.c | 34 ++++
.../riscv/rvv/vsetvl/vlmax_single_block-12.c | 92 +++++++++++
.../riscv/rvv/vsetvl/vlmax_single_block-13.c | 89 ++++++++++
.../riscv/rvv/vsetvl/vlmax_single_block-14.c | 16 ++
.../riscv/rvv/vsetvl/vlmax_single_block-15.c | 42 +++++
.../riscv/rvv/vsetvl/vlmax_single_block-16.c | 147 +++++++++++++++++
.../riscv/rvv/vsetvl/vlmax_single_block-17.c | 32 ++++
.../riscv/rvv/vsetvl/vlmax_single_block-18.c | 32 ++++
.../riscv/rvv/vsetvl/vlmax_single_block-19.c | 105 ++++++++++++
.../riscv/rvv/vsetvl/vlmax_single_block-2.c | 70 ++++++++
.../riscv/rvv/vsetvl/vlmax_single_block-3.c | 70 ++++++++
.../riscv/rvv/vsetvl/vlmax_single_block-4.c | 49 ++++++
.../riscv/rvv/vsetvl/vlmax_single_block-5.c | 49 ++++++
.../riscv/rvv/vsetvl/vlmax_single_block-6.c | 28 ++++
.../riscv/rvv/vsetvl/vlmax_single_block-7.c | 28 ++++
.../riscv/rvv/vsetvl/vlmax_single_block-8.c | 28 ++++
.../riscv/rvv/vsetvl/vlmax_single_block-9.c | 147 +++++++++++++++++
.../riscv/rvv/vsetvl/vlmax_single_vtype-1.c | 86 ++++++++++
.../riscv/rvv/vsetvl/vlmax_single_vtype-2.c | 42 +++++
.../riscv/rvv/vsetvl/vlmax_single_vtype-3.c | 38 +++++
.../riscv/rvv/vsetvl/vlmax_single_vtype-4.c | 31 ++++
.../riscv/rvv/vsetvl/vlmax_single_vtype-5.c | 31 ++++
.../riscv/rvv/vsetvl/vlmax_single_vtype-6.c | 18 ++
.../riscv/rvv/vsetvl/vlmax_single_vtype-7.c | 18 ++
.../riscv/rvv/vsetvl/vlmax_single_vtype-8.c | 18 ++
29 files changed, 1672 insertions(+)
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/dump-1.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-1.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-10.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-11.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-12.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-13.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-14.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-15.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-16.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-17.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-18.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-19.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-2.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-3.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-4.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-5.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-6.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-7.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-8.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-9.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-1.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-2.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-3.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-4.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-5.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-6.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-7.c
create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-8.c
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/rvv.exp b/gcc/testsuite/gcc.target/riscv/rvv/rvv.exp
index 25e09f41d73..2ed29e22606 100644
--- a/gcc/testsuite/gcc.target/riscv/rvv/rvv.exp
+++ b/gcc/testsuite/gcc.target/riscv/rvv/rvv.exp
@@ -42,6 +42,8 @@ dg-init
set CFLAGS "$DEFAULT_CFLAGS -march=$gcc_march -O3"
dg-runtest [lsort [glob -nocomplain $srcdir/$subdir/base/*.\[cS\]]] \
"" $CFLAGS
+gcc-dg-runtest [lsort [glob -nocomplain $srcdir/$subdir/vsetvl/*.\[cS\]]] \
+ "" $CFLAGS
# All done.
dg-finish
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/dump-1.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/dump-1.c
new file mode 100644
index 00000000000..fb4edb459a0
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/dump-1.c
@@ -0,0 +1,33 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32 -fdump-rtl-vsetvl-details" } */
+
+#include "riscv_vector.h"
+
+void f (void * restrict in, void * restrict out, void * restrict in2, void * restrict out2, int n, int cond)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vuint16mf4_t v2;
+ *(vuint16mf4_t*)(out + i + 1000) = v2;
+ vbool32_t v4;
+ *(vbool32_t*)(out + i + 3000) = v4;
+ vbool16_t v5;
+ *(vbool16_t*)(out + i + 4000) = v5;
+ vbool8_t v6;
+ *(vbool8_t*)(out + i + 5000) = v6;
+ vbool4_t v7;
+ *(vbool4_t*)(out + i + 6000) = v7;
+ vbool2_t v8;
+ *(vbool2_t*)(out + i + 7000) = v8;
+ vbool1_t v9;
+ *(vbool1_t*)(out + i + 8000) = v9;
+ vuint32mf2_t v10;
+ *(vuint32mf2_t*)(out + i + 100000) = v10;
+ }
+
+ for (int i = 0; i < n; i++)
+ {
+ vint8mf8_t v1 = *(vint8mf8_t*)(in + i + 100000);
+ *(vint8mf8_t*)(out + i + 10) = v1;
+ }
+}
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-1.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-1.c
new file mode 100644
index 00000000000..016af8fc67b
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-1.c
@@ -0,0 +1,154 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+/* Test insert-vsetvl PASS whether it's able to eliminate vsetvl for same vtype in VLMAX. */
+
+#include "riscv_vector.h"
+
+void foo1 (void * restrict in, void * restrict out)
+{
+ vbool64_t v1 = *(vbool64_t*)(in + 1);
+ vbool64_t v2 = *(vbool64_t*)(in + 2);
+ vbool64_t v3 = *(vbool64_t*)(in + 3);
+ vbool64_t v4 = *(vbool64_t*)(in + 4);
+ vbool64_t v5 = *(vbool64_t*)(in + 5);
+ vbool64_t v6 = *(vbool64_t*)(in + 6);
+ vbool64_t v7 = *(vbool64_t*)(in + 7);
+ vbool64_t v8 = *(vbool64_t*)(in + 8);
+ *(vbool64_t*)(out + 1) = v1;
+ *(vbool64_t*)(out + 2) = v2;
+ *(vbool64_t*)(out + 3) = v3;
+ *(vbool64_t*)(out + 4) = v4;
+ *(vbool64_t*)(out + 5) = v5;
+ *(vbool64_t*)(out + 6) = v6;
+ *(vbool64_t*)(out + 7) = v7;
+ *(vbool64_t*)(out + 8) = v8;
+}
+
+void foo2 (void * restrict in, void * restrict out)
+{
+ vbool32_t v1 = *(vbool32_t*)(in + 1);
+ vbool32_t v2 = *(vbool32_t*)(in + 2);
+ vbool32_t v3 = *(vbool32_t*)(in + 3);
+ vbool32_t v4 = *(vbool32_t*)(in + 4);
+ vbool32_t v5 = *(vbool32_t*)(in + 5);
+ vbool32_t v6 = *(vbool32_t*)(in + 6);
+ vbool32_t v7 = *(vbool32_t*)(in + 7);
+ vbool32_t v8 = *(vbool32_t*)(in + 8);
+ *(vbool32_t*)(out + 1) = v1;
+ *(vbool32_t*)(out + 2) = v2;
+ *(vbool32_t*)(out + 3) = v3;
+ *(vbool32_t*)(out + 4) = v4;
+ *(vbool32_t*)(out + 5) = v5;
+ *(vbool32_t*)(out + 6) = v6;
+ *(vbool32_t*)(out + 7) = v7;
+ *(vbool32_t*)(out + 8) = v8;
+}
+
+void foo3 (void * restrict in, void * restrict out)
+{
+ vbool16_t v1 = *(vbool16_t*)(in + 1);
+ vbool16_t v2 = *(vbool16_t*)(in + 2);
+ vbool16_t v3 = *(vbool16_t*)(in + 3);
+ vbool16_t v4 = *(vbool16_t*)(in + 4);
+ vbool16_t v5 = *(vbool16_t*)(in + 5);
+ vbool16_t v6 = *(vbool16_t*)(in + 6);
+ vbool16_t v7 = *(vbool16_t*)(in + 7);
+ vbool16_t v8 = *(vbool16_t*)(in + 8);
+ *(vbool16_t*)(out + 1) = v1;
+ *(vbool16_t*)(out + 2) = v2;
+ *(vbool16_t*)(out + 3) = v3;
+ *(vbool16_t*)(out + 4) = v4;
+ *(vbool16_t*)(out + 5) = v5;
+ *(vbool16_t*)(out + 6) = v6;
+ *(vbool16_t*)(out + 7) = v7;
+ *(vbool16_t*)(out + 8) = v8;
+}
+
+void foo4 (void * restrict in, void * restrict out)
+{
+ vbool8_t v1 = *(vbool8_t*)(in + 1);
+ vbool8_t v2 = *(vbool8_t*)(in + 2);
+ vbool8_t v3 = *(vbool8_t*)(in + 3);
+ vbool8_t v4 = *(vbool8_t*)(in + 4);
+ vbool8_t v5 = *(vbool8_t*)(in + 5);
+ vbool8_t v6 = *(vbool8_t*)(in + 6);
+ vbool8_t v7 = *(vbool8_t*)(in + 7);
+ vbool8_t v8 = *(vbool8_t*)(in + 8);
+ *(vbool8_t*)(out + 1) = v1;
+ *(vbool8_t*)(out + 2) = v2;
+ *(vbool8_t*)(out + 3) = v3;
+ *(vbool8_t*)(out + 4) = v4;
+ *(vbool8_t*)(out + 5) = v5;
+ *(vbool8_t*)(out + 6) = v6;
+ *(vbool8_t*)(out + 7) = v7;
+ *(vbool8_t*)(out + 8) = v8;
+}
+
+void foo5 (void * restrict in, void * restrict out)
+{
+ vbool4_t v1 = *(vbool4_t*)(in + 1);
+ vbool4_t v2 = *(vbool4_t*)(in + 2);
+ vbool4_t v3 = *(vbool4_t*)(in + 3);
+ vbool4_t v4 = *(vbool4_t*)(in + 4);
+ vbool4_t v5 = *(vbool4_t*)(in + 5);
+ vbool4_t v6 = *(vbool4_t*)(in + 6);
+ vbool4_t v7 = *(vbool4_t*)(in + 7);
+ vbool4_t v8 = *(vbool4_t*)(in + 8);
+ *(vbool4_t*)(out + 1) = v1;
+ *(vbool4_t*)(out + 2) = v2;
+ *(vbool4_t*)(out + 3) = v3;
+ *(vbool4_t*)(out + 4) = v4;
+ *(vbool4_t*)(out + 5) = v5;
+ *(vbool4_t*)(out + 6) = v6;
+ *(vbool4_t*)(out + 7) = v7;
+ *(vbool4_t*)(out + 8) = v8;
