Source code

Revision control

Copy as Markdown

Other Tools

// Copyright 2019 Google LLC
// SPDX-License-Identifier: Apache-2.0
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
#include <stddef.h>
#include <stdint.h>
#undef HWY_TARGET_INCLUDE
#define HWY_TARGET_INCLUDE "tests/reduction_test.cc"
#include "hwy/foreach_target.h" // IWYU pragma: keep
#include "hwy/highway.h"
#include "hwy/tests/test_util-inl.h"
HWY_BEFORE_NAMESPACE();
namespace hwy {
namespace HWY_NAMESPACE {
struct TestSumOfLanes {
template <typename D,
hwy::EnableIf<!IsSigned<TFromD<D>>() ||
((HWY_MAX_LANES_D(D) & 1) != 0)>* = nullptr>
HWY_NOINLINE void SignedEvenLengthVectorTests(D /*d*/) {
// do nothing
}
template <typename D,
hwy::EnableIf<IsSigned<TFromD<D>>() &&
((HWY_MAX_LANES_D(D) & 1) == 0)>* = nullptr>
HWY_NOINLINE void SignedEvenLengthVectorTests(D d) {
using T = TFromD<D>;
const size_t lanes = Lanes(d);
#if HWY_HAVE_SCALABLE
// On platforms that use scalable vectors, it is possible for Lanes(d) to be
// odd but for MaxLanes(d) to be even if Lanes(d) < 2 is true.
if (lanes < 2) return;
#endif
const T pairs = ConvertScalarTo<T>(lanes / 2);
// Lanes are the repeated sequence -2, 1, [...]; each pair sums to -1,
// so the eventual total is just -(N/2).
Vec<decltype(d)> v = InterleaveLower(Set(d, ConvertScalarTo<T>(-2)),
Set(d, ConvertScalarTo<T>(1)));
HWY_ASSERT_VEC_EQ(d, Set(d, ConvertScalarTo<T>(-pairs)), SumOfLanes(d, v));
HWY_ASSERT_EQ(ConvertScalarTo<T>(-pairs), ReduceSum(d, v));
// Similar test with a positive result.
v = InterleaveLower(Set(d, ConvertScalarTo<T>(-2)),
Set(d, ConvertScalarTo<T>(4)));
HWY_ASSERT_VEC_EQ(d,
Set(d, ConvertScalarTo<T>(pairs * ConvertScalarTo<T>(2))),
SumOfLanes(d, v));
HWY_ASSERT_EQ(ConvertScalarTo<T>(pairs * ConvertScalarTo<T>(2)),
ReduceSum(d, v));
}
template <typename T, class D>
HWY_NOINLINE void operator()(T /*unused*/, D d) {
const size_t N = Lanes(d);
auto in_lanes = AllocateAligned<T>(N);
HWY_ASSERT(in_lanes);
// Lane i = bit i, higher lanes 0
T sum = ConvertScalarTo<T>(0);
// Avoid setting sign bit and cap so that f16 precision is not exceeded.
constexpr size_t kBits = HWY_MIN(sizeof(T) * 8 - 1, 9);
for (size_t i = 0; i < N; ++i) {
in_lanes[i] = ConvertScalarTo<T>(i < kBits ? 1ull << i : 0ull);
sum = AddWithWraparound(sum, in_lanes[i]);
}
HWY_ASSERT_VEC_EQ(d, Set(d, sum), SumOfLanes(d, Load(d, in_lanes.get())));
HWY_ASSERT_EQ(T(sum), ReduceSum(d, Load(d, in_lanes.get())));
// Lane i = i (iota) to include upper lanes
sum = ConvertScalarTo<T>(0);
for (size_t i = 0; i < N; ++i) {
sum = AddWithWraparound(sum, ConvertScalarTo<T>(i));
}
HWY_ASSERT_VEC_EQ(d, Set(d, sum), SumOfLanes(d, Iota(d, 0)));
HWY_ASSERT_EQ(T(sum), ReduceSum(d, Iota(d, 0)));
// Run more tests only for signed types with even vector lengths. Some of
// this code may not otherwise compile, so put it in a templated function.
