// Copyright 2013 the V8 project authors. All rights reserved. // Copyright (C) 2015-2017 Apple Inc. All rights reserved. // 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 Google Inc. 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. // Performance.now is used in latency benchmarks, the fallback is Date.now. var performance = performance || {}; performance.now = () => preciseTime() * 1000; // Simple framework for running the benchmark suites and // computing a score based on the timing measurements. // A benchmark has a name (string) and a function that will be run to // do the performance measurement. The optional setup and tearDown // arguments are functions that will be invoked before and after // running the benchmark, but the running time of these functions will // not be accounted for in the benchmark score. function Benchmark(name, doWarmup, doDeterministic, run, setup, tearDown, latencyResult, minIterations) { this.name = name; this.doWarmup = doWarmup; this.doDeterministic = doDeterministic; this.run = run; this.Setup = setup ? setup : function() { }; this.TearDown = tearDown ? tearDown : function() { }; this.latencyResult = latencyResult ? latencyResult : null; this.minIterations = minIterations ? minIterations : 32; } // Benchmark results hold the benchmark and the measured time used to // run the benchmark. The benchmark score is computed later once a // full benchmark suite has run to completion. If latency is set to 0 // then there is no latency score for this benchmark. function BenchmarkResult(benchmark, time, latency) { this.benchmark = benchmark; this.time = time; this.latency = latency; } // Automatically convert results to numbers. Used by the geometric // mean computation. BenchmarkResult.prototype.valueOf = function() { return this.time; } // Suites of benchmarks consist of a name and the set of benchmarks in // addition to the reference timing that the final score will be based // on. This way, all scores are relative to a reference run and higher // scores implies better performance. function BenchmarkSuite(name, reference, benchmarks) { this.name = name; this.reference = reference; this.benchmarks = benchmarks; BenchmarkSuite.suites.push(this); } // Keep track of all declared benchmark suites. BenchmarkSuite.suites = []; // Scores are not comparable across versions. Bump the version if // you're making changes that will affect that scores, e.g. if you add // a new benchmark or change an existing one. BenchmarkSuite.version = '9'; // Override the alert function to throw an exception instead. alert = function(s) { throw "Alert called with argument: " + s; }; // To make the benchmark results predictable, we replace Math.random // with a 100% deterministic alternative. BenchmarkSuite.ResetRNG = function() { Math.random = (function() { var seed = 49734321; return function() { // Robert Jenkins' 32 bit integer hash function. seed = ((seed + 0x7ed55d16) + (seed << 12)) & 0xffffffff; seed = ((seed ^ 0xc761c23c) ^ (seed >>> 19)) & 0xffffffff; seed = ((seed + 0x165667b1) + (seed << 5)) & 0xffffffff; seed = ((seed + 0xd3a2646c) ^ (seed << 9)) & 0xffffffff; seed = ((seed + 0xfd7046c5) + (seed << 3)) & 0xffffffff; seed = ((seed ^ 0xb55a4f09) ^ (seed >>> 16)) & 0xffffffff; return (seed & 0xfffffff) / 0x10000000; }; })(); } // Runs all registered benchmark suites and optionally yields between // each individual benchmark to avoid running for too long in the // context of browsers. Once done, the final score is reported to the // runner. BenchmarkSuite.RunSuites = function(runner) { var continuation = null; var suites = BenchmarkSuite.suites; var length = suites.length; BenchmarkSuite.scores = []; var index = 0; function RunStep() { while (continuation || index < length) { if (continuation) { continuation = continuation(); } else { var suite = suites[index++]; if (runner.NotifyStart) runner.NotifyStart(suite.name); continuation = suite.RunStep(runner); } if (continuation && typeof window != 'undefined' && window.setTimeout) { window.setTimeout(RunStep, 25); return; } } // show