+}
+
+void foo6 (void * restrict in, void * restrict out)
+{
+ vbool2_t v1 = *(vbool2_t*)(in + 1);
+ vbool2_t v2 = *(vbool2_t*)(in + 2);
+ vbool2_t v3 = *(vbool2_t*)(in + 3);
+ vbool2_t v4 = *(vbool2_t*)(in + 4);
+ vbool2_t v5 = *(vbool2_t*)(in + 5);
+ vbool2_t v6 = *(vbool2_t*)(in + 6);
+ vbool2_t v7 = *(vbool2_t*)(in + 7);
+ vbool2_t v8 = *(vbool2_t*)(in + 8);
+ *(vbool2_t*)(out + 1) = v1;
+ *(vbool2_t*)(out + 2) = v2;
+ *(vbool2_t*)(out + 3) = v3;
+ *(vbool2_t*)(out + 4) = v4;
+ *(vbool2_t*)(out + 5) = v5;
+ *(vbool2_t*)(out + 6) = v6;
+ *(vbool2_t*)(out + 7) = v7;
+ *(vbool2_t*)(out + 8) = v8;
+}
+
+void foo7 (void * restrict in, void * restrict out)
+{
+ vbool1_t v1 = *(vbool1_t*)(in + 1);
+ vbool1_t v2 = *(vbool1_t*)(in + 2);
+ vbool1_t v3 = *(vbool1_t*)(in + 3);
+ vbool1_t v4 = *(vbool1_t*)(in + 4);
+ vbool1_t v5 = *(vbool1_t*)(in + 5);
+ vbool1_t v6 = *(vbool1_t*)(in + 6);
+ vbool1_t v7 = *(vbool1_t*)(in + 7);
+ vbool1_t v8 = *(vbool1_t*)(in + 8);
+ *(vbool1_t*)(out + 1) = v1;
+ *(vbool1_t*)(out + 2) = v2;
+ *(vbool1_t*)(out + 3) = v3;
+ *(vbool1_t*)(out + 4) = v4;
+ *(vbool1_t*)(out + 5) = v5;
+ *(vbool1_t*)(out + 6) = v6;
+ *(vbool1_t*)(out + 7) = v7;
+ *(vbool1_t*)(out + 8) = v8;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-10.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-10.c
new file mode 100644
index 00000000000..185db998df1
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-10.c
@@ -0,0 +1,143 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+void f1 (void * restrict in, void * restrict out)
+{
+ vint16mf4_t v1 = *(vint16mf4_t*)(in + 1);
+ vint8mf8_t v2;
+ *(vint16mf4_t*)(out + 1) = v1;
+ *(vint8mf8_t*)(out + 2) = v2;
+}
+
+void f2 (void * restrict in, void * restrict out)
+{
+ vint16mf4_t v1 = *(vint16mf4_t*)(in + 1);
+ vint32mf2_t v2;
+ *(vint16mf4_t*)(out + 1) = v1;
+ *(vint32mf2_t*)(out + 2) = v2;
+}
+
+void f3 (void * restrict in, void * restrict out)
+{
+ vint16mf4_t v1 = *(vint16mf4_t*)(in + 1);
+ vint64m1_t v2;
+ *(vint16mf4_t*)(out + 1) = v1;
+ *(vint64m1_t*)(out + 2) = v2;
+}
+
+void f4 (void * restrict in, void * restrict out)
+{
+ vint16mf2_t v1 = *(vint16mf2_t*)(in + 1);
+ vint8mf4_t v2;
+ *(vint16mf2_t*)(out + 1) = v1;
+ *(vint8mf4_t*)(out + 2) = v2;
+}
+
+void f5 (void * restrict in, void * restrict out)
+{
+ vint16mf2_t v1 = *(vint16mf2_t*)(in + 1);
+ vint32m1_t v2;
+ *(vint16mf2_t*)(out + 1) = v1;
+ *(vint32m1_t*)(out + 2) = v2;
+}
+
+void f6 (void * restrict in, void * restrict out)
+{
+ vint16mf2_t v1 = *(vint16mf2_t*)(in + 1);
+ vint64m2_t v2;
+ *(vint16mf2_t*)(out + 1) = v1;
+ *(vint64m2_t*)(out + 2) = v2;
+}
+
+void f7 (void * restrict in, void * restrict out)
+{
+ vint16m1_t v1 = *(vint16m1_t*)(in + 1);
+ vint8mf2_t v2;
+ *(vint16m1_t*)(out + 1) = v1;
+ *(vint8mf2_t*)(out + 2) = v2;
+}
+
+void f8 (void * restrict in, void * restrict out)
+{
+ vint16m1_t v1 = *(vint16m1_t*)(in + 1);
+ vint32m2_t v2;
+ *(vint16m1_t*)(out + 1) = v1;
+ *(vint32m2_t*)(out + 2) = v2;
+}
+
+void f9 (void * restrict in, void * restrict out)
+{
+ vint16m1_t v1 = *(vint16m1_t*)(in + 1);
+ vint64m4_t v2;
+ *(vint16m1_t*)(out + 1) = v1;
+ *(vint64m4_t*)(out + 2) = v2;
+}
+
+void f10 (void * restrict in, void * restrict out)
+{
+ vint16m2_t v1 = *(vint16m2_t*)(in + 1);
+ vint8m1_t v2;
+ *(vint16m2_t*)(out + 1) = v1;
+ *(vint8m1_t*)(out + 2) = v2;
+}
+
+void f11 (void * restrict in, void * restrict out)
+{
+ vint16m2_t v1 = *(vint16m2_t*)(in + 1);
+ vint32m4_t v2;
+ *(vint16m2_t*)(out + 1) = v1;
+ *(vint32m4_t*)(out + 2) = v2;
+}
+
+void f12 (void * restrict in, void * restrict out)
+{
+ vint16m2_t v1 = *(vint16m2_t*)(in + 1);
+ vint64m8_t v2;
+ *(vint16m2_t*)(out + 1) = v1;
+ *(vint64m8_t*)(out + 2) = v2;
+}
+
+void f13 (void * restrict in, void * restrict out)
+{
+ vint16m4_t v1 = *(vint16m4_t*)(in + 1);
+ vint8m2_t v2;
+ *(vint16m4_t*)(out + 1) = v1;
+ *(vint8m2_t*)(out + 2) = v2;
+}
+
+void f14 (void * restrict in, void * restrict out)
+{
+ vint16m4_t v1 = *(vint16m4_t*)(in + 1);
+ vint32m8_t v2;
+ *(vint16m4_t*)(out + 1) = v1;
+ *(vint32m8_t*)(out + 2) = v2;
+}
+
+void f15 (void * restrict in, void * restrict out)
+{
+ vint16m8_t v1 = *(vint16m8_t*)(in + 1);
+ vint8m4_t v2;
+ *(vint16m8_t*)(out + 1) = v1;
+ *(vint8m4_t*)(out + 2) = v2;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli} 15 { target { no-opts "-O0" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-11.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-11.c
new file mode 100644
index 00000000000..95746b763e0
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-11.c
@@ -0,0 +1,34 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+void f1 (void * restrict in, void * restrict out)
+{
+ vint32mf2_t v1 = *(vint32mf2_t*)(in + 1);
+ vint8mf8_t v2;
+ *(vint32mf2_t*)(out + 1) = v1;
+ *(vint8mf8_t*)(out + 2) = v2;
+}
+
+void f2 (void * restrict in, void * restrict out)
+{
+ vint32mf2_t v1 = *(vint32mf2_t*)(in + 1);
+ vint16mf4_t v2;
+ *(vint32mf2_t*)(out + 1) = v1;
+ *(vint16mf4_t*)(out + 2) = v2;
+}
+
+void f3 (void * restrict in, void * restrict out)
+{
+ vint32mf2_t v1 = *(vint32mf2_t*)(in + 1);
+ vint64m1_t v2;
+ *(vint32mf2_t*)(out + 1) = v1;
+ *(vint64m1_t*)(out + 2) = v2;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli} 3 { target { no-opts "-O0" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-12.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-12.c
new file mode 100644
index 00000000000..d3442457861
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-12.c
@@ -0,0 +1,92 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+void f1 (void * restrict in, void * restrict out)
+{
+ vint8mf8_t v1 = *(vint8mf8_t*)(in + 1);
+ vfloat32mf2_t v2;
+ *(vint8mf8_t*)(out + 1) = v1;
+ *(vfloat32mf2_t*)(out + 2) = v2;
+}
+
+void f2 (void * restrict in, void * restrict out)
+{
+ vint8mf8_t v1 = *(vint8mf8_t*)(in + 1);
+ vfloat64m1_t v2;
+ *(vint8mf8_t*)(out + 1) = v1;
+ *(vfloat64m1_t*)(out + 2) = v2;
+}
+
+void f3 (void * restrict in, void * restrict out)
+{
+ vint8mf4_t v1 = *(vint8mf4_t*)(in + 1);
+ vfloat32m1_t v2;
+ *(vint8mf4_t*)(out + 1) = v1;
+ *(vfloat32m1_t*)(out + 2) = v2;
+}
+
+void f4 (void * restrict in, void * restrict out)
+{
+ vint8mf4_t v1 = *(vint8mf4_t*)(in + 1);
+ vfloat64m2_t v2;
+ *(vint8mf4_t*)(out + 1) = v1;
+ *(vfloat64m2_t*)(out + 2) = v2;
+}
+
+void f5 (void * restrict in, void * restrict out)
+{
+ vint8mf2_t v1 = *(vint8mf2_t*)(in + 1);
+ vfloat32m2_t v2;
+ *(vint8mf2_t*)(out + 1) = v1;
+ *(vfloat32m2_t*)(out + 2) = v2;
+}
+
+void f6 (void * restrict in, void * restrict out)
+{
+ vint8mf2_t v1 = *(vint8mf2_t*)(in + 1);
+ vfloat64m4_t v2;
+ *(vint8mf2_t*)(out + 1) = v1;
+ *(vfloat64m4_t*)(out + 2) = v2;
+}
+
+void f7 (void * restrict in, void * restrict out)
+{
+ vint8m1_t v1 = *(vint8m1_t*)(in + 1);
+ vfloat32m4_t v2;
+ *(vint8m1_t*)(out + 1) = v1;
+ *(vfloat32m4_t*)(out + 2) = v2;
+}
+
+void f8 (void * restrict in, void * restrict out)
+{
+ vint8m1_t v1 = *(vint8m1_t*)(in + 1);
+ vfloat64m8_t v2;
+ *(vint8m1_t*)(out + 1) = v1;
+ *(vfloat64m8_t*)(out + 2) = v2;
+}
+
+void f9 (void * restrict in, void * restrict out)
+{
+ vint8m2_t v1 = *(vint8m2_t*)(in + 1);
+ vfloat32m8_t v2;
+ *(vint8m2_t*)(out + 1) = v1;
+ *(vfloat32m8_t*)(out + 2) = v2;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli} 9 { target { no-opts "-O0" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-13.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-13.c
new file mode 100644
index 00000000000..9ed1020af29
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-13.c
@@ -0,0 +1,89 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+void f1 (void * restrict in, void * restrict out)
+{
+ vint16mf4_t v1 = *(vint16mf4_t*)(in + 1);
+ vfloat32mf2_t v2;
+ *(vint16mf4_t*)(out + 1) = v1;
+ *(vfloat32mf2_t*)(out + 2) = v2;
+}
+
+void f2 (void * restrict in, void * restrict out)
+{
+ vint16mf4_t v1 = *(vint16mf4_t*)(in + 1);
+ vfloat64m1_t v2;
+ *(vint16mf4_t*)(out + 1) = v1;
+ *(vfloat64m1_t*)(out + 2) = v2;
+}
+
+void f3 (void * restrict in, void * restrict out)
+{
+ vint16mf2_t v1 = *(vint16mf2_t*)(in + 1);
+ vfloat32m1_t v2;
+ *(vint16mf2_t*)(out + 1) = v1;
