diff mbox

[9/N,v2,x86_64] Vectorized math functions

Message ID CAMXFM3uqG=KhwofMe7w-ToPHL0ur+apMT0sh4W82KYHyXnEEKg@mail.gmail.com
State New
Headers show

Commit Message

Andrew Senkevich Dec. 2, 2014, 5:14 p.m. UTC 
This is addition of tests for vectorized function cos.

2014-12-02  Andrew Senkevich  <andrew.n.senkevich@gmail.com>

        * math/Makefile: Added rules for tests.
        * sysdeps/x86_64/fpu/Makefile: Likewise.
        * math/test-double-vlen2.h: New file.
        * math/test-double-vlen4.h: New file.
        * sysdeps/x86_64/fpu/test-double-vlen2.c: New file.
        * sysdeps/x86_64/fpu/test-double-vlen4-avx2.c: New file.
        * sysdeps/x86_64/fpu/test-double-vlen4.c: New file.
        * sysdeps/x86_64/fpu/math-tests-arch.h: AVX2 availability runtime
        check set up.
        * sysdeps/x86_64/fpu/libm-test-ulps: Regenarated.



--
WBR,
Andrew

Comments

Joseph Myers Dec. 3, 2014, 6:26 p.m. UTC | #1 
On Tue, 2 Dec 2014, Andrew Senkevich wrote:

> +$(objpfx)test-double-vlen2: $(common-objpfx)mathvec/libmvec.so \
> +    $(objpfx)init-arch.o
> +$(objpfx)test-double-vlen4: $(common-objpfx)mathvec/libmvec.so \
> +    $(objpfx)init-arch.o

Depending on $(common-objpfx)mathvec/libmvec.so would break testing 
--disable-shared (it may be broken anyway, but there's no need to add more 
breakage).  It's better to add a variable libmvec to Makeconfig like the 
existing libm variable, then use it everywhere you have a test using this 
library.  Abstracting out paths to libraries like that may also make it 
easier to support testing installed glibc in future.

> +CFLAGS-test-double-vlen2.c = -fno-inline -ffloat-store -fno-builtin
> -frounding-math \
> +     -D__FAST_MATH__ -DTEST_FAST_MATH -D_OPENMP=201307 \
> +     -Wno-unknown-pragmas
> +CFLAGS-test-double-vlen4.c = -fno-inline -ffloat-store -fno-builtin
> -frounding-math \
> +     -D__FAST_MATH__ -DTEST_FAST_MATH -D_OPENMP=201307 \
> +     -Wno-unknown-pragmas $(arch-ext-cflags)

Why is $(arch-ext-cflags) only in one of these two settings?  Even if you 
don't in fact need those flags for x86_64, this is a generic file, so it 
would seem appropriate to be consistent here.

I think some refactoring is needed to reduce duplication of such settings, 
here and in sysdeps makefiles.

* First a preliminary patch could eliminate existing references to 
-frounding-math other than the global one in Makeconfig - with that global 
one, there's no need for it to be specified for individual files as well.

* Then the existing flag settings could be refactored so that settings 
used for groups of related tests use a single variable rather than being 
duplicated in lots of CFLAGS-* variables.

* Then you might have a libm-test-vector-cflags variable (for example), 
that's based on a common libm-test-fast-math-cflags variable (for 
example), also used for the ifloat.c etc. tests, and adds just the 
vector-test-specific settings such as -D_OPENMP=201307.  Both 
CFLAGS-test-double-vlen2.c and CFLAGS-test-double-vlen4.c could just be 
set to $(libm-test-vector-cflags), as would 
CFLAGS-test-double-vlen4-avx2.c in the sysdeps makefile.

> +#define TEST_VEC_LOOP(len) \
> +  do \
> +    { \
> +      for (i=1; i<len; i++) \
> +        { \
> +          if (((FLOAT *) &mr)[0] != ((FLOAT *) &mr)[i]) \
> +            { \
> +              return ((FLOAT *) &mr)[0]+0.1; \
> +            } \
> +        } \
> +      return ((FLOAT *) &mr)[0]; \
> +    } \
> +  while (0)

Spaces around binary operators ('=', '<', '+').  Put this function in a 
common header rather than duplicating it in test-double-vlen2.h and 
test-double-vlen4.h.

> +  TEST_VEC_LOOP(2); \

Space before '('.

> +  TEST_VEC_LOOP(4); \

Likewise.

