mirror of
https://github.com/VectorCamp/vectorscan.git
synced 2025-06-28 16:41:01 +03:00
454 lines
15 KiB
C++
454 lines
15 KiB
C++
/*
|
|
* Copyright (c) 2015-2017, Intel Corporation
|
|
* Copyright (c) 2020-2021, VectorCamp PC
|
|
*
|
|
* Redistribution and use in source and binary forms, with or without
|
|
* modification, are permitted provided that the following conditions are met:
|
|
*
|
|
* * Redistributions of source code must retain the above copyright notice,
|
|
* this list of conditions and the following disclaimer.
|
|
* * Redistributions in binary form must reproduce the above copyright
|
|
* notice, this list of conditions and the following disclaimer in the
|
|
* documentation and/or other materials provided with the distribution.
|
|
* * Neither the name of Intel Corporation nor the names of its contributors
|
|
* may be used to endorse or promote products derived from this software
|
|
* without specific prior written permission.
|
|
*
|
|
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
|
|
* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|
* ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
|
|
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
|
|
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
|
|
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
|
|
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
|
|
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
|
|
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
|
|
* POSSIBILITY OF SUCH DAMAGE.
|
|
*/
|
|
|
|
#include<iostream>
|
|
#include<cstring>
|
|
#include<time.h>
|
|
#include"gtest/gtest.h"
|
|
#include"ue2common.h"
|
|
#include"util/arch.h"
|
|
#include"util/simd_utils.h"
|
|
#include"util/simd/types.hpp"
|
|
|
|
|
|
typedef union uni128i{__m128i f; int8_t vec[16];}u128i;
|
|
|
|
TEST(SuperVectorUtilsTest, Zero128) {
|
|
m128_t zeroes = SuperVector<16>::Zeroes();
|
|
u128i z;
|
|
z.f = _mm_set1_epi8(0);
|
|
for(int i=0; i<16; i++){ASSERT_EQ(zeroes.u.s8[i],z.vec[i]);}
|
|
}
|
|
|
|
TEST(SuperVectorUtilsTest, Ones128) {
|
|
m128_t ones = SuperVector<16>::Ones();
|
|
u128i z;
|
|
z.f = _mm_set1_epi8(0xff);
|
|
for(int i=0; i<16; i++){ASSERT_EQ(ones.u.s8[i],z.vec[i]);}
|
|
}
|
|
|
|
|
|
TEST(SuperVectorUtilsTest, Loadu128) {
|
|
int vec[4];
|
|
srand(time(NULL));
|
|
for (int i=0; i<4; i++) {vec[i]=rand() %1000 +1;}
|
|
m128_t SP = SuperVector<16>::loadu((__m128i*)vec);
|
|
u128i test_vector;
|
|
test_vector.f = _mm_lddqu_si128((__m128i*)vec);
|
|
for(int i=0; i<16; i++){ ASSERT_EQ(SP.u.s8[i],test_vector.vec[i]);}