+ *(vfloat32m1_t*)(out + 2) = v2;
+}
+
+void f4 (void * restrict in, void * restrict out)
+{
+ vint16mf2_t v1 = *(vint16mf2_t*)(in + 1);
+ vfloat64m2_t v2;
+ *(vint16mf2_t*)(out + 1) = v1;
+ *(vfloat64m2_t*)(out + 2) = v2;
+}
+
+void f5 (void * restrict in, void * restrict out)
+{
+ vint16m1_t v1 = *(vint16m1_t*)(in + 1);
+ vfloat32m2_t v2;
+ *(vint16m1_t*)(out + 1) = v1;
+ *(vfloat32m2_t*)(out + 2) = v2;
+}
+
+void f6 (void * restrict in, void * restrict out)
+{
+ vint16m1_t v1 = *(vint16m1_t*)(in + 1);
+ vfloat64m4_t v2;
+ *(vint16m1_t*)(out + 1) = v1;
+ *(vfloat64m4_t*)(out + 2) = v2;
+}
+
+void f7 (void * restrict in, void * restrict out)
+{
+ vint16m2_t v1 = *(vint16m2_t*)(in + 1);
+ vfloat32m4_t v2;
+ *(vint16m2_t*)(out + 1) = v1;
+ *(vfloat32m4_t*)(out + 2) = v2;
+}
+
+void f8 (void * restrict in, void * restrict out)
+{
+ vint16m2_t v1 = *(vint16m2_t*)(in + 1);
+ vfloat64m8_t v2;
+ *(vint16m2_t*)(out + 1) = v1;
+ *(vfloat64m8_t*)(out + 2) = v2;
+}
+
+void f9 (void * restrict in, void * restrict out)
+{
+ vint16m4_t v1 = *(vint16m4_t*)(in + 1);
+ vfloat32m8_t v2;
+ *(vint16m4_t*)(out + 1) = v1;
+ *(vfloat32m8_t*)(out + 2) = v2;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli} 9 { target { no-opts "-O0" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-14.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-14.c
new file mode 100644
index 00000000000..f5d91c68a16
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-14.c
@@ -0,0 +1,16 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+void f3 (void * restrict in, void * restrict out)
+{
+ vint32mf2_t v1 = *(vint32mf2_t*)(in + 1);
+ vfloat64m1_t v2;
+ *(vint32mf2_t*)(out + 1) = v1;
+ *(vfloat64m1_t*)(out + 2) = v2;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli} 1 { target { no-opts "-O0" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-15.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-15.c
new file mode 100644
index 00000000000..3d9e8370cc9
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-15.c
@@ -0,0 +1,42 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+void f1 (void * restrict in, void * restrict out)
+{
+ vfloat32mf2_t v1 = *(vfloat32mf2_t*)(in + 1);
+ vint8mf8_t v2;
+ *(vfloat32mf2_t*)(out + 1) = v1;
+ *(vint8mf8_t*)(out + 2) = v2;
+}
+
+void f2 (void * restrict in, void * restrict out)
+{
+ vfloat32mf2_t v1 = *(vfloat32mf2_t*)(in + 1);
+ vint16mf4_t v2;
+ *(vfloat32mf2_t*)(out + 1) = v1;
+ *(vint16mf4_t*)(out + 2) = v2;
+}
+
+void f3 (void * restrict in, void * restrict out)
+{
+ vfloat32mf2_t v1 = *(vfloat32mf2_t*)(in + 1);
+ vint64m1_t v2;
+ *(vfloat32mf2_t*)(out + 1) = v1;
+ *(vint64m1_t*)(out + 2) = v2;
+}
+
+void f4 (void * restrict in, void * restrict out)
+{
+ vfloat32mf2_t v1 = *(vfloat32mf2_t*)(in + 1);
+ vfloat64m1_t v2;
+ *(vfloat32mf2_t*)(out + 1) = v1;
+ *(vfloat64m1_t*)(out + 2) = v2;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m1,\s*t[au],\s*m[au]} 2 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli} 4 { target { no-opts "-O0" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-16.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-16.c
new file mode 100644
index 00000000000..6ee416088ca
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-16.c
@@ -0,0 +1,147 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+void f1 (void * restrict in, void * restrict out)
+{
+ vbool64_t v1 = *(vbool64_t*)(in + 1);
+ vint16mf4_t v2;
+ *(vbool64_t*)(out + 1) = v1;
+ *(vint16mf4_t*)(out + 2) = v2;
+}
+
+void f2 (void * restrict in, void * restrict out)
+{
+ vbool64_t v1 = *(vbool64_t*)(in + 1);
+ vint32mf2_t v2;
+ *(vbool64_t*)(out + 1) = v1;
+ *(vint32mf2_t*)(out + 2) = v2;
+}
+
+void f3 (void * restrict in, void * restrict out)
+{
+ vbool64_t v1 = *(vbool64_t*)(in + 1);
+ vint64m1_t v2;
+ *(vbool64_t*)(out + 1) = v1;
+ *(vint64m1_t*)(out + 2) = v2;
+}
+
+void f4 (void * restrict in, void * restrict out)
+{
+ vbool32_t v1 = *(vbool32_t*)(in + 1);
+ vint16mf2_t v2;
+ *(vbool32_t*)(out + 1) = v1;
+ *(vint16mf2_t*)(out + 2) = v2;
+}
+
+void f5 (void * restrict in, void * restrict out)
+{
+ vbool32_t v1 = *(vbool32_t*)(in + 1);
+ vint32m1_t v2;
+ *(vbool32_t*)(out + 1) = v1;
+ *(vint32m1_t*)(out + 2) = v2;
+}
+
+void f6 (void * restrict in, void * restrict out)
+{
+ vbool32_t v1 = *(vbool32_t*)(in + 1);
+ vint64m2_t v2;
+ *(vbool32_t*)(out + 1) = v1;
+ *(vint64m2_t*)(out + 2) = v2;
+}
+
+void f7 (void * restrict in, void * restrict out)
+{
+ vbool16_t v1 = *(vbool16_t*)(in + 1);
+ vint16m1_t v2;
+ *(vbool16_t*)(out + 1) = v1;
+ *(vint16m1_t*)(out + 2) = v2;
+}
+
+void f8 (void * restrict in, void * restrict out)
+{
+ vbool16_t v1 = *(vbool16_t*)(in + 1);
+ vint32m2_t v2;
+ *(vbool16_t*)(out + 1) = v1;
+ *(vint32m2_t*)(out + 2) = v2;
+}
+
+void f9 (void * restrict in, void * restrict out)
+{
+ vbool16_t v1 = *(vbool16_t*)(in + 1);
+ vint64m4_t v2;
+ *(vbool16_t*)(out + 1) = v1;
+ *(vint64m4_t*)(out + 2) = v2;
+}
+
+void f10 (void * restrict in, void * restrict out)
+{
+ vbool8_t v1 = *(vbool8_t*)(in + 1);
+ vint16m2_t v2;
+ *(vbool8_t*)(out + 1) = v1;
+ *(vint16m2_t*)(out + 2) = v2;
+}
+
+void f11 (void * restrict in, void * restrict out)
+{
+ vbool8_t v1 = *(vbool8_t*)(in + 1);
+ vint32m4_t v2;
+ *(vbool8_t*)(out + 1) = v1;
+ *(vint32m4_t*)(out + 2) = v2;
+}
+
+void f12 (void * restrict in, void * restrict out)
+{
+ vbool8_t v1 = *(vbool8_t*)(in + 1);
+ vint64m8_t v2;
+ *(vbool8_t*)(out + 1) = v1;
+ *(vint64m8_t*)(out + 2) = v2;
+}
+
+void f13 (void * restrict in, void * restrict out)
+{
+ vbool4_t v1 = *(vbool4_t*)(in + 1);
+ vint16m4_t v2;
+ *(vbool4_t*)(out + 1) = v1;
+ *(vint16m4_t*)(out + 2) = v2;
+}
+
+void f14 (void * restrict in, void * restrict out)
+{
+ vbool4_t v1 = *(vbool4_t*)(in + 1);
+ vint32m8_t v2;
+ *(vbool4_t*)(out + 1) = v1;
+ *(vint32m8_t*)(out + 2) = v2;
+}
+
+void f15 (void * restrict in, void * restrict out)
+{
+ vbool2_t v1 = *(vbool2_t*)(in + 1);
+ vint16m8_t v2;
+ *(vbool2_t*)(out + 1) = v1;
+ *(vint16m8_t*)(out + 2) = v2;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*m2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*m4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli} 15 { target { no-opts "-O0" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-17.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-17.c
new file mode 100644
index 00000000000..7fba5560f88
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-17.c
@@ -0,0 +1,32 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+void foo7 (void * restrict in, void * restrict out)
+{
+ vbool1_t v1 = *(vbool1_t*)(in + 1);
+ vbool2_t v2 = *(vbool2_t*)(in + 2);
+ vbool4_t v3 = *(vbool4_t*)(in + 3);
+ vbool8_t v4 = *(vbool8_t*)(in + 4);
+ vbool16_t v5 = *(vbool16_t*)(in + 5);
+ vbool32_t v6 = *(vbool32_t*)(in + 6);
+ vbool64_t v7 = *(vbool64_t*)(in + 7);
+ *(vbool1_t*)(out + 1) = v1;
+ *(vbool2_t*)(out + 2) = v2;
+ *(vbool4_t*)(out + 3) = v3;
+ *(vbool8_t*)(out + 4) = v4;
+ *(vbool16_t*)(out + 5) = v5;
+ *(vbool32_t*)(out + 6) = v6;
+ *(vbool64_t*)(out + 7) = v7;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" } } } } */
+/* { dg-final { scan-assembler-times {vlm\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 7 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" } } } } */
+/* { dg-final { scan-assembler-times {vsm\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 7 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-18.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-18.c
new file mode 100644
index 00000000000..01ea59db736
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-18.c
@@ -0,0 +1,32 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+void f (void * restrict in, void * restrict out)
+{
+ vint8mf8_t v1 = *(vint8mf8_t*)(in + 1);
+ vint16mf4_t v2 = *(vint16mf4_t*)(in + 2);
+ vint32mf2_t v3 = *(vint32mf2_t*)(in + 3);
+ vfloat32mf2_t v4 = *(vfloat32mf2_t*)(in + 4);
+
+ vint8mf4_t v5 = *(vint8mf4_t*)(in + 5);
+ vint16mf2_t v6 = *(vint16mf2_t*)(in + 6);
+
+ vint8mf2_t v7 = *(vint8mf2_t*)(in + 7);
+
+ *(vint8mf8_t*)(out + 1) = v1;
+ *(vint16mf4_t*)(out + 2) = v2;
+ *(vint32mf2_t*)(out + 3) = v3;
+ *(vfloat32mf2_t*)(out + 4) = v4;