> +  TEST_VEC_LOOP(4); \

Likewise.
diff mbox

Patch

diff --git a/math/Makefile b/math/Makefile
index a77687e..7478824 100644
--- a/math/Makefile
+++ b/math/Makefile
@@ -114,8 +114,9 @@  tests-static = test-fpucw-static test-fpucw-ieee-static
 test-longdouble-yes = test-ldouble test-ildoubl

 ifneq (no,$(PERL))
+libm-vec-tests = $(addprefix test-,$(libmvec-tests))
 libm-tests = test-float test-double $(test-longdouble-$(long-double-fcts)) \
- test-ifloat test-idouble
+ test-ifloat test-idouble $(libm-vec-tests)
 libm-tests.o = $(addsuffix .o,$(libm-tests))

 tests += $(libm-tests)
@@ -142,8 +143,22 @@  $(objpfx)test-double.o: $(objpfx)libm-test.stmp
 $(objpfx)test-idouble.o: $(objpfx)libm-test.stmp
 $(objpfx)test-ldouble.o: $(objpfx)libm-test.stmp
 $(objpfx)test-ildoubl.o: $(objpfx)libm-test.stmp
+
+$(objpfx)test-double-vlen2.o: $(objpfx)libm-test.stmp
+$(objpfx)test-double-vlen4.o: $(objpfx)libm-test.stmp
+
+$(objpfx)test-double-vlen2: $(common-objpfx)mathvec/libmvec.so \
+    $(objpfx)init-arch.o
+$(objpfx)test-double-vlen4: $(common-objpfx)mathvec/libmvec.so \
+    $(objpfx)init-arch.o
 endif