|
|
}
|
|
|
|
TEST(SuperVectorUtilsTest, Load128) {
|
|
int vec[4] __attribute__((aligned(16)));
|
|
srand(time(NULL));
|
|
for (int i=0; i<4; i++) {vec[i]=rand() %1000 +1;}
|
|
m128_t SP = SuperVector<16>::load((__m128i*)vec);
|
|
u128i test_vector;
|
|
test_vector.f = _mm_load_si128((__m128i*)vec);
|
|
for(int i=0; i<16; i++){ ASSERT_EQ(SP.u.s8[i],test_vector.vec[i]);}
|
|
}
|
|
|
|
TEST(SuperVectorUtilsTest,Equal128){
|
|
int vec[8];
|
|
srand(time(NULL));
|
|
for (int i=0; i<8; i++) {vec[i]=rand() %1000 +1;}
|
|
m128_t SP1 = SuperVector<16>::loadu((__m128i*)vec);
|
|
m128_t SP2 = SuperVector<16>::loadu((__m128i*)vec+4);
|
|
u128i test_vector1;
|
|
u128i test_vector2;
|
|
test_vector1.f = _mm_loadu_si128((__m128i*)vec);
|
|
test_vector2.f = _mm_loadu_si128((__m128i*)vec+4);
|
|
m128_t SPResult = SP1.eq(SP2);
|
|
u128i test_result;
|
|
test_result.f = _mm_cmpeq_epi8(test_vector1.f,test_vector2.f);
|
|
for (int i=0; i<16; i++){ASSERT_EQ(SPResult.u.s8[i],test_result.vec[i]);}
|
|
}
|
|
|
|
TEST(SuperVectorUtilsTest,And128){
|
|
m128_t SPResult = SuperVector<16>::Zeroes() & SuperVector<16>::Ones();
|
|
__m128i test_vector1 = _mm_set1_epi8(0);
|
|
__m128i test_vector2 = _mm_set1_epi8(0xff);
|
|
u128i test_result;
|
|
test_result.f = _mm_and_si128(test_vector1,test_vector2);
|
|
for (int i=0; i<16; i++){ASSERT_EQ(SPResult.u.s8[i],test_result.vec[i]);}
|
|
}
|
|
|
|
TEST(SuperVectorUtilsTest,Movemask128){
|
|
int vec[4];
|
|
srand(time(NULL));
|
|
for (int i=0; i<4; i++) {vec[i]=rand() %1000 +1;}
|
|
m128_t SP = SuperVector<16>::loadu((__m128i*)vec);
|
|
__m128i test_vector = _mm_loadu_si128((__m128i*)vec);
|
|
int SP_Mask = SP.movemask();
|
|
int test_result = _mm_movemask_epi8(test_vector);
|
|
ASSERT_EQ(SP_Mask,test_result);
|
|
}
|
|
|
|
TEST(SuperVectorUtilsTest,Eqmask128){
|
|
int vec[8];
|
|
srand(time(NULL));
|
|
for (int i=0; i<8; i++) {vec[i]=rand() %1000 +1;}
|
|
m128_t SP = SuperVector<16>::loadu((__m128i*)vec);
|
|
m128_t SP1 = SuperVector<16>::loadu((__m128i*)vec+4);
|
|
__m128i test_vector1 = _mm_loadu_si128((__m128i*)vec);
|
|
__m128i test_vector2 = _mm_loadu_si128((__m128i*)vec+4);
|
|
__m128i test_result = _mm_cmpeq_epi8(test_vector1,test_vector2);
|
|
int SP_Mask = SP.eqmask(SP1);
|
|
int test_res = _mm_movemask_epi8(test_result);
|
|
ASSERT_EQ(SP_Mask,test_res);
|
|
}
|
|
|
|
/*Define SHIFT128 macro*/
|
|
#define TEST_SHIFT128(l) \
|
|
SP_after_shift = SP<<(l); \
|
|
test_vector_after_shift.f = _mm_slli_si128(test_vector.f,l); \
|
|
for(int i=0; i<16; i++) {ASSERT_EQ(SP_after_shift.u.s8[i],test_vector_after_shift.vec[i]);} \
