Re: [PATCH 2/2] rs6000: Add tests for SSE4.1 "ceil" intrinsics

Bill Schmidt via Gcc-patches Sun, 11 Jul 2021 10:35:07 -0700

Hi Paul,

On 7/1/21 9:11 PM, Paul A. Clarke via Gcc-patches wrote:

Add the tests for _mm_ceil_pd, _mm_ceil_ps, _mm_ceil_sd, _mm_ceil_ss.


Copy a test for _mm_ceil_pd and _mm_ceil_ps from
gcc/testsuite/gcc.target/i386.

Define __VSX_SSE2__ to pick up some union definitons in

typo ("definitions").

m128-check.h.

2021-07-01  Paul A. Clarke  <p...@us.ibm.com>

gcc/testsuite/ChangeLog:

"gcc/testsuite/" will make the tools happy.

        * gcc/testsuite/gcc.target/powerpc/sse4_1-ceilpd.c: New.


All of these should be relative to gcc/testsuite/, so

    * gcc.target/powerpc/sse4_1-ceilpd.c: New.

and similar.

        * gcc/testsuite/gcc.target/powerpc/sse4_1-ceilps.c: New.
        * gcc/testsuite/gcc.target/powerpc/sse4_1-ceilsd.c: New.
        * gcc/testsuite/gcc.target/powerpc/sse4_1-ceilss.c: New.
        * gcc/testsuite/gcc.target/powerpc/sse4_1-round-data.h: New.
        * gcc/testsuite/gcc.target/powerpc/sse4_1-round.h: New.
        * gcc/testsuite/gcc.target/powerpc/sse4_1-round2.h: New.
        * gcc/testsuite/gcc.target/powerpc/sse4_1-roundpd-3.c: Copy
         from gcc/testsuite/gcc.target/i386.
        * gcc/testsuite/gcc.target/powerpc/sse4_1-check.h
        (__VSX_SSE2__): Define.
---
  .../gcc.target/powerpc/sse4_1-ceilpd.c        |  51 ++++++++
  .../gcc.target/powerpc/sse4_1-ceilps.c        |  33 +++++
  .../gcc.target/powerpc/sse4_1-ceilsd.c        | 119 ++++++++++++++++++
  .../gcc.target/powerpc/sse4_1-ceilss.c        |  95 ++++++++++++++
  .../gcc.target/powerpc/sse4_1-check.h         |   4 +
  .../gcc.target/powerpc/sse4_1-round-data.h    |  20 +++
  .../gcc.target/powerpc/sse4_1-round.h         |  27 ++++
  .../gcc.target/powerpc/sse4_1-round2.h        |  27 ++++
  .../gcc.target/powerpc/sse4_1-roundpd-3.c     |  36 ++++++
  9 files changed, 412 insertions(+)
  create mode 100644 gcc/testsuite/gcc.target/powerpc/sse4_1-ceilpd.c
  create mode 100644 gcc/testsuite/gcc.target/powerpc/sse4_1-ceilps.c
  create mode 100644 gcc/testsuite/gcc.target/powerpc/sse4_1-ceilsd.c
  create mode 100644 gcc/testsuite/gcc.target/powerpc/sse4_1-ceilss.c
  create mode 100644 gcc/testsuite/gcc.target/powerpc/sse4_1-round-data.h
  create mode 100644 gcc/testsuite/gcc.target/powerpc/sse4_1-round.h
  create mode 100644 gcc/testsuite/gcc.target/powerpc/sse4_1-round2.h
  create mode 100644 gcc/testsuite/gcc.target/powerpc/sse4_1-roundpd-3.c