+
+ *(vint8mf4_t*)(out + 5) = v5;
+ *(vint16mf2_t*)(out + 6) = v6;
+
+ *(vint8mf2_t*)(out + 7) = v7;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli} 3 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-19.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-19.c
new file mode 100644
index 00000000000..431ec82846b
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-19.c
@@ -0,0 +1,105 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+void f (void * restrict in, void * restrict out)
+{
+ vint8mf8_t v1 = *(vint8mf8_t*)(in + 1);
+ vint16mf4_t v2;
+ *(vint8mf8_t*)(out + 1) = v1;
+ *(vint16mf4_t*)(out + 2) = v2;
+
+ vint8mf8_t v3 = *(vint8mf8_t*)(in + 3);
+ vint32mf2_t v4;
+ *(vint8mf8_t*)(out + 3) = v3;
+ *(vint32mf2_t*)(out + 4) = v4;
+
+ vint8mf8_t v5 = *(vint8mf8_t*)(in + 5);
+ vint64m1_t v6;
+ *(vint8mf8_t*)(out + 5) = v5;
+ *(vint64m1_t*)(out + 6) = v6;
+
+ vint8mf4_t v7 = *(vint8mf4_t*)(in + 7);
+ vint16mf2_t v8;
+ *(vint8mf4_t*)(out + 7) = v7;
+ *(vint16mf2_t*)(out + 8) = v8;
+
+ vint8mf4_t v9 = *(vint8mf4_t*)(in + 9);
+ vint32m1_t v10;
+ *(vint8mf4_t*)(out + 9) = v9;
+ *(vint32m1_t*)(out + 10) = v10;
+
+ vint8mf4_t v11 = *(vint8mf4_t*)(in + 11);
+ vint64m2_t v12;
+ *(vint8mf4_t*)(out + 11) = v11;
+ *(vint64m2_t*)(out + 12) = v12;
+
+ vint8mf2_t v13 = *(vint8mf2_t*)(in + 13);
+ vint16m1_t v14;
+ *(vint8mf2_t*)(out + 13) = v13;
+ *(vint16m1_t*)(out + 14) = v14;
+
+ vint8mf2_t v15 = *(vint8mf2_t*)(in + 15);
+ vint32m2_t v16;
+ *(vint8mf2_t*)(out + 15) = v15;
+ *(vint32m2_t*)(out + 16) = v16;
+
+ vint8mf2_t v17 = *(vint8mf2_t*)(in + 17);
+ vint64m4_t v18;
+ *(vint8mf2_t*)(out + 17) = v17;
+ *(vint64m4_t*)(out + 18) = v18;
+
+ vint8m1_t v19 = *(vint8m1_t*)(in + 19);
+ vint16m2_t v20;
+ *(vint8m1_t*)(out + 19) = v19;
+ *(vint16m2_t*)(out + 20) = v20;
+
+ vint8m1_t v21 = *(vint8m1_t*)(in + 20);
+ vint32m4_t v22;
+ *(vint8m1_t*)(out + 20) = v21;
+ *(vint32m4_t*)(out + 21) = v22;
+
+ vint8m1_t v23 = *(vint8m1_t*)(in + 23);
+ vint64m8_t v24;
+ *(vint8m1_t*)(out + 21) = v23;
+ *(vint64m8_t*)(out + 22) = v24;
+
+ vint8m2_t v25 = *(vint8m2_t*)(in + 25);
+ vint16m4_t v26;
+ *(vint8m2_t*)(out + 25) = v25;
+ *(vint16m4_t*)(out + 26) = v26;
+
+ vint8m2_t v27 = *(vint8m2_t*)(in + 27);
+ vint32m8_t v28;
+ *(vint8m2_t*)(out + 27) = v27;
+ *(vint32m8_t*)(out + 28) = v28;
+
+ vint8m4_t v29 = *(vint8m4_t*)(in + 29);
+ vint16m8_t v30;
+ *(vint8m4_t*)(out + 29) = v29;
+ *(vint16m8_t*)(out + 30) = v30;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" no-opts "-O2" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+zero,\s*zero,\s*e32,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" no-opts "-O2" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+zero,\s*zero,\s*e64,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" no-opts "-O2" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" no-opts "-O2" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+zero,\s*zero,\s*e32,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" no-opts "-O2" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+zero,\s*zero,\s*e64,\s*m2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" no-opts "-O2" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" no-opts "-O2" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+zero,\s*zero,\s*e32,\s*m2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" no-opts "-O2" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+zero,\s*zero,\s*e64,\s*m4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" no-opts "-O2" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*m2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" no-opts "-O2" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+zero,\s*zero,\s*e32,\s*m4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" no-opts "-O2" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+zero,\s*zero,\s*e64,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" no-opts "-O2" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*m4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" no-opts "-O2" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+zero,\s*zero,\s*e32,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" no-opts "-O2" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" no-opts "-O2" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli} 15 { target { no-opts "-O0" no-opts "-O1" no-opts "-Os" no-opts "-O2" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-2.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-2.c
new file mode 100644
index 00000000000..1660ab7d8d8
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-2.c
@@ -0,0 +1,70 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+/* Test insert-vsetvl PASS whether it's able to eliminate vsetvl for same vtype in VLMAX. */
+
+#include "riscv_vector.h"
+
+void foo1 (void * restrict in, void * restrict out)
+{
+ vint8mf8_t v1 = *(vint8mf8_t*)(in + 1);
+ vint8mf8_t v2 = *(vint8mf8_t*)(in + 2);
+ vint8mf8_t v3 = *(vint8mf8_t*)(in + 3);
+ vint8mf8_t v4 = *(vint8mf8_t*)(in + 4);
+ vint8mf8_t v5 = *(vint8mf8_t*)(in + 5);
+ vint8mf8_t v6 = *(vint8mf8_t*)(in + 6);
+ vint8mf8_t v7 = *(vint8mf8_t*)(in + 7);
+ vint8mf8_t v8 = *(vint8mf8_t*)(in + 8);
+ *(vint8mf8_t*)(out + 1) = v1;
+ *(vint8mf8_t*)(out + 2) = v2;
+ *(vint8mf8_t*)(out + 3) = v3;
+ *(vint8mf8_t*)(out + 4) = v4;
+ *(vint8mf8_t*)(out + 5) = v5;
+ *(vint8mf8_t*)(out + 6) = v6;
+ *(vint8mf8_t*)(out + 7) = v7;
+ *(vint8mf8_t*)(out + 8) = v8;
+}
+
+void foo2 (void * restrict in, void * restrict out)
+{
+ vint8mf4_t v1 = *(vint8mf4_t*)(in + 1);
+ vint8mf4_t v2 = *(vint8mf4_t*)(in + 2);
+ vint8mf4_t v3 = *(vint8mf4_t*)(in + 3);
+ vint8mf4_t v4 = *(vint8mf4_t*)(in + 4);
+ vint8mf4_t v5 = *(vint8mf4_t*)(in + 5);
+ vint8mf4_t v6 = *(vint8mf4_t*)(in + 6);
+ vint8mf4_t v7 = *(vint8mf4_t*)(in + 7);
+ vint8mf4_t v8 = *(vint8mf4_t*)(in + 8);
+ *(vint8mf4_t*)(out + 1) = v1;
+ *(vint8mf4_t*)(out + 2) = v2;
+ *(vint8mf4_t*)(out + 3) = v3;
+ *(vint8mf4_t*)(out + 4) = v4;
+ *(vint8mf4_t*)(out + 5) = v5;
+ *(vint8mf4_t*)(out + 6) = v6;
+ *(vint8mf4_t*)(out + 7) = v7;
+ *(vint8mf4_t*)(out + 8) = v8;
+}
+
+void foo3 (void * restrict in, void * restrict out)
+{
+ vint8mf2_t v1 = *(vint8mf2_t*)(in + 1);
+ vint8mf2_t v2 = *(vint8mf2_t*)(in + 2);
+ vint8mf2_t v3 = *(vint8mf2_t*)(in + 3);
+ vint8mf2_t v4 = *(vint8mf2_t*)(in + 4);
+ vint8mf2_t v5 = *(vint8mf2_t*)(in + 5);
+ vint8mf2_t v6 = *(vint8mf2_t*)(in + 6);
+ vint8mf2_t v7 = *(vint8mf2_t*)(in + 7);
+ vint8mf2_t v8 = *(vint8mf2_t*)(in + 8);
+ *(vint8mf2_t*)(out + 1) = v1;
+ *(vint8mf2_t*)(out + 2) = v2;
+ *(vint8mf2_t*)(out + 3) = v3;
+ *(vint8mf2_t*)(out + 4) = v4;
+ *(vint8mf2_t*)(out + 5) = v5;
+ *(vint8mf2_t*)(out + 6) = v6;
+ *(vint8mf2_t*)(out + 7) = v7;
+ *(vint8mf2_t*)(out + 8) = v8;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-3.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-3.c
new file mode 100644
index 00000000000..13c801fc33e
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-3.c
@@ -0,0 +1,70 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+/* Test insert-vsetvl PASS whether it's able to eliminate vsetvl for same vtype in VLMAX. */
+
+#include "riscv_vector.h"
+
+void foo1 (void * restrict in, void * restrict out)
+{
+ vuint8mf8_t v1 = *(vuint8mf8_t*)(in + 1);
+ vuint8mf8_t v2 = *(vuint8mf8_t*)(in + 2);
+ vuint8mf8_t v3 = *(vuint8mf8_t*)(in + 3);
+ vuint8mf8_t v4 = *(vuint8mf8_t*)(in + 4);
+ vuint8mf8_t v5 = *(vuint8mf8_t*)(in + 5);
+ vuint8mf8_t v6 = *(vuint8mf8_t*)(in + 6);
+ vuint8mf8_t v7 = *(vuint8mf8_t*)(in + 7);
+ vuint8mf8_t v8 = *(vuint8mf8_t*)(in + 8);
+ *(vuint8mf8_t*)(out + 1) = v1;
+ *(vuint8mf8_t*)(out + 2) = v2;
+ *(vuint8mf8_t*)(out + 3) = v3;
+ *(vuint8mf8_t*)(out + 4) = v4;
+ *(vuint8mf8_t*)(out + 5) = v5;
+ *(vuint8mf8_t*)(out + 6) = v6;
+ *(vuint8mf8_t*)(out + 7) = v7;
+ *(vuint8mf8_t*)(out + 8) = v8;
+}
+
+void foo2 (void * restrict in, void * restrict out)
+{
+ vuint8mf4_t v1 = *(vuint8mf4_t*)(in + 1);
+ vuint8mf4_t v2 = *(vuint8mf4_t*)(in + 2);
+ vuint8mf4_t v3 = *(vuint8mf4_t*)(in + 3);
+ vuint8mf4_t v4 = *(vuint8mf4_t*)(in + 4);
+ vuint8mf4_t v5 = *(vuint8mf4_t*)(in + 5);