+CFLAGS-test-double-vlen2.c = -fno-inline -ffloat-store -fno-builtin
-frounding-math \
+     -D__FAST_MATH__ -DTEST_FAST_MATH -D_OPENMP=201307 \
+     -Wno-unknown-pragmas
+CFLAGS-test-double-vlen4.c = -fno-inline -ffloat-store -fno-builtin
-frounding-math \
+     -D__FAST_MATH__ -DTEST_FAST_MATH -D_OPENMP=201307 \
+     -Wno-unknown-pragmas $(arch-ext-cflags)
 CFLAGS-test-float.c = -fno-inline -ffloat-store -fno-builtin -frounding-math
 CFLAGS-test-double.c = -fno-inline -ffloat-store -fno-builtin -frounding-math
 CFLAGS-test-ldouble.c = -fno-inline -ffloat-store -fno-builtin -frounding-math
diff --git a/math/test-double-vlen2.h b/math/test-double-vlen2.h
new file mode 100644
index 0000000..ca0e4a5
--- /dev/null
+++ b/math/test-double-vlen2.h
@@ -0,0 +1,56 @@ 
+/* Copyright (C) 2014 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <http://www.gnu.org/licenses/>.  */
+
+#define FLOAT double
+#define FUNC(function) function
+#define TEST_MSG "testing double vector math (without inline functions)\n"
+#define MATHCONST(x) x
+#define CHOOSE(Clongdouble,Cdouble,Cfloat,Cinlinelongdouble,Cinlinedouble,Cinlinefloat)
Cdouble
+#define PRINTF_EXPR "e"
+#define PRINTF_XEXPR "a"
+#define PRINTF_NEXPR "f"
+#define TEST_DOUBLE 1
+#define TEST_MATHVEC 1
+
+#ifndef __NO_MATH_INLINES
+# define __NO_MATH_INLINES
+#endif
+
+#define EXCEPTION_TESTS_double 0
+#define ROUNDING_TESTS_double(MODE) ((MODE) == FE_TONEAREST)
+
+#define VEC_SUFF _vlen2
+
+#define CONCAT(a, b) __CONCAT (a, b)
+
+#define WRAPPER_NAME(function) CONCAT (function, VEC_SUFF)
+
+#define FUNC_TEST(function) function ## _VEC_SUFF
+
+#define TEST_VEC_LOOP(len) \
+  do \
+    { \
+      for (i=1; i<len; i++) \
+        { \
+          if (((FLOAT *) &mr)[0] != ((FLOAT *) &mr)[i]) \
+            { \
+              return ((FLOAT *) &mr)[0]+0.1; \
+            } \
+        } \
+      return ((FLOAT *) &mr)[0]; \
+    } \
+  while (0)
diff --git a/math/test-double-vlen4.h b/math/test-double-vlen4.h
new file mode 100644
index 0000000..303e562
--- /dev/null
+++ b/math/test-double-vlen4.h
@@ -0,0 +1,54 @@ 
+/* Copyright (C) 2014 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <http://www.gnu.org/licenses/>.  */
+
+#define FLOAT double
+#define FUNC(function) function
+#define TEST_MSG "testing double vector math (without inline functions)\n"
+#define MATHCONST(x) x
+#define CHOOSE(Clongdouble,Cdouble,Cfloat,Cinlinelongdouble,Cinlinedouble,Cinlinefloat)
Cdouble
+#define PRINTF_EXPR "e"
+#define PRINTF_XEXPR "a"
+#define PRINTF_NEXPR "f"
+#define TEST_DOUBLE 1
+#define TEST_MATHVEC 1
+
+#ifndef __NO_MATH_INLINES
+# define __NO_MATH_INLINES
+#endif
+
+#define EXCEPTION_TESTS_double 0
+#define ROUNDING_TESTS_double(MODE) ((MODE) == FE_TONEAREST)
+
+#define CONCAT(a, b) __CONCAT (a, b)
+
+#define WRAPPER_NAME(function) CONCAT (function, VEC_SUFF)
+
+#define FUNC_TEST(function) function ## _VEC_SUFF
+
+#define TEST_VEC_LOOP(len)                              \
+  do                                                    \
+    {                                                   \
+      for (i=1; i<len; i++)                             \
+        {                                               \
+          if (((FLOAT *) &mr)[0] != ((FLOAT *) &mr)[i]) \
+            {                                           \
+              return ((FLOAT *) &mr)[0]+0.1;            \
+            }                                           \
+        }                                               \
+      return ((FLOAT *) &mr)[0];                        \
+    }                                                   \
+  while (0)
diff --git a/sysdeps/x86_64/fpu/Makefile b/sysdeps/x86_64/fpu/Makefile
index 25fe0d4..e6add6d 100644
--- a/sysdeps/x86_64/fpu/Makefile
+++ b/sysdeps/x86_64/fpu/Makefile
@@ -2,3 +2,20 @@  ifeq ($(subdir),mathvec)
 libmvec-support += svml_d_cos2_core svml_d_cos4_core_avx \
    svml_d_cos4_core_avx2 svml_d_cos_data
 endif
+
+# Rules for libmvec tests.
+ifeq ($(subdir),math)
+ifeq ($(build-mathvec),yes)
+libmvec-tests += double-vlen2 double-vlen4 double-vlen4-avx2
+arch-ext-cflags = -mavx
+
+$(objpfx)test-double-vlen4-avx2.o: $(objpfx)libm-test.stmp
+
+$(objpfx)test-double-vlen4-avx2: $(common-objpfx)mathvec/libmvec.so \
+ $(objpfx)init-arch.o
+
+CFLAGS-test-double-vlen4-avx2.c = -fno-inline -ffloat-store
-fno-builtin -frounding-math \
+  -D__FAST_MATH__ -DTEST_FAST_MATH -D_OPENMP=201307 \
+  -Wno-unknown-pragmas -mavx2
+endif
+endif
diff --git a/sysdeps/x86_64/fpu/libm-test-ulps
b/sysdeps/x86_64/fpu/libm-test-ulps
index 36e1b76..e4de5b4 100644
--- a/sysdeps/x86_64/fpu/libm-test-ulps
+++ b/sysdeps/x86_64/fpu/libm-test-ulps
@@ -905,6 +905,15 @@  idouble: 1
 ildouble: 2
 ldouble: 2