diff --git a/gcc/testsuite/gcc.target/powerpc/sse4_1-ceilpd.c 
b/gcc/testsuite/gcc.target/powerpc/sse4_1-ceilpd.c
new file mode 100644
index 000000000000..f532fdb9c285
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/sse4_1-ceilpd.c
@@ -0,0 +1,51 @@
+/* { dg-do run } */
+/* { dg-require-effective-target p8vector_hw } */
+/* { dg-options "-O2 -mpower8-vector -Wno-psabi" } */
+
+#define NO_WARN_X86_INTRINSICS 1
+#include <smmintrin.h>
+
+#define VEC_T __m128d
+#define FP_T double
+
+#define ROUND_INTRIN(x, mode) _mm_ceil_pd (x)
+
+#include "sse4_1-round-data.h"
+
+static struct data data[] = {
+  { .value = { .f = {  0.00,  0.25 } }, .answer = {  0.0,  1.0 } },
+  { .value = { .f = {  0.50,  0.75 } }, .answer = {  1.0,  1.0 } },
+
+  { { .f = {  0x1.ffffffffffffcp+50,  0x1.ffffffffffffdp+50 } },
+           {  0x1.ffffffffffffcp+50,  0x1.0000000000000p+51 } },
+  { { .f = {  0x1.ffffffffffffep+50,  0x1.fffffffffffffp+50 } },
+           {  0x1.0000000000000p+51,  0x1.0000000000000p+51 } },
+  { { .f = {  0x1.0000000000000p+51,  0x1.0000000000001p+51 } },
+           {  0x1.0000000000000p+51,  0x1.0000000000002p+51 } },
+  { { .f = {  0x1.0000000000002p+51,  0x1.0000000000003p+51 } },
+           {  0x1.0000000000002p+51,  0x1.0000000000004p+51 } },
+
+  { { .f = {  0x1.ffffffffffffep+51,  0x1.fffffffffffffp+51 } },
+           {  0x1.ffffffffffffep+51,  0x1.0000000000000p+52 } },
+  { { .f = {  0x1.0000000000000p+52,  0x1.0000000000001p+52 } },
+           {  0x1.0000000000000p+52,  0x1.0000000000001p+52 } },
+
+  { { .f = { -0x1.0000000000001p+52, -0x1.0000000000000p+52 } },
+           { -0x1.0000000000001p+52, -0x1.0000000000000p+52 } },
+  { { .f = { -0x1.fffffffffffffp+51, -0x1.ffffffffffffep+51 } },
+           { -0x1.ffffffffffffep+51, -0x1.ffffffffffffep+51 } },
+
+  { { .f = { -0x1.0000000000003p+51, -0x1.0000000000002p+51 } },
+           { -0x1.0000000000002p+51, -0x1.0000000000002p+51 } },
