diff options
Diffstat (limited to 'src/bench/bench.cpp')
-rw-r--r-- | src/bench/bench.cpp | 39 |
1 files changed, 18 insertions, 21 deletions
diff --git a/src/bench/bench.cpp b/src/bench/bench.cpp index 33631d2d15..dd4ba5ab0e 100644 --- a/src/bench/bench.cpp +++ b/src/bench/bench.cpp @@ -8,29 +8,22 @@ #include <assert.h> #include <iostream> #include <iomanip> -#include <sys/time.h> benchmark::BenchRunner::BenchmarkMap &benchmark::BenchRunner::benchmarks() { static std::map<std::string, benchmark::BenchFunction> benchmarks_map; return benchmarks_map; } -static double gettimedouble(void) { - struct timeval tv; - gettimeofday(&tv, NULL); - return tv.tv_usec * 0.000001 + tv.tv_sec; -} - benchmark::BenchRunner::BenchRunner(std::string name, benchmark::BenchFunction func) { benchmarks().insert(std::make_pair(name, func)); } void -benchmark::BenchRunner::RunAll(double elapsedTimeForOne) +benchmark::BenchRunner::RunAll(benchmark::duration elapsedTimeForOne) { perf_init(); - std::cout << "#Benchmark" << "," << "count" << "," << "min" << "," << "max" << "," << "average" << "," + std::cout << "#Benchmark" << "," << "count" << "," << "min(ns)" << "," << "max(ns)" << "," << "average(ns)" << "," << "min_cycles" << "," << "max_cycles" << "," << "average_cycles" << "\n"; for (const auto &p: benchmarks()) { @@ -46,22 +39,23 @@ bool benchmark::State::KeepRunning() ++count; return true; } - double now; + time_point now; + uint64_t nowCycles; if (count == 0) { - lastTime = beginTime = now = gettimedouble(); + lastTime = beginTime = now = clock::now(); lastCycles = beginCycles = nowCycles = perf_cpucycles(); } else { - now = gettimedouble(); - double elapsed = now - lastTime; - double elapsedOne = elapsed * countMaskInv; + now = clock::now(); + auto elapsed = now - lastTime; + auto elapsedOne = elapsed / (countMask + 1); if (elapsedOne < minTime) minTime = elapsedOne; if (elapsedOne > maxTime) maxTime = elapsedOne; // We only use relative values, so don't have to handle 64-bit wrap-around specially nowCycles = perf_cpucycles(); - uint64_t elapsedOneCycles = (nowCycles - lastCycles) * countMaskInv; + uint64_t elapsedOneCycles = (nowCycles - lastCycles) / (countMask + 1); if (elapsedOneCycles < minCycles) minCycles = elapsedOneCycles; if (elapsedOneCycles > maxCycles) maxCycles = elapsedOneCycles; @@ -69,10 +63,9 @@ bool benchmark::State::KeepRunning() // If the execution was much too fast (1/128th of maxElapsed), increase the count mask by 8x and restart timing. // The restart avoids including the overhead of this code in the measurement. countMask = ((countMask<<3)|7) & ((1LL<<60)-1); - countMaskInv = 1./(countMask+1); count = 0; - minTime = std::numeric_limits<double>::max(); - maxTime = std::numeric_limits<double>::min(); + minTime = duration::max(); + maxTime = duration::zero(); minCycles = std::numeric_limits<uint64_t>::max(); maxCycles = std::numeric_limits<uint64_t>::min(); return true; @@ -81,7 +74,6 @@ bool benchmark::State::KeepRunning() uint64_t newCountMask = ((countMask<<1)|1) & ((1LL<<60)-1); if ((count & newCountMask)==0) { countMask = newCountMask; - countMaskInv = 1./(countMask+1); } } } @@ -96,10 +88,15 @@ bool benchmark::State::KeepRunning() assert(count != 0 && "count == 0 => (now == 0 && beginTime == 0) => return above"); // Output results - double average = (now-beginTime)/count; + // Duration casts are only necessary here because hardware with sub-nanosecond clocks + // will lose precision. + int64_t min_elapsed = std::chrono::duration_cast<std::chrono::nanoseconds>(minTime).count(); + int64_t max_elapsed = std::chrono::duration_cast<std::chrono::nanoseconds>(maxTime).count(); + int64_t avg_elapsed = std::chrono::duration_cast<std::chrono::nanoseconds>((now-beginTime)/count).count(); int64_t averageCycles = (nowCycles-beginCycles)/count; - std::cout << std::fixed << std::setprecision(15) << name << "," << count << "," << minTime << "," << maxTime << "," << average << "," + std::cout << std::fixed << std::setprecision(15) << name << "," << count << "," << min_elapsed << "," << max_elapsed << "," << avg_elapsed << "," << minCycles << "," << maxCycles << "," << averageCycles << "\n"; + std::cout.copyfmt(std::ios(nullptr)); return false; } |