SignedEvenLengthVectorTests(d);
}
};
HWY_NOINLINE void TestAllSumOfLanes() {
ForAllTypes(ForPartialVectors<TestSumOfLanes>());
}
struct TestMinOfLanes {
template <typename T, class D>
HWY_NOINLINE void operator()(T /*unused*/, D d) {
const size_t N = Lanes(d);
auto in_lanes = AllocateAligned<T>(N);
// Lane i = bit i, higher lanes = 2 (not the minimum)
T min = HighestValue<T>();
// Avoid setting sign bit and cap at double precision
constexpr size_t kBits = HWY_MIN(sizeof(T) * 8 - 1, 51);
for (size_t i = 0; i < N; ++i) {
in_lanes[i] = ConvertScalarTo<T>(i < kBits ? 1ull << i : 2ull);
min = HWY_MIN(min, in_lanes[i]);
}
HWY_ASSERT_VEC_EQ(d, Set(d, min), MinOfLanes(d, Load(d, in_lanes.get())));
// Lane i = N - i to include upper lanes
min = HighestValue<T>();
for (size_t i = 0; i < N; ++i) {
in_lanes[i] = ConvertScalarTo<T>(N - i); // no 8-bit T so no wraparound
min = HWY_MIN(min, in_lanes[i]);
}
HWY_ASSERT_VEC_EQ(d, Set(d, min), MinOfLanes(d, Load(d, in_lanes.get())));
// Bug #910: also check negative values
min = HighestValue<T>();
const T input_copy[] = {ConvertScalarTo<T>(-1),
ConvertScalarTo<T>(-2),
1,
2,
3,
4,
5,
6,
7,
8,
9,
10,
11,
12,
13,
14};
size_t i = 0;
for (; i < HWY_MIN(N, sizeof(input_copy) / sizeof(T)); ++i) {
in_lanes[i] = input_copy[i];
min = HWY_MIN(min, input_copy[i]);
}
// Pad with neutral element to full vector (so we can load)
for (; i < N; ++i) {
in_lanes[i] = min;
}
HWY_ASSERT_VEC_EQ(d, Set(d, min), MinOfLanes(d, Load(d, in_lanes.get())));
HWY_ASSERT_EQ(min, ReduceMin(d, Load(d, in_lanes.get())));
}
};
struct TestMaxOfLanes {
template <typename T, class D>
HWY_NOINLINE void operator()(T /*unused*/, D d) {
const size_t N = Lanes(d);
auto in_lanes = AllocateAligned<T>(N);
T max = LowestValue<T>();
// Avoid setting sign bit and cap at double precision
constexpr size_t kBits = HWY_MIN(sizeof(T) * 8 - 1, 51);
for (size_t i = 0; i < N; ++i) {
in_lanes[i] = ConvertScalarTo<T>(i < kBits ? 1ull << i : 0ull);
max = HWY_MAX(max, in_lanes[i]);
}
HWY_ASSERT_VEC_EQ(d, Set(d, max), MaxOfLanes(d, Load(d, in_lanes.get())));
// Lane i = i to include upper lanes
max = LowestValue<T>();
for (size_t i = 0; i < N; ++i) {
in_lanes[i] = ConvertScalarTo<T>(i); // no 8-bit T so no wraparound
max = HWY_MAX(max, in_lanes[i]);
}
HWY_ASSERT_VEC_EQ(d, Set(d, max), MaxOfLanes(d, Load(d, in_lanes.get())));
// Bug #910: also check negative values
max = LowestValue<T>();
const T input_copy[] = {ConvertScalarTo<T>(-1),
ConvertScalarTo<T>(-2),
1,
2,
3,
4,
5,
6,
7,
8,
9,
10,
11,
12,
13,
14};
size_t i = 0;
for (; i < HWY_MIN(N, sizeof(input_copy) / sizeof(T)); ++i) {
in_lanes[i] = input_copy[i];
max = HWY_MAX(max, in_lanes[i]);
}
// Pad with neutral element to full vector (so we can load)
for (; i < N; ++i) {
in_lanes[i] = max;
}
HWY_ASSERT_VEC_EQ(d, Set(d, max), MaxOfLanes(d, Load(d, in_lanes.get())));
HWY_ASSERT_EQ(max, ReduceMax(d, Load(d, in_lanes.get())));
}
};
HWY_NOINLINE void TestAllMinMaxOfLanes() {
ForAllTypes(ForPartialVectors<TestMinOfLanes>());
ForAllTypes(ForPartialVectors<TestMaxOfLanes>());
}
struct TestSumsOf2 {
template <typename T, class D>
HWY_NOINLINE void operator()(T /*unused*/, D d) {
RandomState rng;
using TW = MakeWide<T>;
const size_t N = Lanes(d);
if (N < 2) return;
const RepartitionToWide<D> dw;
auto in_lanes = AllocateAligned<T>(N);
auto sum_lanes = AllocateAligned<TW>(N / 2);