+  { { .f = { -0x1.0000000000001p+51, -0x1.0000000000000p+51 } },
+           { -0x1.0000000000000p+51, -0x1.0000000000000p+51 } },
+  { { .f = { -0x1.fffffffffffffp+50, -0x1.ffffffffffffep+50 } },
+           { -0x1.ffffffffffffcp+50, -0x1.ffffffffffffcp+50 } },
+  { { .f = { -0x1.ffffffffffffdp+50, -0x1.ffffffffffffcp+50 } },
+           { -0x1.ffffffffffffcp+50, -0x1.ffffffffffffcp+50 } },
+
+  { { .f = { -1.00, -0.75 } }, { -1.0,  0.0 } },
+  { { .f = { -0.50, -0.25 } }, {  0.0,  0.0 } }
+};
+
+#include "sse4_1-round.h"
diff --git a/gcc/testsuite/gcc.target/powerpc/sse4_1-ceilps.c 
b/gcc/testsuite/gcc.target/powerpc/sse4_1-ceilps.c
new file mode 100644
index 000000000000..417ac76d8aa9
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/sse4_1-ceilps.c
@@ -0,0 +1,33 @@
+/* { dg-do run } */
+/* { dg-require-effective-target p8vector_hw } */
+/* { dg-options "-O2 -mpower8-vector -Wno-psabi" } */
+
+#define NO_WARN_X86_INTRINSICS 1
+#include <smmintrin.h>
+
+#define VEC_T __m128
+#define FP_T float
+
+#define ROUND_INTRIN(x, mode) _mm_ceil_ps (x)
+
+#include "sse4_1-round-data.h"
+
+static struct data data[] = {
+  { { .f = {  0.00,  0.25,  0.50,  0.75 } }, {  0.0,  1.0,  1.0,  1.0 } },
+
+  { { .f = {  0x1.fffff8p+21,  0x1.fffffap+21,  0x1.fffffcp+21,  
0x1.fffffep+21 } },
+           {  0x1.fffff8p+21,  0x1.000000p+22,  0x1.000000p+22,  
0x1.000000p+22 } },
+
+  { { .f = {  0x1.fffffap+22,  0x1.fffffcp+22,  0x1.fffffep+22,  
0x1.fffffep+23 } },
+           {  0x1.fffffcp+22,  0x1.fffffcp+22,  0x1.000000p+23,  
0x1.fffffep+23 } },
+
+  { { .f = { -0x1.fffffep+23, -0x1.fffffep+22, -0x1.fffffcp+22, 
-0x1.fffffap+22 } },
+           { -0x1.fffffep+23, -0x1.fffffcp+22, -0x1.fffffcp+22, 
-0x1.fffff8p+22 } },
+
+  { { .f = { -0x1.fffffep+21, -0x1.fffffcp+21, -0x1.fffffap+21, 
-0x1.fffff8p+21 } },
+           { -0x1.fffff8p+21, -0x1.fffff8p+21, -0x1.fffff8p+21, 
-0x1.fffff8p+21 } },