|
|
|
|
TEST(SuperVectorUtilsTest,Shift128){
|
|
int vec[4];
|
|
srand(time(NULL));
|
|
for (int i=0; i<4; i++) {vec[i]=rand() %1000 +1;}
|
|
m128_t SP = SuperVector<16>::loadu((__m128i*)vec);
|
|
u128i test_vector;
|
|
test_vector.f = _mm_loadu_si128((__m128i*)vec);
|
|
u128i test_vector_after_shift;
|
|
m128_t SP_after_shift = SP<<(0);
|
|
TEST_SHIFT128(1)
|
|
TEST_SHIFT128(2)
|
|
TEST_SHIFT128(3)
|
|
TEST_SHIFT128(4)
|
|
TEST_SHIFT128(5)
|
|
TEST_SHIFT128(6)
|
|
TEST_SHIFT128(7)
|
|
TEST_SHIFT128(8)
|
|
TEST_SHIFT128(9)
|
|
TEST_SHIFT128(10)
|
|
TEST_SHIFT128(11)
|
|
TEST_SHIFT128(12)
|
|
TEST_SHIFT128(13)
|
|
TEST_SHIFT128(14)
|
|
TEST_SHIFT128(15)
|
|
TEST_SHIFT128(16)
|
|
}
|
|
|
|
#define ALIGNR128(l) \
|
|
al_test.f = _mm_alignr_epi8(test_vector1,test_vector2,l); \
|
|
SP_test = SP.alignr(SP1,l); \
|
|
for (int i=0; i<16; i++) {ASSERT_EQ(SP_test.u.s8[i],al_test.vec[i]);} \
|
|
|
|
TEST(SuperVectorUtilsTest,Alignr128){
|
|
int vec[8];
|
|
srand(time(NULL));
|
|
for (int i=0; i<8; i++) {vec[i]=rand() %1000 +1;}
|
|
m128_t SP = SuperVector<16>::loadu((__m128i*)vec);
|
|
m128_t SP1 = SuperVector<16>::loadu((__m128i*)vec+4);
|
|
__m128i test_vector1 = _mm_loadu_si128((__m128i*)vec);
|
|
__m128i test_vector2 = _mm_loadu_si128((__m128i*)vec+4);
|
|
u128i al_test;
|
|
m128_t SP_test = SP.alignr(SP1,0);
|
|
ALIGNR128(1);
|
|
ALIGNR128(2);
|
|
ALIGNR128(3);
|
|
ALIGNR128(4);
|
|
ALIGNR128(5);
|
|
ALIGNR128(6);
|
|
ALIGNR128(7);
|
|
ALIGNR128(8);
|
|
ALIGNR128(9);
|
|
ALIGNR128(10);
|
|
ALIGNR128(11);
|
|
ALIGNR128(12);
|
|
ALIGNR128(13);
|
|
ALIGNR128(14);
|
|
ALIGNR128(15);
|
|
ALIGNR128(16);
|
|
}
|
|
|
|
|
|
#if defined(HAVE_AVX2)
|
|
typedef union uni256i{__m256i f; int8_t vec[32];}u256i;
|
|
|
|
TEST(SuperVectorUtilsTest, Ones256) {
|
|
m256_t zeroes = SuperVector<32>::Ones();
|
|
u256i z;
|
|
z.f = _mm256_set1_epi8(0xff);
|
|
for(int i=0; i<32; i++){ASSERT_EQ(zeroes.u.s8[i],z.vec[i]);}
|
|
}
|
|
|
|
TEST(SuperVectorUtilsTest, Zero256) {
|
|
m256_t ones = SuperVector<32>::Zeroes();
|
|
u256i z;
|
|
z.f = _mm256_set1_epi8(0);
|
|
for(int i=0; i<32; i++){ASSERT_EQ(ones.u.s8[i],z.vec[i]);}
|
|
}
|
|
|
|
TEST(SuperVectorUtilsTest, Load256) {
|
|
int vec[8] __attribute__((aligned(16)));
|
|
srand(time(NULL));
|
|
for (int i=0; i<8; i++) {vec[i]=rand() %1000 +1;}
|
|
m256_t SP = SuperVector<32>::load((__m256i*)vec);
|
|
u256i test_vector;
|
|
test_vector.f = _mm256_load_si256((__m256i*)vec);
|
|
for(int i=0; i<32; i++){ ASSERT_EQ(SP.u.s8[i],test_vector.vec[i]);}
|
|
}
|
|
|
|
