From 479abae9cdf76f6676fbfd8695b0de9de41f81f8 Mon Sep 17 00:00:00 2001
From: Chris Vest <christianvest_hansen@apple.com>
Date: Wed, 29 Jul 2020 10:34:21 +0200
Subject: [PATCH] Access benchmark for BBuf

Motivation:
We want to be able to compare the performance of the existing ByteBuf implementation, and the new MemorySegment based BBuf.

Modification:
Copy the existing access benchmark for ByteBuf, and modify the copy to use the new BBuf instead.

Result:
We are able to get our first benchmark runs with BBuf.
The cost of accessing memory in BBuf is roughly twice that of the comparable ByteBuf implementations.
I think we'll need to improve this if we want BBuf to be a viable path forward.
---
 .../io/netty/buffer/BBufAccessBenchmark.java  | 120 ++++++++++++++++++
 1 file changed, 120 insertions(+)
 create mode 100644 microbench/src/main/java/io/netty/buffer/BBufAccessBenchmark.java

diff --git a/microbench/src/main/java/io/netty/buffer/BBufAccessBenchmark.java b/microbench/src/main/java/io/netty/buffer/BBufAccessBenchmark.java
new file mode 100644
index 0000000..47da07f
--- /dev/null
+++ b/microbench/src/main/java/io/netty/buffer/BBufAccessBenchmark.java
@@ -0,0 +1,120 @@
+/*
+* Copyright 2019 The Netty Project
+*
+* The Netty Project licenses this file to you 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.
+*/
+package io.netty.buffer;
+
+import io.netty.buffer.b2.Allocator;
+import io.netty.buffer.b2.BBuf;
+import io.netty.microbench.util.AbstractMicrobenchmark;
+import org.openjdk.jmh.annotations.Benchmark;
+import org.openjdk.jmh.annotations.BenchmarkMode;
+import org.openjdk.jmh.annotations.Fork;
+import org.openjdk.jmh.annotations.Measurement;
+import org.openjdk.jmh.annotations.Mode;
+import org.openjdk.jmh.annotations.OutputTimeUnit;
+import org.openjdk.jmh.annotations.Param;
+import org.openjdk.jmh.annotations.Setup;
+import org.openjdk.jmh.annotations.TearDown;
+import org.openjdk.jmh.annotations.Warmup;
+
+import java.util.concurrent.TimeUnit;
+
+@Warmup(iterations = 5, time = 1500, timeUnit = TimeUnit.MILLISECONDS)
+@Measurement(iterations = 10, time = 500, timeUnit = TimeUnit.MILLISECONDS)
+@Fork(1)
+@BenchmarkMode(Mode.AverageTime)
+@OutputTimeUnit(TimeUnit.NANOSECONDS)
+public class BBufAccessBenchmark extends AbstractMicrobenchmark {
+    public enum BBufType {
+        UNSAFE {
+            @Override
+            BBuf newBuffer() {
+                return Allocator.direct().allocate(64);
+            }
+        },
+//        UNSAFE_SLICE {
+//            @Override
+//            BBuf newBuffer() {
+//                return UNSAFE.newBuffer().slice(16, 48);
+//            }
+//        },
+        HEAP {
+            @Override
+            BBuf newBuffer() {
+                return Allocator.heap().allocate(64);
+            }
+        },
+//        COMPOSITE {
+//            @Override
+//            BBuf newBuffer() {
+//                return Unpooled.wrappedBuffer(UNSAFE.newBuffer(), HEAP.newBuffer());
+//            }
+//        },
+//        NIO {
+//            @Override
+//            BBuf newBuffer() {
+//                return new NioFacade(BBuffer.allocateDirect(64));
+//            }
+//        }
+        ;
+        abstract BBuf newBuffer();
+    }
+
+    @Param
+    public BBufType bufferType;
+
+    @Param({ "8" })
+    public int batchSize; // applies only to readBatch benchmark
+
+    @Setup
+    public void setup() {
+        buffer = bufferType.newBuffer();
+    }
+
+    private BBuf buffer;
+
+    @TearDown
+    public void tearDown() {
+        buffer.close();
+    }
+
+    @Benchmark
+    public long setGetLong() {
+        return buffer.setLong(0, 1).getLong(0);
+    }
+
+    @Benchmark
+    public BBuf setLong() {
+        return buffer.setLong(0, 1);
+    }
+
+    @Benchmark
+    public int readBatch() {
+        buffer.readerIndex(0).touch();
+        int result = 0;
+        // WARNING!
+        // Please do not replace this sum loop with a BlackHole::consume loop:
+        // BlackHole::consume could prevent the JVM to perform certain optimizations
+        // forcing ByteBuf::readByte to be executed in order.
+        // The purpose of the benchmark is to mimic accesses on ByteBuf
+        // as in a real (single-threaded) case ie without (compiler) memory barriers that would
+        // disable certain optimizations or would make bounds checks (if enabled)
+        // to happen on each access.
+        for (int i = 0, size = batchSize; i < size; i++) {
+            result += buffer.readByte();
+        }
+        return result;
+    }
+}