Above lines are too long, please reformat in pairs.

The rest LGTM.  Thanks for the very thorough test cases!

I can't approve, but recommend approval with the above addressed.  Thanks!
Bill

+
+  { { .f = { -1.00, -0.75, -0.50, -0.25 } }, { -1.0,  0.0,  0.0,  0.0 } }
+};
+
+#include "sse4_1-round.h"
diff --git a/gcc/testsuite/gcc.target/powerpc/sse4_1-ceilsd.c 
b/gcc/testsuite/gcc.target/powerpc/sse4_1-ceilsd.c
new file mode 100644
index 000000000000..cc0d9c1d0afe
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/sse4_1-ceilsd.c
@@ -0,0 +1,119 @@
+/* { dg-do run } */
+/* { dg-require-effective-target p8vector_hw } */
+/* { dg-options "-O2 -mpower8-vector -Wno-psabi" } */
+
+#define NO_WARN_X86_INTRINSICS 1
+#include <smmintrin.h>
+
+#define VEC_T __m128d
+#define FP_T double
+
+#define ROUND_INTRIN(x, y) _mm_ceil_sd (x, y)
+
+#include "sse4_1-round-data.h"
+
+static struct data2 data[] = {
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = {  0.00, IGNORED } },
+    .answer = {  0.0, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = {  0.25, IGNORED } },
+    .answer = {  1.0, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = {  0.50, IGNORED } },
+    .answer = {  1.0, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = {  0.75, IGNORED } },
+    .answer = {  1.0, PASSTHROUGH } },
+
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.ffffffffffffcp+50, IGNORED } },
+    .answer = {  0x1.ffffffffffffcp+50, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.ffffffffffffdp+50, IGNORED } },
+    .answer = {  0x1.0000000000000p+51, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.ffffffffffffep+50, IGNORED } },
+    .answer = {  0x1.0000000000000p+51, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.fffffffffffffp+50, IGNORED } },
+    .answer = {  0x1.0000000000000p+51, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.0000000000000p+51, IGNORED } },
+    .answer = {  0x1.0000000000000p+51, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.0000000000001p+51, IGNORED } },
+    .answer = {  0x1.0000000000002p+51, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.0000000000002p+51, IGNORED } },
+    .answer = {  0x1.0000000000002p+51, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.0000000000003p+51, IGNORED } },
+    .answer = {  0x1.0000000000004p+51, PASSTHROUGH } },
+
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.ffffffffffffep+51, IGNORED } },
+    .answer = {  0x1.ffffffffffffep+51, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.fffffffffffffp+51, IGNORED } },
+    .answer = {  0x1.0000000000000p+52, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.0000000000000p+52, IGNORED } },
+    .answer = {  0x1.0000000000000p+52, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.0000000000001p+52, IGNORED } },
+    .answer = {  0x1.0000000000001p+52, PASSTHROUGH } },
+
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.0000000000001p+52, IGNORED } },
+    .answer = { -0x1.0000000000001p+52, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.0000000000000p+52, IGNORED } },
+    .answer = { -0x1.0000000000000p+52, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.fffffffffffffp+51, IGNORED } },
+    .answer = { -0x1.ffffffffffffep+51, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.ffffffffffffep+51, IGNORED } },
+    .answer = { -0x1.ffffffffffffep+51, PASSTHROUGH } },
+
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.0000000000003p+51, IGNORED } },
+    .answer = { -0x1.0000000000002p+51, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.0000000000002p+51, IGNORED } },
+    .answer = { -0x1.0000000000002p+51, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.0000000000001p+51, IGNORED } },
+    .answer = { -0x1.0000000000000p+51, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.0000000000000p+51, IGNORED } },
+    .answer = { -0x1.0000000000000p+51, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.ffffffffffffcp+50, IGNORED } },
+    .answer = { -0x1.ffffffffffffcp+50, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.ffffffffffffep+50, IGNORED } },
+    .answer = { -0x1.ffffffffffffcp+50, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.ffffffffffffdp+50, IGNORED } },
+    .answer = { -0x1.ffffffffffffcp+50, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.ffffffffffffcp+50, IGNORED } },
+    .answer = { -0x1.ffffffffffffcp+50, PASSTHROUGH } },
+
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = { -1.00, IGNORED } },