final result if (runner.NotifyScore) { var score = BenchmarkSuite.GeometricMean(BenchmarkSuite.scores); var formatted = BenchmarkSuite.FormatScore(100 * score); runner.NotifyScore(formatted); } } RunStep(); } // Counts the total number of registered benchmarks. Useful for // showing progress as a percentage. BenchmarkSuite.CountBenchmarks = function() { var result = 0; var suites = BenchmarkSuite.suites; for (var i = 0; i < suites.length; i++) { result += suites[i].benchmarks.length; } return result; } // Computes the geometric mean of a set of numbers. BenchmarkSuite.GeometricMean = function(numbers) { var log = 0; for (var i = 0; i < numbers.length; i++) { log += Math.log(numbers[i]); } return Math.pow(Math.E, log / numbers.length); } // Computes the geometric mean of a set of throughput time measurements. BenchmarkSuite.GeometricMeanTime = function(measurements) { var log = 0; for (var i = 0; i < measurements.length; i++) { log += Math.log(measurements[i].time); } return Math.pow(Math.E, log / measurements.length); } // Computes the average of the worst samples. For example, if percentile is 99, this will report the // average of the worst 1% of the samples. BenchmarkSuite.AverageAbovePercentile = function(numbers, percentile) { // Don't change the original array. numbers = numbers.slice(); // Sort in ascending order. numbers.sort(function(a, b) { return a - b; }); // Now the elements we want are at the end. Keep removing them until the array size shrinks too much. // Examples assuming percentile = 99: // // - numbers.length starts at 100: we will remove just the worst entry and then not remove anymore, // since then numbers.length / originalLength = 0.99. // // - numbers.length starts at 1000: we will remove the ten worst. // // - numbers.length starts at 10: we will remove just the worst. var numbersWeWant = []; var originalLength = numbers.length; while (numbers.length / originalLength > percentile / 100) numbersWeWant.push(numbers.pop()); var sum = 0; for (var i = 0; i < numbersWeWant.length; ++i) sum += numbersWeWant[i]; var result = sum / numbersWeWant.length; // Do a sanity check. if (numbers.length && result < numbers[numbers.length - 1]) { throw "Sanity check fail: the worst case result is " + result + " but we didn't take into account " + numbers; } return result; } // Computes the geometric mean of a set of latency measurements. BenchmarkSuite.GeometricMeanLatency = function(measurements) { var log = 0; var hasLatencyResult = false; for (var i = 0; i < measurements.length; i++) { if (measurements[i].latency != 0) { log += Math.log(measurements[i].latency); hasLatencyResult = true; } } if (hasLatencyResult) { return Math.pow(Math.E, log / measurements.length); } else { return 0; } } // Converts a score value to a string with at least three significant // digits. BenchmarkSuite.FormatScore = function(value) { if (value > 100) { return value.toFixed(0); } else { return value.toPrecision(3); } } // Notifies the runner that we're done running a single benchmark in // the benchmark suite. This can be useful to report progress. BenchmarkSuite.prototype.NotifyStep = function(result) { this.results.push(result); if (this.runner.NotifyStep) this.runner.NotifyStep(result.benchmark.name); } // Notifies the runner that we're done with running a suite and that // we have a result which can be reported to the user if needed. BenchmarkSuite.prototype.NotifyResult = function() { var mean = BenchmarkSuite.GeometricMeanTime(this.results); var score = this.reference[0] / mean; BenchmarkSuite.scores.push(score); if (this.runner.NotifyResult) { var formatted = BenchmarkSuite.FormatScore(100 * score); this.runner.NotifyResult(this.name, formatted); } if (this.reference.length == 2) { var meanLatency = BenchmarkSuite.GeometricMeanLatency(this.results); if (meanLatency != 0) { var scoreLatency = this.reference[1] / meanLatency; BenchmarkSuite.scores.push(scoreLatency); if (this.runner.NotifyResult) { var formattedLatency = BenchmarkSuite.FormatScore(100 * scoreLatency) this.runner.NotifyResult(this.name + "Latency", formattedLatency); } } } } // Notifies the runner that running a benchmark resulted in an error. BenchmarkSuite.prototype.NotifyError = function(error) { if (this.runner.NotifyError) { this.runner.NotifyError(this.name, error); } if (this.runner.NotifyStep) { this.runner.NotifyStep(this.name); } } // Runs a single benchmark for at least a second and computes the // average time it takes to run a single iteration. BenchmarkSuite.prototype.RunSingleBenchmark = function(benchmark, data) { function Measure(data) { var elapsed = 0; var start = new Date(); // Run either for 1 second or for the number of iterations specified // by minIterations, depending on the config flag doDeterministic. for (var i = 0; (benchmark.doDeterministic ? i= keys[i + 1]) { throw new Error("Splay tree not sorted"); } } } function SplayRun() { // Replace a few nodes in the splay tree. for (var i = 0; i < kSplayTreeModifications; i++) { var key = InsertNewNode(); var greatest = splayTree.findGreatestLessThan(key); if (greatest == null) splayTree.remove(key); else splayTree.remove(greatest.key); } SplayUpdateStats(performance.now()); } /** * Constructs a Splay tree. A splay tree is a self-balancing binary * search tree with the additional property that recently accessed * elements are quick to access again. It performs basic operations * such as insertion, look-up and removal in O(log(n)) amortized time. * * @constructor */ function SplayTree() { }; /** * Pointer to the root node of the tree. * * @type {SplayTree.Node} * @private */ SplayTree.prototype.root_ = null; /** * @return {boolean} Whether the tree is empty. */ SplayTree.prototype.isEmpty = function() { return !this.root_; }; /** * Inserts a node into the tree with the specified key and value if * the tree does not already contain a node with the specified key. If * the value is inserted, it becomes the root of the tree. * * @param {number} key Key to insert into the tree. * @param {*} value Value to insert into the tree. */ SplayTree.prototype.insert = function(key, value) { if (this.isEmpty()) { this.root_ = new SplayTree.Node(key, value); return; } // Splay on the key to move the last node on the search path for // the key to the root of the tree. this.splay_(key); if (this.root_.key == key) { return; } var node = new SplayTree.Node(key, value); if (key > this.root_.key) { node.left = this.root_; node.right = this.root_.right; this.root_.right = null; } else { node.right = this.root_; node.left = this.root_.left; this.root_.left = null; } this.root_ = node; }; /** * Removes a node with the specified key from the tree if the tree * contains a node with this key. The removed node is returned. If the * key is not found, an exception is thrown. * * @param {number} key Key to find and remove from the tree. * @return {SplayTree.Node} The removed node. */ SplayTree.prototype.remove = function(key) { if (this.isEmpty()) { throw Error('Key not found: ' + key); } this.splay_(key); if (this.root_.key != key) { throw Error('Key not found: ' + key); } var removed = this.root_; if (!this.root_.left) { this.root_ = this.root_.right; } else { var right = this.root_.right; this.root_ = this.root_.left; // Splay to make sure that the new root has an empty right child. this.splay_(key); // Insert the original right child as the right child of the new // root. this.root_.right = right; } return removed; }; /** * Returns the node having the specified key or null if the tree doesn't contain * a node with the specified key. * * @param {number} key Key to find in the tree. * @return {SplayTree.Node} Node having the specified key. */ SplayTree.prototype.find = function(key) { if (this.isEmpty()) { return null; } this.splay_(key); return this.root_.key == key ? this.root_ : null; }; /** * @return {SplayTree.Node} Node having the maximum key value. */ SplayTree.prototype.findMax = function(opt_startNode) { if (this.isEmpty()) { return null; } var current = opt_startNode || this.root_; while (current.right) { current = current.right; } return current; }; /** * @return {SplayTree.Node} Node having