benchmark/test/perf_counters_gtest.cc

308 lines
11 KiB
C++
Raw Normal View History

#include <random>
#include <thread>
#include "../src/perf_counters.h"
#include "gmock/gmock.h"
#include "gtest/gtest.h"
#ifndef GTEST_SKIP
struct MsgHandler {
2021-11-10 16:22:31 +00:00
void operator=(std::ostream&) {}
};
#define GTEST_SKIP() return MsgHandler() = std::cout
#endif
using benchmark::internal::PerfCounters;
using benchmark::internal::PerfCountersMeasurement;
using benchmark::internal::PerfCounterValues;
using ::testing::AllOf;
using ::testing::Gt;
using ::testing::Lt;
namespace {
const char kGenericPerfEvent1[] = "CYCLES";
const char kGenericPerfEvent2[] = "INSTRUCTIONS";
TEST(PerfCountersTest, Init) {
EXPECT_EQ(PerfCounters::Initialize(), PerfCounters::kSupported);
}
TEST(PerfCountersTest, OneCounter) {
if (!PerfCounters::kSupported) {
GTEST_SKIP() << "Performance counters not supported.\n";
}
EXPECT_TRUE(PerfCounters::Initialize());
EXPECT_EQ(PerfCounters::Create({kGenericPerfEvent1}).num_counters(), 1);
}
TEST(PerfCountersTest, NegativeTest) {
if (!PerfCounters::kSupported) {
EXPECT_FALSE(PerfCounters::Initialize());
return;
}
EXPECT_TRUE(PerfCounters::Initialize());
// Sanity checks
// Create() will always create a valid object, even if passed no or
// wrong arguments as the new behavior is to warn and drop unsupported
// counters
EXPECT_EQ(PerfCounters::Create({}).num_counters(), 0);
EXPECT_EQ(PerfCounters::Create({""}).num_counters(), 0);
EXPECT_EQ(PerfCounters::Create({"not a counter name"}).num_counters(), 0);
{
// Try sneaking in a bad egg to see if it is filtered out. The
// number of counters has to be two, not zero
auto counter =
PerfCounters::Create({kGenericPerfEvent2, "", kGenericPerfEvent1});
EXPECT_EQ(counter.num_counters(), 2);
EXPECT_EQ(counter.names(), std::vector<std::string>(
{kGenericPerfEvent2, kGenericPerfEvent1}));
}
{
// Try sneaking in an outrageous counter, like a fat finger mistake
auto counter = PerfCounters::Create(
{kGenericPerfEvent2, "not a counter name", kGenericPerfEvent1});
EXPECT_EQ(counter.num_counters(), 2);
EXPECT_EQ(counter.names(), std::vector<std::string>(
{kGenericPerfEvent2, kGenericPerfEvent1}));
}
{
// Finally try a golden input - it should like both of them
EXPECT_EQ(PerfCounters::Create({kGenericPerfEvent1, kGenericPerfEvent2})
.num_counters(),
2);
}
{
// Add a bad apple in the end of the chain to check the edges
auto counter = PerfCounters::Create(
{kGenericPerfEvent1, kGenericPerfEvent2, "bad event name"});
EXPECT_EQ(counter.num_counters(), 2);
EXPECT_EQ(counter.names(), std::vector<std::string>(
{kGenericPerfEvent1, kGenericPerfEvent2}));
}
}
TEST(PerfCountersTest, Read1Counter) {
if (!PerfCounters::kSupported) {
GTEST_SKIP() << "Test skipped because libpfm is not supported.\n";
}
EXPECT_TRUE(PerfCounters::Initialize());
auto counters = PerfCounters::Create({kGenericPerfEvent1});
EXPECT_EQ(counters.num_counters(), 1);
PerfCounterValues values1(1);
EXPECT_TRUE(counters.Snapshot(&values1));
EXPECT_GT(values1[0], 0);
PerfCounterValues values2(1);
EXPECT_TRUE(counters.Snapshot(&values2));
EXPECT_GT(values2[0], 0);
EXPECT_GT(values2[0], values1[0]);
}
TEST(PerfCountersTest, Read2Counters) {
if (!PerfCounters::kSupported) {
GTEST_SKIP() << "Test skipped because libpfm is not supported.\n";
}
EXPECT_TRUE(PerfCounters::Initialize());
auto counters =
PerfCounters::Create({kGenericPerfEvent1, kGenericPerfEvent2});
EXPECT_EQ(counters.num_counters(), 2);
PerfCounterValues values1(2);
EXPECT_TRUE(counters.Snapshot(&values1));
EXPECT_GT(values1[0], 0);
EXPECT_GT(values1[1], 0);
PerfCounterValues values2(2);
EXPECT_TRUE(counters.Snapshot(&values2));
EXPECT_GT(values2[0], 0);
EXPECT_GT(values2[1], 0);
}
TEST(PerfCountersTest, ReopenExistingCounters) {
// This test works in recent and old Intel hardware, Pixel 3, and Pixel 6.