+Function: "cos_vlen2":
+double: 1
+
+Function: "cos_vlen4_avx":
+double: 1
+
+Function: "cos_vlen4_avx2":
+double: 1
+
 Function: "cosh":
 double: 1
 float: 1
diff --git a/sysdeps/x86_64/fpu/test-double-vlen2.c
b/sysdeps/x86_64/fpu/test-double-vlen2.c
new file mode 100644
index 0000000..1d38a1f
--- /dev/null
+++ b/sysdeps/x86_64/fpu/test-double-vlen2.c
@@ -0,0 +1,39 @@ 
+/* Tests for SSE4 ISA versions of vector math functions.
+   Copyright (C) 2014 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <http://www.gnu.org/licenses/>.  */
+
+#include "test-double-vlen2.h"
+#include <immintrin.h>
+
+
+
+// Wrapper from scalar to vector function implemented in SSE4.
+#define VECTOR_WRAPPER(scalar_func, vector_func) \
+extern __m128d vector_func (__m128d); \
+FLOAT scalar_func (FLOAT x) \
+{ \
+  int i; \
+  __m128d mx = _mm_set1_pd (x); \
+  __m128d mr = vector_func (mx); \
+  TEST_VEC_LOOP(2); \
+}
+
+VECTOR_WRAPPER (WRAPPER_NAME (cos), _ZGVbN2v_cos)
+
+#define TEST_VECTOR_cos 1
+
+#include "libm-test.c"
diff --git a/sysdeps/x86_64/fpu/test-double-vlen4-avx2.c
b/sysdeps/x86_64/fpu/test-double-vlen4-avx2.c
new file mode 100644
index 0000000..0bf0be1
--- /dev/null
+++ b/sysdeps/x86_64/fpu/test-double-vlen4-avx2.c
@@ -0,0 +1,41 @@ 
+/* Tests for AVX2 ISA versions of vector math functions.
+   Copyright (C) 2014 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <http://www.gnu.org/licenses/>.  */
+
+#include "test-double-vlen4.h"
+#include <immintrin.h>
+
+// Wrapper from scalar to vector function implemented in AVX2.
+#define VECTOR_WRAPPER(scalar_func, vector_func) \
+extern __m256d vector_func (__m256d); \
+FLOAT scalar_func (FLOAT x) \
+{ \
+  int i; \
+  __m256d mx = _mm256_set1_pd (x); \
+  __m256d mr = vector_func (mx); \
+  TEST_VEC_LOOP(4); \
+}
+
+#define VEC_SUFF _vlen4_avx2
+
+VECTOR_WRAPPER (WRAPPER_NAME (cos), _ZGVdN4v_cos)
+
+#define TEST_VECTOR_cos 1
+
+#define REQUIRE_AVX2
+
+#include "libm-test.c"
diff --git a/sysdeps/x86_64/fpu/test-double-vlen4.c
b/sysdeps/x86_64/fpu/test-double-vlen4.c
new file mode 100644
index 0000000..fd289ad
--- /dev/null
+++ b/sysdeps/x86_64/fpu/test-double-vlen4.c
@@ -0,0 +1,39 @@ 
+/* Tests for AVX ISA versions of vector math functions.
+   Copyright (C) 2014 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <http://www.gnu.org/licenses/>.  */
+
+#include "test-double-vlen4.h"
+#include <immintrin.h>
+
+// Wrapper from scalar to vector function implemented in AVX.
+#define VECTOR_WRAPPER(scalar_func, vector_func) \
+extern __m256d vector_func (__m256d); \
+FLOAT scalar_func (FLOAT x) \
+{ \
+  int i; \
+  __m256d mx = _mm256_set1_pd (x); \
+  __m256d mr = vector_func (mx); \
+  TEST_VEC_LOOP(4); \
+}
+
+#define VEC_SUFF _vlen4_avx
+
+VECTOR_WRAPPER (WRAPPER_NAME (cos), _ZGVcN4v_cos)
+
+#define TEST_VECTOR_cos 1
+
+#include "libm-test.c"
diff --git a/sysdeps/x86_64/fpu/math-tests-arch.h
b/sysdeps/x86_64/fpu/math-tests-arch.h
new file mode 100644
index 0000000..4c2f372
--- /dev/null
+++ b/sysdeps/x86_64/fpu/math-tests-arch.h
@@ -0,0 +1,43 @@ 
+/* Runtime architecture check for math tests. x86_64 version.
+   Copyright (C) 2013-2014 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <http://www.gnu.org/licenses/>.  */
+
+#ifdef REQUIRE_AVX2
+# include <init-arch.h>
+
+  /* Set to 1 if AVX2 supported. */
+  static int avx2_usable;
+
+# define INIT_ARCH_EXT                                                 \
+  do                                                           \
+    {                                                          \
+      __init_cpu_features ();                                  \
+      avx2_usable = __cpu_features.feature[index_AVX2_Usable]  \
+                   & bit_AVX2_Usable;                          \
+    }                                                          \
+  while (0)
+
+# define CHECK_ARCH_EXT                                                \
+  do                                                           \
+    {                                                          \
+      if (!avx2_usable) return;                                        \
+    }                                                          \
+  while (0)
+
+#else
+# include <sysdeps/generic/math-tests-arch.h>
+#endif