Modules

node_modules/bytebuffer/tests/bench.js

@@ -0,0 +1,99 @@
+// a simple benchmark utility comparing performance between the different builds.
+
+// Uint8Array observations (compared to node Buffers)
+// - seems to be pretty much equal for byte ops
+
+// DataView observations (compared to Uint8Array):
+// - allocation is about 2 times slower
+// - writing is about 5 times (int32) to 10 times (varint) slower
+// - reading is about 3 times slower
+
+// UTF8 encoding observations (compared to node's string/buffer API)
+// - the longer the string, the poorer is read/write performance
+// - either utfx doesn't cut it yet, or node's bindings simply outperform everything js here
+
+var ByteBuffer = require("../index.js"),
+    prettyHrTime = require("pretty-hrtime");
+
+var impls = [
+    { name: "ByteBufferNB", impl: ByteBuffer.ByteBufferNB },
+    { name: "ByteBufferAB", impl: ByteBuffer.ByteBufferAB },
+    { name: "ByteBufferAB_DataView", impl: ByteBuffer.ByteBufferAB_DataView }
+];
+
+var bench = {};
+
+bench["allocate"] = function(ByteBuffer, n) {
+    n = n || 10000;
+    for (var i=0; i<n; ++i)
+        new ByteBuffer();
+    return n;
+};
+
+bench["writeInt32"] = function(ByteBuffer, n) {
+    n = n || 1000000;
+    var bb = new ByteBuffer(4);
+    for (var i=0; i<n; ++i)
+        bb.writeInt32(0x7fffffff, 0);
+    return n;
+};
+
+bench["readInt32"] = function(ByteBuffer, n) {
+    n = n || 1000000;
+    var bb = new ByteBuffer(4).writeInt32(0x7fffffff).flip();
+    for (var i=0; i<n; ++i)
+        bb.readInt32(0);
+    return n;
+};
+
+bench["writeVarint32"] = function(ByteBuffer, n) {
+    n = n || 1000000;
+    var bb = new ByteBuffer(6);
+    for (var i=0; i<n; ++i)
+        bb.writeVarint32(0x7fffffff, 0);
+    return n;
+};
+
+bench["readVarint32"] = function(ByteBuffer, n) {
+    n = n || 1000000;
+    var bb = new ByteBuffer(6).writeInt32(0x7fffffff).flip();
+    for (var i=0; i<n; ++i)
+        bb.readVarint32(0);
+    return n;
+};
+
+bench["writeString"] = function(ByteBuffer, n) {
+    n = n || 100000;
+    var bb = new ByteBuffer(26);
+    for (var i=0; i<n; ++i)
+        bb.writeString("abcdefghijklmnopqrstuvwxyz", 0);
+    return n;
+};
+
+bench["readString"] = function(ByteBuffer, n) {
+    n = n || 100000;
+    var bb = new ByteBuffer(26).writeString("abcdefghijklmnopqrstuvwxyz").flip();
+    for (var i=0; i<n; ++i)
+        bb.readString(26, ByteBuffer.METRICS_BYTES, 0);
+    return n;
+};
+
+bench["calculateString"] = function(ByteBuffer, n) {
+    n = n || 100000;
+    for (var i=0; i<n; ++i)
+        ByteBuffer.calculateString("abcdefghijklmnopqrstuvwxyz");
+    return n;
+};
+
+Object.keys(bench).forEach(function(key) {
+    var func = bench[key];
+    console.log(key);
+    impls.forEach(function(impl) {
+        var n, diff;
+        var start = process.hrtime();
+        n = func(impl.impl);
+        diff = process.hrtime(start);
+        console.log("- "+impl.name+": "+prettyHrTime(diff));
+    });
+    console.log();
+});