+ vuint8mf4_t v6 = *(vuint8mf4_t*)(in + 6);
+ vuint8mf4_t v7 = *(vuint8mf4_t*)(in + 7);
+ vuint8mf4_t v8 = *(vuint8mf4_t*)(in + 8);
+ *(vuint8mf4_t*)(out + 1) = v1;
+ *(vuint8mf4_t*)(out + 2) = v2;
+ *(vuint8mf4_t*)(out + 3) = v3;
+ *(vuint8mf4_t*)(out + 4) = v4;
+ *(vuint8mf4_t*)(out + 5) = v5;
+ *(vuint8mf4_t*)(out + 6) = v6;
+ *(vuint8mf4_t*)(out + 7) = v7;
+ *(vuint8mf4_t*)(out + 8) = v8;
+}
+
+void foo3 (void * restrict in, void * restrict out)
+{
+ vuint8mf2_t v1 = *(vuint8mf2_t*)(in + 1);
+ vuint8mf2_t v2 = *(vuint8mf2_t*)(in + 2);
+ vuint8mf2_t v3 = *(vuint8mf2_t*)(in + 3);
+ vuint8mf2_t v4 = *(vuint8mf2_t*)(in + 4);
+ vuint8mf2_t v5 = *(vuint8mf2_t*)(in + 5);
+ vuint8mf2_t v6 = *(vuint8mf2_t*)(in + 6);
+ vuint8mf2_t v7 = *(vuint8mf2_t*)(in + 7);
+ vuint8mf2_t v8 = *(vuint8mf2_t*)(in + 8);
+ *(vuint8mf2_t*)(out + 1) = v1;
+ *(vuint8mf2_t*)(out + 2) = v2;
+ *(vuint8mf2_t*)(out + 3) = v3;
+ *(vuint8mf2_t*)(out + 4) = v4;
+ *(vuint8mf2_t*)(out + 5) = v5;
+ *(vuint8mf2_t*)(out + 6) = v6;
+ *(vuint8mf2_t*)(out + 7) = v7;
+ *(vuint8mf2_t*)(out + 8) = v8;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-4.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-4.c
new file mode 100644
index 00000000000..2a597986e2a
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-4.c
@@ -0,0 +1,49 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+/* Test insert-vsetvl PASS whether it's able to eliminate vsetvl for same vtype in VLMAX. */
+
+#include "riscv_vector.h"
+
+void foo2 (void * restrict in, void * restrict out)
+{
+ vint16mf4_t v1 = *(vint16mf4_t*)(in + 1);
+ vint16mf4_t v2 = *(vint16mf4_t*)(in + 2);
+ vint16mf4_t v3 = *(vint16mf4_t*)(in + 3);
+ vint16mf4_t v4 = *(vint16mf4_t*)(in + 4);
+ vint16mf4_t v5 = *(vint16mf4_t*)(in + 5);
+ vint16mf4_t v6 = *(vint16mf4_t*)(in + 6);
+ vint16mf4_t v7 = *(vint16mf4_t*)(in + 7);
+ vint16mf4_t v8 = *(vint16mf4_t*)(in + 8);
+ *(vint16mf4_t*)(out + 1) = v1;
+ *(vint16mf4_t*)(out + 2) = v2;
+ *(vint16mf4_t*)(out + 3) = v3;
+ *(vint16mf4_t*)(out + 4) = v4;
+ *(vint16mf4_t*)(out + 5) = v5;
+ *(vint16mf4_t*)(out + 6) = v6;
+ *(vint16mf4_t*)(out + 7) = v7;
+ *(vint16mf4_t*)(out + 8) = v8;
+}
+
+void foo3 (void * restrict in, void * restrict out)
+{
+ vint16mf2_t v1 = *(vint16mf2_t*)(in + 1);
+ vint16mf2_t v2 = *(vint16mf2_t*)(in + 2);
+ vint16mf2_t v3 = *(vint16mf2_t*)(in + 3);
+ vint16mf2_t v4 = *(vint16mf2_t*)(in + 4);
+ vint16mf2_t v5 = *(vint16mf2_t*)(in + 5);
+ vint16mf2_t v6 = *(vint16mf2_t*)(in + 6);
+ vint16mf2_t v7 = *(vint16mf2_t*)(in + 7);
+ vint16mf2_t v8 = *(vint16mf2_t*)(in + 8);
+ *(vint16mf2_t*)(out + 1) = v1;
+ *(vint16mf2_t*)(out + 2) = v2;
+ *(vint16mf2_t*)(out + 3) = v3;
+ *(vint16mf2_t*)(out + 4) = v4;
+ *(vint16mf2_t*)(out + 5) = v5;
+ *(vint16mf2_t*)(out + 6) = v6;
+ *(vint16mf2_t*)(out + 7) = v7;
+ *(vint16mf2_t*)(out + 8) = v8;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-5.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-5.c
new file mode 100644
index 00000000000..982ec8c2567
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-5.c
@@ -0,0 +1,49 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+/* Test insert-vsetvl PASS whether it's able to eliminate vsetvl for same vtype in VLMAX. */
+
+#include "riscv_vector.h"
+
+void foo2 (void * restrict in, void * restrict out)
+{
+ vuint16mf4_t v1 = *(vuint16mf4_t*)(in + 1);
+ vuint16mf4_t v2 = *(vuint16mf4_t*)(in + 2);
+ vuint16mf4_t v3 = *(vuint16mf4_t*)(in + 3);
+ vuint16mf4_t v4 = *(vuint16mf4_t*)(in + 4);
+ vuint16mf4_t v5 = *(vuint16mf4_t*)(in + 5);
+ vuint16mf4_t v6 = *(vuint16mf4_t*)(in + 6);
+ vuint16mf4_t v7 = *(vuint16mf4_t*)(in + 7);
+ vuint16mf4_t v8 = *(vuint16mf4_t*)(in + 8);
+ *(vuint16mf4_t*)(out + 1) = v1;
+ *(vuint16mf4_t*)(out + 2) = v2;
+ *(vuint16mf4_t*)(out + 3) = v3;
+ *(vuint16mf4_t*)(out + 4) = v4;
+ *(vuint16mf4_t*)(out + 5) = v5;
+ *(vuint16mf4_t*)(out + 6) = v6;
+ *(vuint16mf4_t*)(out + 7) = v7;
+ *(vuint16mf4_t*)(out + 8) = v8;
+}
+
+void foo3 (void * restrict in, void * restrict out)
+{
+ vuint16mf2_t v1 = *(vuint16mf2_t*)(in + 1);
+ vuint16mf2_t v2 = *(vuint16mf2_t*)(in + 2);
+ vuint16mf2_t v3 = *(vuint16mf2_t*)(in + 3);
+ vuint16mf2_t v4 = *(vuint16mf2_t*)(in + 4);
+ vuint16mf2_t v5 = *(vuint16mf2_t*)(in + 5);
+ vuint16mf2_t v6 = *(vuint16mf2_t*)(in + 6);
+ vuint16mf2_t v7 = *(vuint16mf2_t*)(in + 7);
+ vuint16mf2_t v8 = *(vuint16mf2_t*)(in + 8);
+ *(vuint16mf2_t*)(out + 1) = v1;
+ *(vuint16mf2_t*)(out + 2) = v2;
+ *(vuint16mf2_t*)(out + 3) = v3;
+ *(vuint16mf2_t*)(out + 4) = v4;
+ *(vuint16mf2_t*)(out + 5) = v5;
+ *(vuint16mf2_t*)(out + 6) = v6;
+ *(vuint16mf2_t*)(out + 7) = v7;
+ *(vuint16mf2_t*)(out + 8) = v8;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-6.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-6.c
new file mode 100644
index 00000000000..f75449e7345
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-6.c
@@ -0,0 +1,28 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+/* Test insert-vsetvl PASS whether it's able to eliminate vsetvl for same vtype in VLMAX. */
+
+#include "riscv_vector.h"
+
+void foo3 (void * restrict in, void * restrict out)
+{
+ vint32mf2_t v1 = *(vint32mf2_t*)(in + 1);
+ vint32mf2_t v2 = *(vint32mf2_t*)(in + 2);
+ vint32mf2_t v3 = *(vint32mf2_t*)(in + 3);
+ vint32mf2_t v4 = *(vint32mf2_t*)(in + 4);
+ vint32mf2_t v5 = *(vint32mf2_t*)(in + 5);
+ vint32mf2_t v6 = *(vint32mf2_t*)(in + 6);
+ vint32mf2_t v7 = *(vint32mf2_t*)(in + 7);
+ vint32mf2_t v8 = *(vint32mf2_t*)(in + 8);
+ *(vint32mf2_t*)(out + 1) = v1;
+ *(vint32mf2_t*)(out + 2) = v2;
+ *(vint32mf2_t*)(out + 3) = v3;
+ *(vint32mf2_t*)(out + 4) = v4;
+ *(vint32mf2_t*)(out + 5) = v5;
+ *(vint32mf2_t*)(out + 6) = v6;
+ *(vint32mf2_t*)(out + 7) = v7;
+ *(vint32mf2_t*)(out + 8) = v8;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-7.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-7.c
new file mode 100644
index 00000000000..08017e24ef3
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-7.c
@@ -0,0 +1,28 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+/* Test insert-vsetvl PASS whether it's able to eliminate vsetvl for same vtype in VLMAX. */
+
+#include "riscv_vector.h"
+
+void foo3 (void * restrict in, void * restrict out)
+{
+ vuint32mf2_t v1 = *(vuint32mf2_t*)(in + 1);
+ vuint32mf2_t v2 = *(vuint32mf2_t*)(in + 2);
+ vuint32mf2_t v3 = *(vuint32mf2_t*)(in + 3);
+ vuint32mf2_t v4 = *(vuint32mf2_t*)(in + 4);
+ vuint32mf2_t v5 = *(vuint32mf2_t*)(in + 5);
+ vuint32mf2_t v6 = *(vuint32mf2_t*)(in + 6);
+ vuint32mf2_t v7 = *(vuint32mf2_t*)(in + 7);
+ vuint32mf2_t v8 = *(vuint32mf2_t*)(in + 8);
+ *(vuint32mf2_t*)(out + 1) = v1;
+ *(vuint32mf2_t*)(out + 2) = v2;
+ *(vuint32mf2_t*)(out + 3) = v3;
+ *(vuint32mf2_t*)(out + 4) = v4;
+ *(vuint32mf2_t*)(out + 5) = v5;
+ *(vuint32mf2_t*)(out + 6) = v6;
+ *(vuint32mf2_t*)(out + 7) = v7;
+ *(vuint32mf2_t*)(out + 8) = v8;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-8.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-8.c
new file mode 100644
index 00000000000..f03cc5b1b2b
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-8.c
@@ -0,0 +1,28 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+/* Test insert-vsetvl PASS whether it's able to eliminate vsetvl for same vtype in VLMAX. */
+
+#include "riscv_vector.h"
+
+void foo3 (void * restrict in, void * restrict out)
+{
+ vfloat32mf2_t v1 = *(vfloat32mf2_t*)(in + 1);
+ vfloat32mf2_t v2 = *(vfloat32mf2_t*)(in + 2);
+ vfloat32mf2_t v3 = *(vfloat32mf2_t*)(in + 3);
+ vfloat32mf2_t v4 = *(vfloat32mf2_t*)(in + 4);
+ vfloat32mf2_t v5 = *(vfloat32mf2_t*)(in + 5);
+ vfloat32mf2_t v6 = *(vfloat32mf2_t*)(in + 6);
+ vfloat32mf2_t v7 = *(vfloat32mf2_t*)(in + 7);
+ vfloat32mf2_t v8 = *(vfloat32mf2_t*)(in + 8);
+ *(vfloat32mf2_t*)(out + 1) = v1;
+ *(vfloat32mf2_t*)(out + 2) = v2;
+ *(vfloat32mf2_t*)(out + 3) = v3;
+ *(vfloat32mf2_t*)(out + 4) = v4;
+ *(vfloat32mf2_t*)(out + 5) = v5;
+ *(vfloat32mf2_t*)(out + 6) = v6;
+ *(vfloat32mf2_t*)(out + 7) = v7;
+ *(vfloat32mf2_t*)(out + 8) = v8;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-9.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-9.c
new file mode 100644