// However we cannot make assumptions beyond 2 HW counters due to Pixel 6.
if (!PerfCounters::kSupported) {
GTEST_SKIP() << "Test skipped because libpfm is not supported.\n";
}
EXPECT_TRUE(PerfCounters::Initialize());
std::vector<std::string> kMetrics({kGenericPerfEvent1});
std::vector<PerfCounters> counters(2);
for (auto& counter : counters) {
counter = PerfCounters::Create(kMetrics);
}
PerfCounterValues values(1);
EXPECT_TRUE(counters[0].Snapshot(&values));
EXPECT_TRUE(counters[1].Snapshot(&values));
}
TEST(PerfCountersTest, CreateExistingMeasurements) {
// The test works (i.e. causes read to fail) for the assumptions
// about hardware capabilities (i.e. small number (2) hardware
// counters) at this date,
// the same as previous test ReopenExistingCounters.
if (!PerfCounters::kSupported) {
GTEST_SKIP() << "Test skipped because libpfm is not supported.\n";
}
EXPECT_TRUE(PerfCounters::Initialize());
// This means we will try 10 counters but we can only guarantee
// for sure at this time that only 3 will work. Perhaps in the future
// we could use libpfm to query for the hardware limits on this
// particular platform.
const int kMaxCounters = 10;
const int kMinValidCounters = 2;
// Let's use a ubiquitous counter that is guaranteed to work
// on all platforms
const std::vector<std::string> kMetrics{"cycles"};
// Cannot create a vector of actual objects because the
// copy constructor of PerfCounters is deleted - and so is
// implicitly deleted on PerfCountersMeasurement too
std::vector<std::unique_ptr<PerfCountersMeasurement>>
perf_counter_measurements;
perf_counter_measurements.reserve(kMaxCounters);
for (int j = 0; j < kMaxCounters; ++j) {
perf_counter_measurements.emplace_back(
new PerfCountersMeasurement(kMetrics));
}
std::vector<std::pair<std::string, double>> measurements;
// Start all counters together to see if they hold
size_t max_counters = kMaxCounters;
for (size_t i = 0; i < kMaxCounters; ++i) {
auto& counter(*perf_counter_measurements[i]);
EXPECT_EQ(counter.num_counters(), 1);
if (!counter.Start()) {
max_counters = i;
break;
};
}
ASSERT_GE(max_counters, kMinValidCounters);
// Start all together
for (size_t i = 0; i < max_counters; ++i) {
auto& counter(*perf_counter_measurements[i]);
EXPECT_TRUE(counter.Stop(measurements) || (i >= kMinValidCounters));
}
// Start/stop individually
for (size_t i = 0; i < max_counters; ++i) {
auto& counter(*perf_counter_measurements[i]);
measurements.clear();
counter.Start();
EXPECT_TRUE(counter.Stop(measurements) || (i >= kMinValidCounters));
}
}
// We try to do some meaningful work here but the compiler
// insists in optimizing away our loop so we had to add a
// no-optimize macro. In case it fails, we added some entropy
// to this pool as well.
BENCHMARK_DONT_OPTIMIZE size_t do_work() {
static std::mt19937 rd{std::random_device{}()};
static std::uniform_int_distribution<size_t> mrand(0, 10);
const size_t kNumLoops = 1000000;
size_t sum = 0;
for (size_t j = 0; j < kNumLoops; ++j) {
sum += mrand(rd);
}
benchmark::DoNotOptimize(sum);
return sum;
}
void measure(size_t threadcount, PerfCounterValues* before,
PerfCounterValues* after) {
BM_CHECK_NE(before, nullptr);
BM_CHECK_NE(after, nullptr);
std::vector<std::thread> threads(threadcount);
2021-06-24 17:21:59 +00:00
auto work = [&]() { BM_CHECK(do_work() > 1000); };
// We need to first set up the counters, then start the threads, so the
// threads would inherit the counters. But later, we need to first destroy
// the thread pool (so all the work finishes), then measure the counters. So
// the scopes overlap, and we need to explicitly control the scope of the
// threadpool.
auto counters =
PerfCounters::Create({kGenericPerfEvent1, kGenericPerfEvent2});
for (auto& t : threads) t = std::thread(work);
counters.Snapshot(before);
for (auto& t : threads) t.join();
counters.Snapshot(after);
}
TEST(PerfCountersTest, MultiThreaded) {
if (!PerfCounters::kSupported) {
GTEST_SKIP() << "Test skipped because libpfm is not supported.";
}
EXPECT_TRUE(PerfCounters::Initialize());
PerfCounterValues before(2);
PerfCounterValues after(2);
// Notice that this test will work even if we taskset it to a single CPU
// In this case the threads will run sequentially
// Start two threads and measure the number of combined cycles and
// instructions
measure(2, &before, &after);
std::vector<double> Elapsed2Threads{
static_cast<double>(after[0] - before[0]),
static_cast<double>(after[1] - before[1])};
// Start four threads and measure the number of combined cycles and
// instructions
measure(4, &before, &after);
std::vector<double> Elapsed4Threads{
static_cast<double>(after[0] - before[0]),
static_cast<double>(after[1] - before[1])};
// The following expectations fail (at least on a beefy workstation with lots
// of cpus) - it seems that in some circumstances the runtime of 4 threads
// can even be better than with 2.
