/*
* Copyright (c) Meta Platforms, Inc. and affiliates.
*
* 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
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* 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 <random>
#include <folly/Benchmark.h>
#include <folly/Fingerprint.h>
#include <folly/Format.h>
#include <folly/detail/SlowFingerprint.h>
using namespace folly;
using folly::detail::SlowFingerprint;
namespace {
constexpr int kMaxIds = 64 << 10; // 64Ki
constexpr int kMaxTerms = 64 << 10;
// Globals are generally bad, but this is a benchmark, so there.
uint64_t ids[kMaxIds];
std::string terms[kMaxTerms];
void initialize() {
std::mt19937 rng;
for (int i = 0; i < kMaxIds; i++) {
ids[i] = (((uint64_t)rng()) << 32) | rng();
}
// Use randomly generated words. These numbers are out of my hat and
// probably wrong.
// word length = uniformly distributed between 1 and 10
// charset = 0x20 - 0x7f
std::uniform_int_distribution<size_t> term_len(1, 10);
std::uniform_int_distribution<uint16_t> term_char(0x20, 0x7f);
for (int i = 0; i < kMaxTerms; i++) {
std::string& term = terms[i];
int len = term_len(rng);
term.reserve(len);
for (int j = 0; j < len; j++) {
term.append(1, (char)term_char(rng));
}
}
}
template <template <int> class T, int Bits>
constexpr size_t fpBits(tag_t<T<Bits>>) {
return Bits;
}
template <class FP>
void fingerprintIds(int num_iterations, int num_ids) {
for (int iter = 0; iter < num_iterations; iter++) {
FP fp;
for (int i = 0; i < num_ids; i++) {
fp.update64(ids[i]);
}
// GOTCHA: if we don't actually call write(), compiler optimizes
// away the inner loop!
uint64_t out[fpBits(tag<FP>) + 63 / 64]; // ceil(bits / 64.0)
fp.write(out);
compiler_must_not_elide(out);
}
}
template <class FP>
void fingerprintTerms(int num_iterations, int num_terms) {
for (int iter = 0; iter < num_iterations; iter++) {
FP fp;
for (int i = 0; i < num_terms; i++) {
fp.update(terms[i]);
}
// GOTCHA: if we don't actually call write(), compiler optimizes
// away the inner loop!
uint64_t out[fpBits(tag<FP>) + 63 / 64]; // ceil(bits / 64.0)
fp.write(out);
compiler_must_not_elide(out);
}
}
void fastFingerprintIds64(int num_iterations, int num_ids) {
fingerprintIds<Fingerprint<64>>(num_iterations, num_ids);
}
void slowFingerprintIds64(int num_iterations, int num_ids) {
fingerprintIds<SlowFingerprint<64>>(num_iterations, num_ids);
}
void fastFingerprintIds96(int num_iterations, int num_ids) {
fingerprintIds<Fingerprint<96>>(num_iterations, num_ids);
}
void fastFingerprintIds128(int num_iterations, int num_ids) {
fingerprintIds<Fingerprint<128>>(num_iterations, num_ids);
}
void fastFingerprintTerms64(int num_iterations, int num_ids) {
fingerprintTerms<Fingerprint<64>>(num_iterations, num_ids);
}
void slowFingerprintTerms64(int num_iterations, int num_ids) {
fingerprintTerms<SlowFingerprint<64>>(num_iterations, num_ids);
}
void fastFingerprintTerms96(int num_iterations, int num_ids) {
fingerprintTerms<Fingerprint<96>>(num_iterations, num_ids);
}
void fastFingerprintTerms128(int num_iterations, int num_ids) {
fingerprintTerms<Fingerprint<128>>(num_iterations, num_ids);
}
} // namespace
// Only benchmark one size of slowFingerprint; it's significantly slower
// than fastFingeprint (as you can see for 64 bits) and it just slows down
// the benchmark without providing any useful data.
int main(int argc, char** argv) {
gflags::ParseCommandLineFlags(&argc, &argv, true);
#define BM(name, min, max) \
for (size_t i = min; i <= max; i *= 2) { \
addBenchmark( \
__FILE__, sformat("{}_{}", #name, i).c_str(), [=](int iters) { \
name(iters, i); \
return iters; \
}); \
}
BM(fastFingerprintIds64, 1, kMaxIds)
BM(slowFingerprintIds64, 1, kMaxIds)
BM(fastFingerprintIds96, 1, kMaxIds)
BM(fastFingerprintIds128, 1, kMaxIds)
BM(fastFingerprintTerms64, 1, kMaxTerms)
BM(slowFingerprintTerms64, 1, kMaxTerms)
BM(fastFingerprintTerms96, 1, kMaxTerms)
BM(fastFingerprintTerms128, 1, kMaxTerms)
#undef BM
initialize();
runBenchmarks();
return 0;
}