index 00000000000..e4a734ce616
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_block-9.c
@@ -0,0 +1,147 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+void f1 (void * restrict in, void * restrict out)
+{
+ vint8mf8_t v1 = *(vint8mf8_t*)(in + 1);
+ vint16mf4_t v2;
+ *(vint8mf8_t*)(out + 1) = v1;
+ *(vint16mf4_t*)(out + 2) = v2;
+}
+
+void f2 (void * restrict in, void * restrict out)
+{
+ vint8mf8_t v1 = *(vint8mf8_t*)(in + 1);
+ vint32mf2_t v2;
+ *(vint8mf8_t*)(out + 1) = v1;
+ *(vint32mf2_t*)(out + 2) = v2;
+}
+
+void f3 (void * restrict in, void * restrict out)
+{
+ vint8mf8_t v1 = *(vint8mf8_t*)(in + 1);
+ vint64m1_t v2;
+ *(vint8mf8_t*)(out + 1) = v1;
+ *(vint64m1_t*)(out + 2) = v2;
+}
+
+void f4 (void * restrict in, void * restrict out)
+{
+ vint8mf4_t v1 = *(vint8mf4_t*)(in + 1);
+ vint16mf2_t v2;
+ *(vint8mf4_t*)(out + 1) = v1;
+ *(vint16mf2_t*)(out + 2) = v2;
+}
+
+void f5 (void * restrict in, void * restrict out)
+{
+ vint8mf4_t v1 = *(vint8mf4_t*)(in + 1);
+ vint32m1_t v2;
+ *(vint8mf4_t*)(out + 1) = v1;
+ *(vint32m1_t*)(out + 2) = v2;
+}
+
+void f6 (void * restrict in, void * restrict out)
+{
+ vint8mf4_t v1 = *(vint8mf4_t*)(in + 1);
+ vint64m2_t v2;
+ *(vint8mf4_t*)(out + 1) = v1;
+ *(vint64m2_t*)(out + 2) = v2;
+}
+
+void f7 (void * restrict in, void * restrict out)
+{
+ vint8mf2_t v1 = *(vint8mf2_t*)(in + 1);
+ vint16m1_t v2;
+ *(vint8mf2_t*)(out + 1) = v1;
+ *(vint16m1_t*)(out + 2) = v2;
+}
+
+void f8 (void * restrict in, void * restrict out)
+{
+ vint8mf2_t v1 = *(vint8mf2_t*)(in + 1);
+ vint32m2_t v2;
+ *(vint8mf2_t*)(out + 1) = v1;
+ *(vint32m2_t*)(out + 2) = v2;
+}
+
+void f9 (void * restrict in, void * restrict out)
+{
+ vint8mf2_t v1 = *(vint8mf2_t*)(in + 1);
+ vint64m4_t v2;
+ *(vint8mf2_t*)(out + 1) = v1;
+ *(vint64m4_t*)(out + 2) = v2;
+}
+
+void f10 (void * restrict in, void * restrict out)
+{
+ vint8m1_t v1 = *(vint8m1_t*)(in + 1);
+ vint16m2_t v2;
+ *(vint8m1_t*)(out + 1) = v1;
+ *(vint16m2_t*)(out + 2) = v2;
+}
+
+void f11 (void * restrict in, void * restrict out)
+{
+ vint8m1_t v1 = *(vint8m1_t*)(in + 1);
+ vint32m4_t v2;
+ *(vint8m1_t*)(out + 1) = v1;
+ *(vint32m4_t*)(out + 2) = v2;
+}
+
+void f12 (void * restrict in, void * restrict out)
+{
+ vint8m1_t v1 = *(vint8m1_t*)(in + 1);
+ vint64m8_t v2;
+ *(vint8m1_t*)(out + 1) = v1;
+ *(vint64m8_t*)(out + 2) = v2;
+}
+
+void f13 (void * restrict in, void * restrict out)
+{
+ vint8m2_t v1 = *(vint8m2_t*)(in + 1);
+ vint16m4_t v2;
+ *(vint8m2_t*)(out + 1) = v1;
+ *(vint16m4_t*)(out + 2) = v2;
+}
+
+void f14 (void * restrict in, void * restrict out)
+{
+ vint8m2_t v1 = *(vint8m2_t*)(in + 1);
+ vint32m8_t v2;
+ *(vint8m2_t*)(out + 1) = v1;
+ *(vint32m8_t*)(out + 2) = v2;
+}
+
+void f15 (void * restrict in, void * restrict out)
+{
+ vint8m4_t v1 = *(vint8m4_t*)(in + 1);
+ vint16m8_t v2;
+ *(vint8m4_t*)(out + 1) = v1;
+ *(vint16m8_t*)(out + 2) = v2;
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*m2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e64,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*m4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli} 15 { target { no-opts "-O0" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-1.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-1.c
new file mode 100644
index 00000000000..6f0da3a1581
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-1.c
@@ -0,0 +1,86 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32 -O3" } */
+
+#include "riscv_vector.h"
+
+/* Check vsetvl instruction is hoisted outside the loop, so it should
+ stay before label. */
+
+void foo1 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vbool64_t v = *(vbool64_t*)(in + i);
+ *(vbool64_t*)(out + i) = v;
+ }
+}
+
+void foo2 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vbool32_t v = *(vbool32_t*)(in + i);
+ *(vbool32_t*)(out + i) = v;
+ }
+}
+
+void foo3 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vbool16_t v = *(vbool16_t*)(in + i);
+ *(vbool16_t*)(out + i) = v;
+ }
+}
+
+void foo4 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vbool8_t v = *(vbool8_t*)(in + i);
+ *(vbool8_t*)(out + i) = v;
+ }
+}
+
+void foo5 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vbool4_t v = *(vbool4_t*)(in + i);
+ *(vbool4_t*)(out + i) = v;
+ }
+}
+
+void foo6 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vbool2_t v = *(vbool2_t*)(in + i);
+ *(vbool2_t*)(out + i) = v;
+ }
+}
+
+void foo7 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vbool1_t v = *(vbool1_t*)(in + i);
+ *(vbool1_t*)(out + i) = v;
+ }
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf8,\s*t[au],\s*m[au]\s+\.L[0-9]\:\s+vlm\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf4,\s*t[au],\s*m[au]\s+\.L[0-9]\:\s+vlm\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf2,\s*t[au],\s*m[au]\s+\.L[0-9][0-9]\:\s+vlm\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m1,\s*t[au],\s*m[au]\s+\.L[0-9][0-9]\:\s+vlm\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m2,\s*t[au],\s*m[au]\s+\.L[0-9][0-9]\:\s+vlm\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m4,\s*t[au],\s*m[au]\s+\.L[0-9][0-9]\:\s+vlm\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m8,\s*t[au],\s*m[au]\s+\.L[0-9][0-9]\:\s+vlm\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m1,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*m8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-2.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-2.c
new file mode 100644
index 00000000000..b2dcda61d3d
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-2.c
@@ -0,0 +1,42 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+/* Check vsetvl instruction is hoisted outside the loop, so it should
+ stay before label. */
+
+void foo1 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vint8mf8_t v = *(vint8mf8_t*)(in + i);
+ *(vint8mf8_t*)(out + i) = v;
+ }
+}
+
+void foo2 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vint8mf4_t v = *(vint8mf4_t*)(in + i);
+ *(vint8mf4_t*)(out + i) = v;
+ }
+}
+
+void foo3 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vint8mf2_t v = *(vint8mf2_t*)(in + i);
+ *(vint8mf2_t*)(out + i) = v;
+ }
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf8,\s*t[au],\s*m[au]\s+\.L[0-9]\:\s+vle8\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf4,\s*t[au],\s*m[au]\s+\.L[0-9]\:\s+vle8\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf2,\s*t[au],\s*m[au]\s+\.L[0-9][0-9]\:\s+vle8\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf8,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-3.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-3.c
new file mode 100644
index 00000000000..28215d7b033
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-3.c
@@ -0,0 +1,38 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+/* Check vsetvl instruction is hoisted outside the loop, so it should
+ stay before label. */
+
+void foo1 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vuint8mf8_t v = *(vuint8mf8_t*)(in + i);
+ *(vuint8mf8_t*)(out + i) = v;
+ }
+}
+
+void foo2 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vuint8mf4_t v = *(vuint8mf4_t*)(in + i);
+ *(vuint8mf4_t*)(out + i) = v;
+ }
+}
+
+void foo3 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vuint8mf2_t v = *(vuint8mf2_t*)(in + i);
+ *(vuint8mf2_t*)(out + i) = v;
+ }
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf8,\s*t[au],\s*m[au]\s+\.L[0-9]\:\s+vle8\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf4,\s*t[au],\s*m[au]\s+\.L[0-9]\:\s+vle8\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e8,\s*mf2,\s*t[au],\s*m[au]\s+\.L[0-9][0-9]\:\s+vle8\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-4.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-4.c
new file mode 100644
index 00000000000..6c485369ca5
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-4.c
@@ -0,0 +1,31 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+/* Check vsetvl instruction is hoisted outside the loop, so it should
+ stay before label. */
+
+void foo2 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vint16mf4_t v = *(vint16mf4_t*)(in + i);
+ *(vint16mf4_t*)(out + i) = v;
+ }
+}
+