TEST(SuperVectorUtilsTest, Loadu256) {
|
|
int vec[8];
|
|
srand(time(NULL));
|
|
for (int i=0; i<8; i++) {vec[i]=rand() %1000 +1;}
|
|
m256_t SP = SuperVector<32>::loadu((__m256i*)vec);
|
|
u256i test_vector;
|
|
test_vector.f = _mm256_lddqu_si256((__m256i*)vec);
|
|
for(int i=0; i<32; i++){ ASSERT_EQ(SP.u.s8[i],test_vector.vec[i]);}
|
|
}
|
|
|
|
TEST(SuperVectorUtilsTest,Equal256){
|
|
int vec[16];
|
|
srand(time(NULL));
|
|
for (int i=0; i<16; i++) {vec[i]=rand() %1000 +1;}
|
|
m256_t SP1 = SuperVector<32>::loadu((__m256i*)vec);
|
|
m256_t SP2 = SuperVector<32>::loadu((__m256i*)vec+8);
|
|
u256i test_vector1;
|
|
u256i test_vector2;
|
|
test_vector1.f = _mm256_loadu_si256((__m256i*)vec);
|
|
test_vector2.f = _mm256_loadu_si256((__m256i*)vec+8);
|
|
m256_t SPResult = SP1.eq(SP2);
|
|
u256i test_result;
|
|
test_result.f = _mm256_cmpeq_epi8(test_vector1.f,test_vector2.f);
|
|
for (int i=0; i<32; i++){ASSERT_EQ(SPResult.u.s8[i],test_result.vec[i]);}
|
|
}
|
|
|
|
TEST(SuperVectorUtilsTest,And256){
|
|
m256_t SPResult = SuperVector<32>::Zeroes() & SuperVector<32>::Ones();
|
|
__m256i test_vector1 = _mm256_set1_epi8(0);
|
|
__m256i test_vector2 = _mm256_set1_epi8(0xff);
|
|
u256i test_result;
|
|
test_result.f = _mm256_and_si256(test_vector1,test_vector2);
|
|
for (int i=0; i<32; i++){ASSERT_EQ(SPResult.u.s8[i],test_result.vec[i]);}
|
|
}
|
|
|
|
TEST(SuperVectorUtilsTest,Movemask256){
|
|
int vec[8];
|
|
srand(time(NULL));
|
|
for (int i=0; i<8; i++) {vec[i]=rand() %1000 +1;}
|
|
m256_t SP = SuperVector<32>::loadu((__m256i*)vec);
|
|
__m256i test_vector = _mm256_loadu_si256((__m256i*)vec);
|
|
int SP_Mask = SP.movemask();
|
|
int test_result = _mm256_movemask_epi8(test_vector);
|
|
ASSERT_EQ(SP_Mask,test_result);
|
|
}
|
|
|
|
TEST(SuperVectorUtilsTest,Eqmask256){
|
|
int vec[16];
|
|
srand(time(NULL));
|
|
for (int i=0; i<16; i++) {vec[i]=rand() %1000 +1;}
|
|
m256_t SP = SuperVector<32>::loadu((__m256i*)vec);
|
|
m256_t SP1 = SuperVector<32>::loadu((__m256i*)vec+8);
|
|
__m256i test_vector1 = _mm256_loadu_si256((__m256i*)vec);
|
|
__m256i test_vector2 = _mm256_loadu_si256((__m256i*)vec+8);
|
|
__m256i test_result = _mm256_cmpeq_epi8(test_vector1,test_vector2);
|
|
int SP_Mask = SP.eqmask(SP1);
|
|
int test_res = _mm256_movemask_epi8(test_result);
|
|
ASSERT_EQ(SP_Mask,test_res);
|
|
}
|
|
|
|
/*Define SHIFT256 macro*/
|
|
#define TEST_SHIFT256(l) \
|
|
SP_after_shift = SP<<(l); \
|
|
test_vector_after_shift.f = _mm256_slli_si256(test_vector.f,l); \
|
|
for(int i=0; i<32; i++) {ASSERT_EQ(SP_after_shift.u.s8[i],test_vector_after_shift.vec[i]);} \