// So instead of expecting 4 threads to be slower, let's just make sure they
// do not differ too much in general (one is not more than 10x than the
// other).
EXPECT_THAT(Elapsed4Threads[0] / Elapsed2Threads[0], AllOf(Gt(0.1), Lt(10)));
EXPECT_THAT(Elapsed4Threads[1] / Elapsed2Threads[1], AllOf(Gt(0.1), Lt(10)));
}
Implement unlimited number of performance counters (#1552) * Implement unlimited number of performance counters Linux performance counters will limit the number of hardware counters per reading group. For that reason the implementation of PerfCounters is limited to 3. However if only software counters are added, there is no reason to limit the counters. For hardware counters, we create multiple groups and store a vector or leaders in the PerfCounters object. When reading, there is an extra time waste by iterating through all the group leaders. However this should be the same performance as with today. Reading is done by groups and it had to be heavily adjusted with the logic being moved to PerfCounterValues. I created a test for x86-64 and took care of filtering out the events in case it runs in a platform that does not support those counters - the test will not fail. The current tests were already failing (ReOpenExistingCounters, CreateExistingMeasurements and MultiThreaded) on the main branch and they continue to fail after this implementation - I did not fix those not to conflate all here. * Moved the PerfCounterValues::Read() implementation from header to body. * Added missing implementation of PerfCounters::IsCounterSupported when HAVE_LIBPFM is not defined. * Changed comments to reflect the implementation * Removed arg name so it does not generate an error when HAVE_LIBPBM is not defined. * Made loop counter a const reference for clang-tidy * Added missig BENCHMARK_EXPORT to PerfCounterValues
2023-03-01 15:30:41 +00:00
TEST(PerfCountersTest, HardwareLimits) {
// The test works (i.e. causes read to fail) for the assumptions
// about hardware capabilities (i.e. small number (3-4) hardware
// counters) at this date,
// the same as previous test ReopenExistingCounters.
if (!PerfCounters::kSupported) {
GTEST_SKIP() << "Test skipped because libpfm is not supported.\n";
}
EXPECT_TRUE(PerfCounters::Initialize());
// Taken from `perf list`, but focusses only on those HW events that actually
// were reported when running `sudo perf stat -a sleep 10`, intersected over
// several platforms. All HW events listed in the first command not reported
// in the second seem to not work. This is sad as we don't really get to test
// the grouping here (groups can contain up to 6 members)...
std::vector<std::string> counter_names{
"cycles", // leader
"instructions", //
"branch-misses", //
};
Implement unlimited number of performance counters (#1552) * Implement unlimited number of performance counters Linux performance counters will limit the number of hardware counters per reading group. For that reason the implementation of PerfCounters is limited to 3. However if only software counters are added, there is no reason to limit the counters. For hardware counters, we create multiple groups and store a vector or leaders in the PerfCounters object. When reading, there is an extra time waste by iterating through all the group leaders. However this should be the same performance as with today. Reading is done by groups and it had to be heavily adjusted with the logic being moved to PerfCounterValues. I created a test for x86-64 and took care of filtering out the events in case it runs in a platform that does not support those counters - the test will not fail. The current tests were already failing (ReOpenExistingCounters, CreateExistingMeasurements and MultiThreaded) on the main branch and they continue to fail after this implementation - I did not fix those not to conflate all here. * Moved the PerfCounterValues::Read() implementation from header to body. * Added missing implementation of PerfCounters::IsCounterSupported when HAVE_LIBPFM is not defined. * Changed comments to reflect the implementation * Removed arg name so it does not generate an error when HAVE_LIBPBM is not defined. * Made loop counter a const reference for clang-tidy * Added missig BENCHMARK_EXPORT to PerfCounterValues
2023-03-01 15:30:41 +00:00
// In the off-chance that some of these values are not supported,
// we filter them out so the test will complete without failure
// albeit it might not actually test the grouping on that platform
std::vector<std::string> valid_names;
for (const std::string& name : counter_names) {
if (PerfCounters::IsCounterSupported(name)) {
valid_names.push_back(name);
}
}
PerfCountersMeasurement counter(valid_names);
std::vector<std::pair<std::string, double>> measurements;
counter.Start();
EXPECT_TRUE(counter.Stop(measurements));
}
} // namespace