Add a configure-time check for QT_COMPILER_SUPPORTS_SIMD_ALWAYS

This has two main benefits:
 1) introduces a qmake CONFIG we can use in .pro/.pri/.prf files
 2) removes the need to keep an up-to-date list of which compilers
    support the feature

The test is implemented as trying to compile every single SIMD test we
currently have, but without passing the -mXXX option. The reason for
trying all of them is that some people may have modified their mkspecs
to add -mXXX options or -march=XXX, which could enable the particular
feature we tried, resulting in a false positive outcome.

Change-Id: I938b024e38bf4aac9154fffd14f7784dc8d1f020
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
This commit is contained in:
Thiago Macieira 2017-11-15 21:01:15 -08:00
parent ff1ffa7577
commit a09fc184ac
5 changed files with 291 additions and 6 deletions

View File

@ -0,0 +1,253 @@
/****************************************************************************
**
** Copyright (C) 2017 Intel Corporation.
** Contact: https://www.qt.io/licensing/
**
** This file is part of the QtCore module of the Qt Toolkit.
**
** $QT_BEGIN_LICENSE:LGPL$
** Commercial License Usage
** Licensees holding valid commercial Qt licenses may use this file in
** accordance with the commercial license agreement provided with the
** Software or, alternatively, in accordance with the terms contained in
** a written agreement between you and The Qt Company. For licensing terms
** and conditions see https://www.qt.io/terms-conditions. For further
** information use the contact form at https://www.qt.io/contact-us.
**
** GNU Lesser General Public License Usage
** Alternatively, this file may be used under the terms of the GNU Lesser
** General Public License version 3 as published by the Free Software
** Foundation and appearing in the file LICENSE.LGPL3 included in the
** packaging of this file. Please review the following information to
** ensure the GNU Lesser General Public License version 3 requirements
** will be met: https://www.gnu.org/licenses/lgpl-3.0.html.
**
** GNU General Public License Usage
** Alternatively, this file may be used under the terms of the GNU
** General Public License version 2.0 or (at your option) the GNU General
** Public license version 3 or any later version approved by the KDE Free
** Qt Foundation. The licenses are as published by the Free Software
** Foundation and appearing in the file LICENSE.GPL2 and LICENSE.GPL3
** included in the packaging of this file. Please review the following
** information to ensure the GNU General Public License requirements will
** be met: https://www.gnu.org/licenses/gpl-2.0.html and
** https://www.gnu.org/licenses/gpl-3.0.html.
**
** $QT_END_LICENSE$
**
****************************************************************************/
// All of our supported compilers support <immintrin.h>
#include <immintrin.h>
#define T(x) (QT_COMPILER_SUPPORTS_ ## x)
#if !defined(__INTEL_COMPILER) && !defined(_MSC_VER)
/* GCC requires attributes for a function */
# define attribute_target(x) __attribute__((__target__(x)))
#else
# define attribute_target(x)
#endif
#if T(SSE2)
attribute_target("sse2") void test_sse2()
{
__m128i a = _mm_setzero_si128();
_mm_maskmoveu_si128(a, _mm_setzero_si128(), 0);
}
#endif
#if T(SSE3)
attribute_target("sse3") void test_sse3()
{
__m128d a = _mm_set1_pd(6.28);
__m128d b = _mm_set1_pd(3.14);
__m128d result = _mm_addsub_pd(a, b);
(void) _mm_movedup_pd(result);
}
#endif
#if T(SSSE3)
attribute_target("ssse3") void test_ssse3()