for (size_t rep = 0; rep < 100; ++rep) {
for (size_t i = 0; i < N; ++i) {
in_lanes[i] = RandomFiniteValue<T>(&rng);
}
for (size_t idx_sum = 0; idx_sum < N / 2; ++idx_sum) {
TW sum = static_cast<TW>(static_cast<TW>(in_lanes[idx_sum * 2]) +
static_cast<TW>(in_lanes[idx_sum * 2 + 1]));
sum_lanes[idx_sum] = sum;
}
const Vec<D> in = Load(d, in_lanes.get());
HWY_ASSERT_VEC_EQ(dw, sum_lanes.get(), SumsOf2(in));
}
}
};
HWY_NOINLINE void TestAllSumsOf2() {
ForGEVectors<16, TestSumsOf2>()(int8_t());
ForGEVectors<16, TestSumsOf2>()(uint8_t());
ForGEVectors<32, TestSumsOf2>()(int16_t());
ForGEVectors<32, TestSumsOf2>()(uint16_t());
#if HWY_HAVE_FLOAT16
ForGEVectors<32, TestSumsOf2>()(float16_t());
#endif
#if HWY_HAVE_INTEGER64
ForGEVectors<64, TestSumsOf2>()(int32_t());
ForGEVectors<64, TestSumsOf2>()(uint32_t());
#endif
#if HWY_HAVE_FLOAT64
ForGEVectors<64, TestSumsOf2>()(float());
#endif
}
struct TestSumsOf4 {
template <typename T, class D>
HWY_NOINLINE void operator()(T /*unused*/, D d) {
RandomState rng;
using TW = MakeWide<T>;
using TW2 = MakeWide<TW>;
const size_t N = Lanes(d);
if (N < 4) return;
const Repartition<TW2, D> dw2;
auto in_lanes = AllocateAligned<T>(N);
auto sum_lanes = AllocateAligned<TW2>(N / 4);
for (size_t rep = 0; rep < 100; ++rep) {
for (size_t i = 0; i < N; ++i) {
in_lanes[i] = RandomFiniteValue<T>(&rng);
}
for (size_t idx_sum = 0; idx_sum < N / 4; ++idx_sum) {
TW2 sum = static_cast<TW2>(static_cast<TW>(in_lanes[idx_sum * 4]) +
static_cast<TW>(in_lanes[idx_sum * 4 + 1]) +
static_cast<TW>(in_lanes[idx_sum * 4 + 2]) +
static_cast<TW>(in_lanes[idx_sum * 4 + 3]));
sum_lanes[idx_sum] = sum;
}
const Vec<D> in = Load(d, in_lanes.get());
HWY_ASSERT_VEC_EQ(dw2, sum_lanes.get(), SumsOf4(in));
}
}
};
HWY_NOINLINE void TestAllSumsOf4() {
ForGEVectors<32, TestSumsOf4>()(int8_t());
ForGEVectors<32, TestSumsOf4>()(uint8_t());
#if HWY_HAVE_INTEGER64
ForGEVectors<64, TestSumsOf4>()(int16_t());
ForGEVectors<64, TestSumsOf4>()(uint16_t());
#endif
}
struct TestSumsOf8 {
template <typename T, class D>
HWY_NOINLINE void operator()(T /*unused*/, D d) {
RandomState rng;
using TW = MakeWide<MakeWide<MakeWide<T>>>;
const size_t N = Lanes(d);
if (N < 8) return;
const Repartition<TW, D> d64;
auto in_lanes = AllocateAligned<T>(N);
auto sum_lanes = AllocateAligned<TW>(N / 8);
for (size_t rep = 0; rep < 100; ++rep) {
for (size_t i = 0; i < N; ++i) {
in_lanes[i] = ConvertScalarTo<T>(Random64(&rng) & 0xFF);
}
for (size_t idx_sum = 0; idx_sum < N / 8; ++idx_sum) {
TW sum = 0;
for (size_t i = 0; i < 8; ++i) {
sum += in_lanes[idx_sum * 8 + i];
}
sum_lanes[idx_sum] = sum;
}
const Vec<D> in = Load(d, in_lanes.get());
HWY_ASSERT_VEC_EQ(d64, sum_lanes.get(), SumsOf8(in));
}
}
};
HWY_NOINLINE void TestAllSumsOf8() {
ForGEVectors<64, TestSumsOf8>()(int8_t());
ForGEVectors<64, TestSumsOf8>()(uint8_t());
}
// NOLINTNEXTLINE(google-readability-namespace-comments)
} // namespace HWY_NAMESPACE
} // namespace hwy
HWY_AFTER_NAMESPACE();
#if HWY_ONCE
namespace hwy {
HWY_BEFORE_TEST(HwyReductionTest);
HWY_EXPORT_AND_TEST_P(HwyReductionTest, TestAllSumOfLanes);
HWY_EXPORT_AND_TEST_P(HwyReductionTest, TestAllMinMaxOfLanes);
HWY_EXPORT_AND_TEST_P(HwyReductionTest, TestAllSumsOf2);
HWY_EXPORT_AND_TEST_P(HwyReductionTest, TestAllSumsOf4);
HWY_EXPORT_AND_TEST_P(HwyReductionTest, TestAllSumsOf8);
} // namespace hwy
#endif