+    .answer = { -1.0, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = { -0.75, IGNORED } },
+    .answer = { -0.0, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = { -0.50, IGNORED } },
+    .answer = { -0.0, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH } },
+    .value2 = { .f = { -0.25, IGNORED } },
+    .answer = { -0.0, PASSTHROUGH } }
+};
+
+#include "sse4_1-round2.h"
diff --git a/gcc/testsuite/gcc.target/powerpc/sse4_1-ceilss.c 
b/gcc/testsuite/gcc.target/powerpc/sse4_1-ceilss.c
new file mode 100644
index 000000000000..cf1a0392990e
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/sse4_1-ceilss.c
@@ -0,0 +1,95 @@
+/* { dg-do run } */
+/* { dg-require-effective-target p8vector_hw } */
+/* { dg-options "-O2 -mpower8-vector -Wno-psabi" } */
+
+#define NO_WARN_X86_INTRINSICS 1
+#include <smmintrin.h>
+
+#define VEC_T __m128
+#define FP_T float
+
+#define ROUND_INTRIN(x, y) _mm_ceil_ss (x, y)
+
+#include "sse4_1-round-data.h"
+
+static struct data2 data[] = {
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = {  0.00,  IGNORED, IGNORED, IGNORED } },
+    .answer = {  0.0, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = {  0.25, IGNORED, IGNORED, IGNORED } },
+    .answer = {  1.0, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = {  0.50, IGNORED, IGNORED, IGNORED } },
+    .answer = {  1.0, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = {  0.75, IGNORED, IGNORED, IGNORED } },
+    .answer = {  1.0, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.fffff8p+21, IGNORED, IGNORED, IGNORED } },
+    .answer = {  0x1.fffff8p+21, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.fffffap+21, IGNORED, IGNORED, IGNORED } },
+    .answer = {  0x1.000000p+22, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.fffffcp+21, IGNORED, IGNORED, IGNORED } },
+    .answer = {  0x1.000000p+22, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.fffffep+21, IGNORED, IGNORED, IGNORED } },
+    .answer = {  0x1.000000p+22, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.fffffap+22, IGNORED, IGNORED, IGNORED } },
+    .answer = {  0x1.fffffcp+22, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.fffffcp+22, IGNORED, IGNORED, IGNORED } },
+    .answer = {  0x1.fffffcp+22, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.fffffep+22, IGNORED, IGNORED, IGNORED } },
+    .answer = {  0x1.000000p+23, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = {  0x1.fffffep+23, IGNORED, IGNORED, IGNORED } },
+    .answer = {  0x1.fffffep+23, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.fffffep+23, IGNORED, IGNORED, IGNORED } },
+    .answer = { -0x1.fffffep+23, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.fffffep+22, IGNORED, IGNORED, IGNORED } },
+    .answer = { -0x1.fffffcp+22, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.fffffcp+22, IGNORED, IGNORED, IGNORED } },
+    .answer = { -0x1.fffffcp+22, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.fffffap+22, IGNORED, IGNORED, IGNORED } },
+    .answer = { -0x1.fffff8p+22, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.fffffep+21, IGNORED, IGNORED, IGNORED } },
+    .answer = { -0x1.fffff8p+21, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.fffffcp+21, IGNORED, IGNORED, IGNORED } },
+    .answer = { -0x1.fffff8p+21, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.fffffap+21, IGNORED, IGNORED, IGNORED } },
+    .answer = { -0x1.fffff8p+21, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = { -0x1.fffff8p+21, IGNORED, IGNORED, IGNORED } },
+    .answer = { -0x1.fffff8p+21, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = { -1.00, IGNORED, IGNORED, IGNORED } },
+    .answer = { -1.0, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = { -0.75, IGNORED, IGNORED, IGNORED } },
+    .answer = {  0.0, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = { -0.50, IGNORED, IGNORED, IGNORED } },
+    .answer = {  0.0, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+  { .value1 = { .f = { IGNORED, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } },
+    .value2 = { .f = { -0.25, IGNORED, IGNORED, IGNORED } },
+    .answer = {  0.0, PASSTHROUGH, PASSTHROUGH, PASSTHROUGH } }
+};
+
+#include "sse4_1-round2.h"
diff --git a/gcc/testsuite/gcc.target/powerpc/sse4_1-check.h 
b/gcc/testsuite/gcc.target/powerpc/sse4_1-check.h
index 5f855b9fd53a..16330533e50a 100644
--- a/gcc/testsuite/gcc.target/powerpc/sse4_1-check.h
+++ b/gcc/testsuite/gcc.target/powerpc/sse4_1-check.h
@@ -1,6 +1,10 @@
  #include <stdio.h>
  #include <stdlib.h>