the maximum key value that * is less than the specified key value. */ SplayTree.prototype.findGreatestLessThan = function(key) { if (this.isEmpty()) { return null; } // Splay on the key to move the node with the given key or the last // node on the search path to the top of the tree. this.splay_(key); // Now the result is either the root node or the greatest node in // the left subtree. if (this.root_.key < key) { return this.root_; } else if (this.root_.left) { return this.findMax(this.root_.left); } else { return null; } }; /** * @return {Array<*>} An array containing all the keys of tree's nodes. */ SplayTree.prototype.exportKeys = function() { var result = []; if (!this.isEmpty()) { this.root_.traverse_(function(node) { result.push(node.key); }); } return result; }; /** * Perform the splay operation for the given key. Moves the node with * the given key to the top of the tree. If no node has the given * key, the last node on the search path is moved to the top of the * tree. This is the simplified top-down splaying algorithm from: * "Self-adjusting Binary Search Trees" by Sleator and Tarjan * * @param {number} key Key to splay the tree on. * @private */ SplayTree.prototype.splay_ = function(key) { if (this.isEmpty()) { return; } // Create a dummy node. The use of the dummy node is a bit // counter-intuitive: The right child of the dummy node will hold // the L tree of the algorithm. The left child of the dummy node // will hold the R tree of the algorithm. Using a dummy node, left // and right will always be nodes and we avoid special cases. var dummy, left, right; dummy = left = right = new SplayTree.Node(null, null); var current = this.root_; while (true) { if (key < current.key) { if (!current.left) { break; } if (key < current.left.key) { // Rotate right. var tmp = current.left; current.left = tmp.right; tmp.right = current; current = tmp; if (!current.left) { break; } } // Link right. right.left = current; right = current; current = current.left; } else if (key > current.key) { if (!current.right) { break; } if (key > current.right.key) { // Rotate left. var tmp = current.right; current.right = tmp.left; tmp.left = current; current = tmp; if (!current.right) { break; } } // Link left. left.right = current; left = current; current = current.right; } else { break; } } // Assemble. left.right = current.left; right.left = current.right; current.left = dummy.right; current.right = dummy.left; this.root_ = current; }; /** * Constructs a Splay tree node. * * @param {number} key Key. * @param {*} value Value. */ SplayTree.Node = function(key, value) { this.key = key; this.value = value; }; /** * @type {SplayTree.Node} */ SplayTree.Node.prototype.left = null; /** * @type {SplayTree.Node} */ SplayTree.Node.prototype.right = null; /** * Performs an ordered traversal of the subtree starting at * this SplayTree.Node. * * @param {function(SplayTree.Node)} f Visitor function. * @private */ SplayTree.Node.prototype.traverse_ = function(f) { var current = this; while (current) { var left = current.left; if (left) left.traverse_(f); f(current); current = current.right; } }; function report(msg) { } function start(resultObject) { SplaySetup(); var samples = []; var before = performance.now(); for (var i = 0; i < 10000; ++i) { SplayRun(); var after = performance.now(); samples.push(after - before); before = after; } SplayTearDown(); var scatterData = []; for (var i = 0; i < samples.length; ++i) scatterData.push({x: i + 1, y: samples[i]}); report("JetStream-like Latency Score: " + Math.round(4000 / BenchmarkSuite.AverageAbovePercentile(samples, 99.5))); var sumOfSquares = 0; for (var i = 0; i < samples.length; ++i) sumOfSquares += samples[i] * samples[i]; report("Octane-like Latency Score: " + Math.round(27395.14 / Math.sqrt(sumOfSquares / samples.length))); for (var percentile of [99.5, 95, 87, 75, 50, 0]) report("Average above " + percentile + "%: " + BenchmarkSuite.AverageAbovePercentile(samples, percentile)); resultObject.value = BenchmarkSuite.AverageAbovePercentile(samples, 99.5); } start(arguments[0]);