{
__m128i a = _mm_set1_epi32(42);
_mm_abs_epi8(a);
(void) _mm_sign_epi16(a, _mm_set1_epi32(64));
}
#endif
#if T(SSE4_1)
attribute_target("sse4.1") void test_sse4_1()
{
__m128 a = _mm_setzero_ps();
_mm_ceil_ps(a);
__m128i result = _mm_mullo_epi32(_mm_set1_epi32(42), _mm_set1_epi32(64));
(void)result;
}
#endif
#if T(SSE4_2)
attribute_target("sse4.2") void test_sse4_2()
{
__m128i a = _mm_setzero_si128();
__m128i b = _mm_set1_epi32(42);
(void) _mm_cmpestrm(a, 16, b, 16, 0);
}
#endif
#if T(AESNI)
attribute_target("aes,sse4.2") void test_aesni()
{
__m128i a = _mm_setzero_si128();
__m128i b = _mm_aesenc_si128(a, a);
__m128i c = _mm_aesdec_si128(a, b);
(void)c;
}
#endif
#if T(F16C)
attribute_target("f16c") void test_f16c()
{
__m128i a = _mm_setzero_si128();
__m128 b = _mm_cvtph_ps(a);
__m256 b256 = _mm256_cvtph_ps(a);
(void) _mm_cvtps_ph(b, 0);
(void) _mm256_cvtps_ph(b256, 0);
}
#endif
#if T(RDRND)
attribute_target("rdrnd") int test_rdrnd()
{
unsigned short us;
unsigned int ui;
if (_rdrand16_step(&us))
return 1;
if (_rdrand32_step(&ui))
return 1;
# if defined(__x86_64) || defined(__x86_64__) || defined(__amd64) || defined(_M_X64)
unsigned long long ull;
if (_rdrand64_step(&ull))
return 1;
# endif
}
#endif
#if T(SHANI)
attribute_target("sha") void test_shani()
{
__m128i a = _mm_setzero_si128();
__m128i b = _mm_sha1rnds4_epu32(a, a, 0);
__m128i c = _mm_sha1msg1_epu32(a, b);
__m128i d = _mm_sha256msg2_epu32(b, c);
(void)d;
}
#endif
#if T(AVX)
attribute_target("avx") void test_avx()
{
__m256d a = _mm256_setzero_pd();
__m256d b = _mm256_set1_pd(42.42);
(void) _mm256_add_pd(a, b);
}
#endif
#if T(AVX2)
attribute_target("avx2") void test_avx2()
{
_mm256_zeroall();
__m256i a = _mm256_setzero_si256();
__m256i b = _mm256_and_si256(a, a);
(void) _mm256_add_epi8(a, b);
}
#endif
#if T(AVX512F)
attribute_target("avx512f") void test_avx512f(char *ptr)
{
/* AVX512 Foundation */
__mmask16 m = ~1;
__m512i i;
__m512d d;
__m512 f;
i = _mm512_maskz_loadu_epi32(0, ptr);
d = _mm512_loadu_pd((double *)ptr + 64);
f = _mm512_loadu_ps((float *)ptr + 128);
_mm512_mask_storeu_epi64(ptr, m, i);
_mm512_mask_storeu_ps(ptr + 64, m, f);
_mm512_mask_storeu_pd(ptr + 128, m, d);
}
#endif
#if T(AVX512ER)
attribute_target("avx512er") void test_avx512er()
{
/* AVX512 Exponential and Reciprocal */
__m512 f;
f = _mm512_exp2a23_round_ps(f, 8);
}
#endif
#if T(AVX512CD)
attribute_target("avx512cd") void test_avx512cd()
{
/* AVX512 Conflict Detection */
__mmask16 m = ~1;
__m512i i;
i = _mm512_maskz_conflict_epi32(m, i);
}
#endif
#if T(AVX512PF)
attribute_target("avx512pf") void test_avx512pf(void *ptr)
{
/* AVX512 Prefetch */
__m512i i;
__mmask16 m = 0xf;
_mm512_mask_prefetch_i64scatter_pd(ptr, m, i, 2, 2);
}
#endif
#if T(AVX512DQ)
attribute_target("avx512dq") void test_avx512dq()
{
/* AVX512 Doubleword and Quadword support */
__m512i i;
__mmask16 m = ~1;
m = _mm512_movepi32_mask(i);
}
#endif
#if T(AVX512BW)
attribute_target("avx512bw") void test_avx512bw(char *ptr)
{
/* AVX512 Byte and Word support */
__m512i i;
__mmask16 m = ~1;
i = _mm512_mask_loadu_epi8(i, m, ptr - 8);
}
#endif
#if T(AVX512VL)
attribute_target("avx512vl") void test_avx512vl(char *ptr)
{
/* AVX512 Vector Length */
__mmask16 m = ~1;
__m256i i2 = _mm256_maskz_loadu_epi32(0, ptr);
_mm256_mask_storeu_epi32(ptr + 1, m, i2);
}
#endif
#if T(AVX512IFMA)
attribute_target("avx512ifma") void test_avx512ifma()
{
/* AVX512 Integer Fused Multiply-Add */
__m512i i;
i = _mm512_madd52lo_epu64(i, i, i);
}
#endif
int main()
{
return 0;
}