|
|
|
|
TEST(SuperVectorUtilsTest,Shift256){
|
|
int vec[8];
|
|
srand(time(NULL));
|
|
for (int i=0; i<8; i++) {vec[i]=rand() %1000 +1;}
|
|
m256_t SP = SuperVector<32>::loadu((__m128i*)vec);
|
|
u256i test_vector;
|
|
test_vector.f = _mm256_loadu_si256((__m256i*)vec);
|
|
u256i test_vector_after_shift;
|
|
m256_t SP_after_shift = SP<<(0);
|
|
TEST_SHIFT256(1)
|
|
TEST_SHIFT256(2)
|
|
TEST_SHIFT256(3)
|
|
TEST_SHIFT256(4)
|
|
TEST_SHIFT256(5)
|
|
TEST_SHIFT256(6)
|
|
TEST_SHIFT256(7)
|
|
TEST_SHIFT256(8)
|
|
TEST_SHIFT256(9)
|
|
TEST_SHIFT256(10)
|
|
TEST_SHIFT256(11)
|
|
TEST_SHIFT256(12)
|
|
TEST_SHIFT256(13)
|
|
TEST_SHIFT256(14)
|
|
TEST_SHIFT256(15)
|
|
TEST_SHIFT256(16)
|
|
}
|
|
|
|
#define ALIGNR256(l) \
|
|
al_test.f = _mm256_alignr_epi8(test_vector1,test_vector2,l); \
|
|
SP_test = SP.alignr(SP1,l); \
|
|
for (int i=0; i<32; i++) {ASSERT_EQ(SP_test.u.s8[i],al_test.vec[i]);} \
|
|
|
|
TEST(SuperVectorUtilsTest,Alignr256){
|
|
int vec[16];
|
|
srand(time(NULL));
|
|
for (int i=0; i<16; i++) {vec[i]=rand() %1000 +1;}
|
|
m256_t SP = SuperVector<32>::loadu((__m256i*)vec);
|
|
m256_t SP1 = SuperVector<32>::loadu((__m256i*)vec+8);
|
|
__m256i test_vector1 = _mm256_loadu_si256((__m256i*)vec);
|
|
__m256i test_vector2 = _mm256_loadu_si256((__m256i*)vec+8);
|
|
u256i al_test;
|
|
m256_t SP_test = SP.alignr(SP1,0);
|
|
ALIGNR256(1);
|
|
ALIGNR256(2);
|
|
ALIGNR256(3);
|
|
ALIGNR256(4);
|
|
ALIGNR256(5);
|
|
ALIGNR256(6);
|
|
ALIGNR256(7);
|
|
ALIGNR256(8);
|
|
ALIGNR256(9);
|
|
ALIGNR256(10);
|
|
ALIGNR256(11);
|
|
ALIGNR256(12);
|
|
ALIGNR256(13);
|
|
ALIGNR256(14);
|
|
ALIGNR256(15);
|
|
ALIGNR256(16);
|
|
}
|
|
#endif
|
|
|
|
#if defined(HAVE_AVX512)
|
|
typedef union uni512i{__m512i f; int8_t vec[64];}u512i;
|
|
|
|
TEST(SuperVectorUtilsTest, Ones512) {
|
|
m512_t zeroes = SuperVector<64>::Ones();
|
|
u512i z;
|
|
z.f = _mm512_set1_epi8(0xff);
|
|
for(int i=0; i<64; i++){ASSERT_EQ(zeroes.u.s8[i],z.vec[i]);}
|
|
}
|
|
|
|
TEST(SuperVectorUtilsTest, Zero512) {
|
|
m512_t ones = SuperVector<64>::Zeroes();
|
|
u512i z;
|
|
z.f = _mm512_set1_epi8(0);
|
|
for(int i=0; i<64; i++){ ASSERT_EQ(ones.u.s8[i],z.vec[i]);}
|
|
}
|
|
|
|
TEST(SuperVectorUtilsTest, Load512) {
|
|
int vec[16] __attribute__((aligned(64)));
|
|
srand(time(NULL));
|
|
for (int i=0; i<16; i++) {vec[i]=rand() %1000 +1;}
|
|
m512_t SP = SuperVector<64>::load((__m512i*)vec);
|
|
u512i test_vector;
|
|
test_vector.f = _mm512_load_si512((__m512i*)vec);
|
|
for(int i=0; i<64; i++){ ASSERT_EQ(SP.u.s8[i],test_vector.vec[i]);}
|
|
}
|
|
|
|