+void foo3 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vint16mf2_t v = *(vint16mf2_t*)(in + i);
+ *(vint16mf2_t*)(out + i) = v;
+ }
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf2,\s*t[au],\s*m[au]\s+\.L[0-9]\:\s+vle16\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf4,\s*t[au],\s*m[au]\s+\.L[0-9]\:\s+vle16\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-5.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-5.c
new file mode 100644
index 00000000000..d87c4fb075e
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-5.c
@@ -0,0 +1,31 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+/* Check vsetvl instruction is hoisted outside the loop, so it should
+ stay before label. */
+
+void foo2 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vuint16mf4_t v = *(vuint16mf4_t*)(in + i);
+ *(vuint16mf4_t*)(out + i) = v;
+ }
+}
+
+void foo3 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vuint16mf2_t v = *(vuint16mf2_t*)(in + i);
+ *(vuint16mf2_t*)(out + i) = v;
+ }
+}
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf4,\s*t[au],\s*m[au]\s+\.L[0-9]\:\s+vle16\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf2,\s*t[au],\s*m[au]\s+\.L[0-9]\:\s+vle16\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf4,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-6.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-6.c
new file mode 100644
index 00000000000..d8b00b5fce8
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-6.c
@@ -0,0 +1,18 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+/* Check vsetvl instruction is hoisted outside the loop, so it should
+ stay before label. */
+
+void foo2 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vint32mf2_t v = *(vint32mf2_t*)(in + i);
+ *(vint32mf2_t*)(out + i) = v;
+ }
+}
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*mf2,\s*t[au],\s*m[au]\s+\.L[0-9]\:\s+vle32\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-7.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-7.c
new file mode 100644
index 00000000000..602062c54ae
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-7.c
@@ -0,0 +1,18 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+/* Check vsetvl instruction is hoisted outside the loop, so it should
+ stay before label. */
+
+void foo2 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vuint32mf2_t v = *(vuint32mf2_t*)(in + i);
+ *(vuint32mf2_t*)(out + i) = v;
+ }
+}
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*mf2,\s*t[au],\s*m[au]\s+\.L[0-9]\:\s+vle32\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-8.c b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-8.c
new file mode 100644
index 00000000000..53659787802
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-8.c
@@ -0,0 +1,18 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32" } */
+
+#include "riscv_vector.h"
+
+/* Check vsetvl instruction is hoisted outside the loop, so it should
+ stay before label. */
+
+void foo2 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vfloat32mf2_t v = *(vfloat32mf2_t*)(in + i);
+ *(vfloat32mf2_t*)(out + i) = v;
+ }
+}
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*mf2,\s*t[au],\s*m[au]\s+\.L[0-9]\:\s+vle32\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\)} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
+/* { dg-final { scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e32,\s*mf2,\s*t[au],\s*m[au]} 1 { target { no-opts "-O0" no-opts "-Os" no-opts "-funroll-loops" no-opts "-g" } } } } */
--
2.36.3
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] RISC-V: Add testcases for VSETVL PASS
2022-12-14 8:09 [PATCH] RISC-V: Add testcases for VSETVL PASS juzhe.zhong
@ 2022-12-16 20:07 ` Jeff Law
2022-12-17 1:31 ` 钟居哲
2022-12-26 9:20 ` Andreas Schwab
1 sibling, 1 reply; 9+ messages in thread
From: Jeff Law @ 2022-12-16 20:07 UTC (permalink / raw)
To: juzhe.zhong, gcc-patches; +Cc: kito.cheng, palmer
On 12/14/22 01:09, juzhe.zhong@rivai.ai wrote:
> From: Ju-Zhe Zhong <juzhe.zhong@rivai.ai>
>
> gcc/testsuite/ChangeLog:
>
> * gcc.target/riscv/rvv/rvv.exp: Adjust to enable tests for VSETVL PASS.
> * gcc.target/riscv/rvv/vsetvl/dump-1.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-1.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-10.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-11.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-12.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-13.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-14.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-15.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-16.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-17.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-18.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-19.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-2.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-3.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-4.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-5.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-6.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-7.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-8.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-9.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-1.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-2.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-3.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-4.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-5.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-6.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-7.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-8.c: New test.
So it looks like the assembler strings you're searching for are highly
specific (across all 5 testsuite patches). How sensitive do we expect
these tests to be to things like register allocation giving us different
registers and such? I'd hate to be in a position where we're constantly
needing to update these tests because the output is changing in
unimportant ways.
Jeff
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: Re: [PATCH] RISC-V: Add testcases for VSETVL PASS
2022-12-16 20:07 ` Jeff Law
@ 2022-12-17 1:31 ` 钟居哲
2022-12-17 1:42 ` Jeff Law
0 siblings, 1 reply; 9+ messages in thread
From: 钟居哲 @ 2022-12-17 1:31 UTC (permalink / raw)
To: Jeff Law, gcc-patches; +Cc: kito.cheng, palmer
[-- Attachment #1: Type: text/plain, Size: 4341 bytes --]
Register allocation (RA) doesn't affect the assembler checks since I relax the registers in assmebler checks,
all assmebler checks have their own goal. For example:
The code like this:
+void foo2 (void * restrict in, void * restrict out, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ vuint16mf4_t v = *(vuint16mf4_t*)(in + i);
+ *(vuint16mf4_t*)(out + i) = v;
+ }
+}
Assembler check:
scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf4,\s*t[au],\s*m[au]\s+\.L[0-9]\:\s+vle16\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\I don't care about which vector register is using since I relax register in assembler : (?:v[0-9]|v[1-2][0-9]|v3[0-1]), this means any vector register v0-v31But also I relax scalar register : (?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]), so could be any x0 - x31 of them.The only strict check is that make sure the vsetvl is hoist outside the loop meaning the location of vsetvl is outside of the Lable L[0-9]:vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf4,\s*t[au],\s*m[au]\s+\.L[0-9]You can see the last assembler is \s+\.L[0-9] to make sure VSETVL PASS successfully do the optimization that hoist the vsetvl instruction outside the loopI try to use check-function-body but it fails since it can not recognize the Lable which is most important for such cases.
juzhe.zhong@rivai.ai
From: Jeff Law
Date: 2022-12-17 04:07
To: juzhe.zhong; gcc-patches
CC: kito.cheng; palmer
Subject: Re: [PATCH] RISC-V: Add testcases for VSETVL PASS
On 12/14/22 01:09, juzhe.zhong@rivai.ai wrote:
> From: Ju-Zhe Zhong <juzhe.zhong@rivai.ai>
>
> gcc/testsuite/ChangeLog:
>
> * gcc.target/riscv/rvv/rvv.exp: Adjust to enable tests for VSETVL PASS.