View File

@ -0,0 +1,5 @@
SOURCES = main.cpp
for (config, SIMD) {
uc = $$upper($$config)
DEFINES += QT_COMPILER_SUPPORTS_$${uc}
}

View File

@ -218,7 +218,7 @@
},
"testTypeAliases": {
"compile": [ "library", "architecture" ]
"compile": [ "library", "architecture", "x86SimdAlways" ]
},
"tests": {
@ -563,6 +563,11 @@
"test": "avx512",
"args": "AVX512=VBMI"
},
"x86SimdAlways": {
"label": "Intrinsics without -mXXX argument",
"type": "x86SimdAlways",
"test": "x86_simd"
},
"mips_dsp": {
"label": "MIPS DSP instructions",
"type": "subarch",
@ -1111,6 +1116,14 @@
{ "type": "define", "name": "QT_COMPILER_SUPPORTS_SHA", "value": 1 }
]
},
"x86SimdAlways": {
"label": "Intrinsics without -mXXX option",
"condition": "(arch.i386 || arch.x86_64) && tests.x86SimdAlways",
"output": [
"privateConfig",
{ "type": "define", "name": "QT_COMPILER_SUPPORTS_SIMD_ALWAYS", "value": 1 }
]
},
"mips_dsp": {
"label": "DSP",
"condition": "arch.mips && tests.mips_dsp",
@ -1397,6 +1410,11 @@ Configure with '-qreal float' to create a build that is binary-compatible with 5
"args": "aesni f16c rdrnd shani",
"condition": "(arch.i386 || arch.x86_64)"
},
{
"type": "feature",
"args": "x86SimdAlways",
"condition": "(arch.i386 || arch.x86_64) && !config.msvc"
},
{
"type": "feature",
"args": "neon",

View File

@ -395,6 +395,19 @@ defineTest(qtConfTest_buildParts) {
return(true)
}
defineTest(qtConfTest_x86SimdAlways) {
configs =
fpfx = $${currentConfig}.features
simd = sse2 sse3 ssse3 sse4_1 sse4_2 rdrnd aesni shani avx avx2 avx512f \
avx512er avx512cd avx512pf avx512dq avx512bw avx512vl avx512ifma avx512vbmi
for (f, simd) {
qtConfCheckFeature($$f)
equals($${fpfx}.$${f}.available, true): configs += $$f
}
$${1}.literal_args = $$system_quote(SIMD=$$join(configs, " "))
qtConfTest_compile($${1})
}
# custom outputs
# this reloads the qmakespec as completely as reasonably possible.

View File

@ -164,11 +164,7 @@
# if !defined(__MIPS_DSPR2__) && defined(__mips_dspr2) && defined(Q_PROCESSOR_MIPS_32)
# define __MIPS_DSPR2__
# endif
#elif (defined(Q_CC_INTEL) || defined(Q_CC_MSVC) \
|| (defined(Q_CC_GNU) && !defined(Q_CC_CLANG) && Q_CC_GNU >= 409) \
|| (defined(Q_CC_CLANG) && Q_CC_CLANG >= 308)) \
&& !defined(QT_BOOTSTRAPPED)
# define QT_COMPILER_SUPPORTS_SIMD_ALWAYS
#elif defined(Q_PROCESSOR_X86) && defined(QT_COMPILER_SUPPORTS_SIMD_ALWAYS)
# define QT_COMPILER_SUPPORTS_HERE(x) ((__ ## x ## __) || QT_COMPILER_SUPPORTS(x))
# if defined(Q_CC_GNU) && !defined(Q_CC_INTEL)
/* GCC requires attributes for a function */