TEST(SuperVectorUtilsTest, Loadu512) {
|
|
int vec[16];
|
|
srand(time(NULL));
|
|
for (int i=0; i<16; i++) {vec[i]=rand() %1000 +1;}
|
|
m512_t SP = SuperVector<64>::loadu((__m512i*)vec);
|
|
u512i test_vector;
|
|
test_vector.f = _mm512_loadu_si512((__m512i*)vec);
|
|
for(int i=0; i<64; i++){ ASSERT_EQ(SP.u.s8[i],test_vector.vec[i]);}
|
|
}
|
|
|
|
/* This method is under construction
|
|
TEST(SuperVectorUtilsTest,Equal512){}
|
|
*/
|
|
|
|
TEST(SuperVectorUtilsTest,And512){
|
|
m512_t SPResult = SuperVector<64>::Zeroes() & SuperVector<64>::Ones();
|
|
__m512i test_vector1 = _mm512_set1_epi8(0);
|
|
__m512i test_vector2 = _mm512_set1_epi8(0xff);
|
|
u512i test_result;
|
|
test_result.f = _mm512_and_si512(test_vector1,test_vector2);
|
|
for (int i=0; i<64; i++){ASSERT_EQ(SPResult.u.s8[i],test_result.vec[i]);}
|
|
}
|
|
|
|
/* This methos is under construction
|
|
TEST(SuperVectorUtilsTest,Movemask256){}
|
|
*/
|
|
|
|
TEST(SuperVectorUtilsTest,Eqmask512){
|
|
int vec[16];
|
|
srand(time(NULL));
|
|
for (int i=0; i<16; i++) {vec[i]=rand() %1000 +1;}
|
|
m256_t SP = SuperVector<32>::loadu((__m256i*)vec);
|
|
m256_t SP1 = SuperVector<32>::loadu((__m256i*)vec+8);
|
|
__m256i test_vector1 = _mm256_loadu_si256((__m256i*)vec);
|
|
__m256i test_vector2 = _mm256_loadu_si256((__m256i*)vec+8);
|
|
__m256i test_result = _mm256_cmpeq_epi8(test_vector1,test_vector2);
|
|
int SP_Mask = SP.eqmask(SP1);
|
|
int test_res = _mm256_movemask_epi8(test_result);
|
|
ASSERT_EQ(SP_Mask,test_res);
|
|
}
|
|
/*
|
|
This methos is under construction
|
|
TEST(SuperVectorUtilsTest,Shift256){}
|
|
*/
|
|
|
|
#define ALIGNR512(l) \
|
|
al_test.f = _mm512_alignr_epi8(test_vector1,test_vector2,l); \
|
|
SP_test = SP.alignr(SP1,l); \
|
|
for (int i=0; i<64; i++) {ASSERT_EQ(SP_test.u.s8[i],al_test.vec[i]);} \
|
|
|
|
TEST(SuperVectorUtilsTest,Alignr512){
|
|
int vec[32];
|
|
srand(time(NULL));
|
|
for (int i=0; i<32; i++) {vec[i]=rand() %1000 +1;}
|
|
m512_t SP = SuperVector<64>::loadu((__m512i*)vec);
|
|
m512_t SP1 = SuperVector<64>::loadu((__m512i*)vec+16);
|
|
__m512i test_vector1 = _mm512_loadu_si512((__m512i*)vec);
|
|
__m512i test_vector2 = _mm512_loadu_si512((__m512i*)vec+16);
|
|
u512i al_test;
|
|
m512_t SP_test = SP.alignr(SP1,0);
|
|
ALIGNR512(1);
|
|
ALIGNR512(2);
|
|
ALIGNR512(3);
|
|
ALIGNR512(4);
|
|
ALIGNR512(5);
|
|
ALIGNR512(6);
|
|
ALIGNR512(7);
|
|
ALIGNR512(8);
|
|
ALIGNR512(9);
|
|
ALIGNR512(10);
|
|
ALIGNR512(11);
|
|
ALIGNR512(12);
|
|
ALIGNR512(13);
|
|
ALIGNR512(14);
|
|
ALIGNR512(15);
|
|
ALIGNR512(16);
|
|
}
|
|
|
|
#endif
|