+/* Define this to enable the combination of VSX vector double and
+   SSE2 data types.  */
+#define __VSX_SSE2__ 1
+
  #include "m128-check.h"

  //#define DEBUG 1
diff --git a/gcc/testsuite/gcc.target/powerpc/sse4_1-round-data.h 
b/gcc/testsuite/gcc.target/powerpc/sse4_1-round-data.h
new file mode 100644
index 000000000000..543f5bc2181b
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/sse4_1-round-data.h
@@ -0,0 +1,20 @@
+/* Pick a few numbers at random which are not in the input data and
+   unlikely to show up naturally.  */
+#define PASSTHROUGH -29.5
+#define IGNORED -61.5
+
+union value {
+  VEC_T x;
+  FP_T f[sizeof (VEC_T) / sizeof (FP_T)];
+};
+
+struct data {
+  union value value;
+  double answer[sizeof (VEC_T) / sizeof (FP_T)];
+};
+
+struct data2 {
+  union value value1;
+  union value value2;
+  double answer[sizeof (VEC_T) / sizeof (FP_T)];
+};
diff --git a/gcc/testsuite/gcc.target/powerpc/sse4_1-round.h 
b/gcc/testsuite/gcc.target/powerpc/sse4_1-round.h
new file mode 100644
index 000000000000..6acf8da8b766
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/sse4_1-round.h
@@ -0,0 +1,27 @@
+#include <fenv.h>
+#include <smmintrin.h>
+#include "sse4_1-check.h"
+
+#define DIM(a) (sizeof (a) / sizeof ((a)[0]))
+
+static int modes[] = { FE_TONEAREST, FE_UPWARD, FE_DOWNWARD, FE_TOWARDZERO };
+
+static void
+TEST (void)
+{
+  int i, j, ri, round_save;
+
+  round_save = fegetround ();
+  for (ri = 0; ri < DIM (modes); ri++) {
+    (void) fesetround (modes[ri]);
+    for (i = 0; i < DIM (data); i++) {
+      union value guess;
+      guess.x = ROUND_INTRIN (data[i].value.x, /* Ignored.  */);
+      for (j = 0; j < DIM (data[i].value.f); j++) {
+        if (guess.f[j] != data[i].answer[j])
+          abort ();
+      }
+    }
+  }
+  (void) fesetround (round_save);
+}
diff --git a/gcc/testsuite/gcc.target/powerpc/sse4_1-round2.h 
b/gcc/testsuite/gcc.target/powerpc/sse4_1-round2.h
new file mode 100644
index 000000000000..859574e11d9a
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/sse4_1-round2.h
@@ -0,0 +1,27 @@
+#include <fenv.h>
+#include <smmintrin.h>
+#include "sse4_1-check.h"
+
+#define DIM(a) (sizeof (a) / sizeof ((a)[0]))
+
+static int modes[] = { FE_TONEAREST, FE_UPWARD, FE_DOWNWARD, FE_TOWARDZERO };
+
+static void
+TEST (void)
+{
+  int i, j, ri, round_save;
+
+  round_save = fegetround ();
+  for (ri = 0; ri < DIM (modes); ri++) {
+    (void) fesetround (modes[ri]);
+    for (i = 0; i < DIM (data); i++) {
+      union value guess;
+      guess.x = ROUND_INTRIN (data[i].value1.x, data[i].value2.x);
+      for (j = 0; j < DIM (data[i].value1.f); j++) {
+        if (guess.f[j] != data[i].answer[j])
+          abort ();
+      }
+    }
+  }
+  (void) fesetround (round_save);
+}
diff --git a/gcc/testsuite/gcc.target/powerpc/sse4_1-roundpd-3.c 
b/gcc/testsuite/gcc.target/powerpc/sse4_1-roundpd-3.c
new file mode 100644
index 000000000000..88a5f0718ebb
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/sse4_1-roundpd-3.c
@@ -0,0 +1,36 @@
+/* { dg-do run } */
+/* { dg-require-effective-target p8vector_hw } */
+/* { dg-options "-O2 -mpower8-vector -Wno-psabi" } */
+
+#ifndef CHECK_H
+#define CHECK_H "sse4_1-check.h"
+#endif
+
+#ifndef TEST
+#define TEST sse4_1_test
+#endif
+
+#include CHECK_H
+
+#include <smmintrin.h>
+
+static void
+TEST (void)
+{
+  union128d u, s;
+  double e[2] = {0.0};
+  int i;
+
+  s.x = _mm_set_pd (1.1234, -2.3478);
+  u.x = _mm_ceil_pd (s.x);
+
+  for (i = 0; i < 2; i++)
+    {
+      __m128d tmp = _mm_load_sd (&s.a[i]);
+      tmp = _mm_ceil_sd (tmp, tmp);
+      _mm_store_sd (&e[i], tmp);
+    }
+
+  if (check_union128d (u, e))
+    abort ();
+}

Re: [PATCH 2/2] rs6000: Add tests for SSE4.1 "ceil" intrinsics

Reply via email to