> * gcc.target/riscv/rvv/vsetvl/dump-1.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-1.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-10.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-11.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-12.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-13.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-14.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-15.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-16.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-17.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-18.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-19.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-2.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-3.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-4.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-5.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-6.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-7.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-8.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_block-9.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-1.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-2.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-3.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-4.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-5.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-6.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-7.c: New test.
> * gcc.target/riscv/rvv/vsetvl/vlmax_single_vtype-8.c: New test.
So it looks like the assembler strings you're searching for are highly
specific (across all 5 testsuite patches). How sensitive do we expect
these tests to be to things like register allocation giving us different
registers and such? I'd hate to be in a position where we're constantly
needing to update these tests because the output is changing in
unimportant ways.
Jeff
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] RISC-V: Add testcases for VSETVL PASS
2022-12-17 1:31 ` 钟居哲
@ 2022-12-17 1:42 ` Jeff Law
2022-12-19 3:22 ` Kito Cheng
0 siblings, 1 reply; 9+ messages in thread
From: Jeff Law @ 2022-12-17 1:42 UTC (permalink / raw)
To: 钟居哲, gcc-patches; +Cc: kito.cheng, palmer
On 12/16/22 18:31, 钟居哲 wrote:
> Register allocation (RA) doesn't affect the assembler checks since I
> relax the registers in assmebler checks,
> all assmebler checks have their own goal. For example:
>
> The code like this:
>
> +void foo2 (void * restrict in, void * restrict out, int n)
> +{
> + for (int i = 0; i < n; i++)
> + {
> + vuint16mf4_t v = *(vuint16mf4_t*)(in + i);
> + *(vuint16mf4_t*)(out + i) = v;
> + }
> +}
>
> Assembler check:
>
> scan-assembler-times {vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf4,\s*t[au],\s*m[au]\s+\.L[0-9]\:\s+vle16\.v\s+(?:v[0-9]|v[1-2][0-9]|v3[0-1]),0\s*\((?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7])\
>
> I don't care about which vector register is using since I relax register in assembler : (?:v[0-9]|v[1-2][0-9]|v3[0-1]), this means any vector register v0-v31
>
> But also I relax scalar register : (?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]), so could be any x0 - x31 of them.
>
> The only strict check is that make sure the vsetvl is hoist outside the loop meaning the location of vsetvl is outside of the Lable L[0-9]:
>
> vsetvli\s+(?:ra|[sgtf]p|t[0-6]|s[0-9]|s10|s11|a[0-7]),\s*zero,\s*e16,\s*mf4,\s*t[au],\s*m[au]\s+\.L[0-9]
>
> You can see the last assembler is \s+\.L[0-9] to make sure VSETVL PASS successfully do the optimization that hoist the vsetvl instruction outside the loop
>
> I try to use check-function-body but it fails since it can not recognize the Lable which is most important for such cases.
Ah, I should have looked at those regexps closer. Understood about the
checking for hoisting the vsetvl. Though it makes me wonder if we'd be
better off dumping information out of the vsetvl pass.
In the case of hoisting we could dump the loop nest of the original
evaluation block and the loop nest of the new vsetvl location, then we
scan for those in the vsetvl pass dump. While it doesn't check the
assembly code, it's probably just as good if not better.
Consider that as an alternative. But I'm not going to insist on it. I
just know we've had a lot of trouble through the years where assembly
code changes slightly, causing test fails. So I try to avoid too much
assembly scanning if it can be avoided. Often the easiest way to get
the same basic effect is to dump at the transformation point and scan
for those markers in the dump file.
Jeff
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] RISC-V: Add testcases for VSETVL PASS
2022-12-17 1:42 ` Jeff Law
@ 2022-12-19 3:22 ` Kito Cheng
0 siblings, 0 replies; 9+ messages in thread
From: Kito Cheng @ 2022-12-19 3:22 UTC (permalink / raw)
To: Jeff Law; +Cc: 钟居哲, gcc-patches, palmer
Hi Jeff:
> Ah, I should have looked at those regexps closer. Understood about the
> checking for hoisting the vsetvl. Though it makes me wonder if we'd be
> better off dumping information out of the vsetvl pass.
I've discussed adding an extra verify pass and a kind of rating
mechnish to vsetvli pass with Ju-Zhe,
so that we don't need to do such asm scan, which is relatively fragile
to broken.
And we'll implement that in the next few weeks :)
Thanks
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] RISC-V: Add testcases for VSETVL PASS
2022-12-14 8:09 [PATCH] RISC-V: Add testcases for VSETVL PASS juzhe.zhong
2022-12-16 20:07 ` Jeff Law
@ 2022-12-26 9:20 ` Andreas Schwab
2022-12-26 9:28 ` 钟居哲
2022-12-27 15:04 ` Jeff Law
1 sibling, 2 replies; 9+ messages in thread
From: Andreas Schwab @ 2022-12-26 9:20 UTC (permalink / raw)
To: juzhe.zhong; +Cc: gcc-patches, kito.cheng, palmer
FAIL: gcc.target/riscv/rvv/vsetvl/dump-1.c -O0 (test for excess errors)
Excess errors:
/usr/include/gnu/stubs.h:8:11: fatal error: gnu/stubs-ilp32.h: No such file or directory
compilation terminated.
--
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510 2552 DF73 E780 A9DA AEC1
"And now for something completely different."
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: Re: [PATCH] RISC-V: Add testcases for VSETVL PASS
2022-12-26 9:20 ` Andreas Schwab
@ 2022-12-26 9:28 ` 钟居哲
2022-12-27 15:04 ` Jeff Law
1 sibling, 0 replies; 9+ messages in thread
From: 钟居哲 @ 2022-12-26 9:28 UTC (permalink / raw)
To: Andreas Schwab; +Cc: gcc-patches, kito.cheng, palmer
[-- Attachment #1: Type: text/plain, Size: 840 bytes --]
This is another issue and I have no idea. I think palmer or kito may have know how to solve it.
It seems this patch https://gcc.gnu.org/pipermail/gcc-patches/2022-December/609045.html
fixed previous issue which is more important. I think it's time to merge it.
juzhe.zhong@rivai.ai
From: Andreas Schwab
Date: 2022-12-26 17:20
To: juzhe.zhong
CC: gcc-patches; kito.cheng; palmer
Subject: Re: [PATCH] RISC-V: Add testcases for VSETVL PASS
FAIL: gcc.target/riscv/rvv/vsetvl/dump-1.c -O0 (test for excess errors)
Excess errors:
/usr/include/gnu/stubs.h:8:11: fatal error: gnu/stubs-ilp32.h: No such file or directory
compilation terminated.
--
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510 2552 DF73 E780 A9DA AEC1
"And now for something completely different."
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] RISC-V: Add testcases for VSETVL PASS
2022-12-26 9:20 ` Andreas Schwab
2022-12-26 9:28 ` 钟居哲
@ 2022-12-27 15:04 ` Jeff Law
2022-12-27 15:12 ` Kito Cheng
1 sibling, 1 reply; 9+ messages in thread
From: Jeff Law @ 2022-12-27 15:04 UTC (permalink / raw)
To: Andreas Schwab, juzhe.zhong; +Cc: gcc-patches, kito.cheng, palmer
On 12/26/22 02:20, Andreas Schwab wrote:
> FAIL: gcc.target/riscv/rvv/vsetvl/dump-1.c -O0 (test for excess errors)
> Excess errors:
> /usr/include/gnu/stubs.h:8:11: fatal error: gnu/stubs-ilp32.h: No such file or directory
> compilation terminated.
Isn't this really an indicator that the ILP32 bits from glibc aren't
installed and/or the test needs to be conditionalized?
jeff
>
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] RISC-V: Add testcases for VSETVL PASS
2022-12-27 15:04 ` Jeff Law
@ 2022-12-27 15:12 ` Kito Cheng
0 siblings, 0 replies; 9+ messages in thread
From: Kito Cheng @ 2022-12-27 15:12 UTC (permalink / raw)
To: Jeff Law; +Cc: Andreas Schwab, juzhe.zhong, gcc-patches, palmer
It should be same issue as
https://github.com/gcc-mirror/gcc/commit/d0bbecb1c418b680505faa998fe420f0fd4bbfc1,
I gonna commit a fix for that.
On Tue, Dec 27, 2022 at 11:04 PM Jeff Law via Gcc-patches
<gcc-patches@gcc.gnu.org> wrote:
>
>
>
> On 12/26/22 02:20, Andreas Schwab wrote:
> > FAIL: gcc.target/riscv/rvv/vsetvl/dump-1.c -O0 (test for excess errors)
> > Excess errors:
> > /usr/include/gnu/stubs.h:8:11: fatal error: gnu/stubs-ilp32.h: No such file or directory
> > compilation terminated.
> Isn't this really an indicator that the ILP32 bits from glibc aren't
> installed and/or the test needs to be conditionalized?
>
> jeff
> >
^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2022-12-27 15:12 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-12-14 8:09 [PATCH] RISC-V: Add testcases for VSETVL PASS juzhe.zhong
2022-12-16 20:07 ` Jeff Law
2022-12-17 1:31 ` 钟居哲
2022-12-17 1:42 ` Jeff Law
2022-12-19 3:22 ` Kito Cheng
2022-12-26 9:20 ` Andreas Schwab
2022-12-26 9:28 ` 钟居哲
2022-12-27 15:04 ` Jeff Law
2022-12-27 15:12 ` Kito Cheng
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).