Java performance tuning

Table of Contents; Preface; Contents of This Book; New in the Second Edition; Virtual Machine (VM) Versions; Conventions Used in This Book; Comments and Questions; Acknowledgments; Chapter 1. Introduction; Why Is It Slow?; The Tuning Game; System Limitations and What to Tune; A Tuning Strategy; Perceived Performance; Threading to Appear Quicker; Streaming to Appear Quicker; Caching to Appear Quicker; Starting to Tune; User Agreements; Setting Benchmarks; The Benchmark Harness; Taking Measurements; What to Measure; Don't Tune What You Don't Need to Tune; Performance Checklist

Chapter 2. Profiling ToolsMeasurements and Timings; Garbage Collection; Using -verbosegc to Analyze Garbage Collection; Tools for Analyzing -verbosegc Output; Method Calls; Profiling Methodology; Java 2 cpu=samples Profile Output; HotSpot and 1.3 -Xprof Profile Output; JDK 1.1.x -prof and Java 2 cpu=old ProfileOutput; Java 2 -Xhprof Option; Java 2 -Xaprof option; Object-Creation Profiling; Monitoring Gross Memory Usage; Client/Server Communications; Replacing Sockets; Performance Checklist; Chapter 3. Underlying JDK Improvements; Garbage Collection; Tuning the Heap; Gross Tuning

Problems with a Larger HeapStarting Versus Maximum Heap Size; Benchmarking Considerations; Fine-Tuning the Heap; Expanding the Heap; Minimizing Pauses; Incremental or "train" GC; Concurrent GC; Enlarge the "new" space; Disabling System.gc() Calls; Tuning RMI Garbage Collection; Extreme Heap and Intimate Shared Memory; Loading a Huge Number of Classes; Per-Thread Stack Size; Eliminate Finalizers; Sharing Memory; Replacing JDK Classes; Faster VMs; VM Speed Variations; VMs with JIT Compilers; VM Startup Time; Other VM Optimizations; Better Optimizing Compilers

What Optimizing Compilers Cannot DoWhat Optimizing Compilers Can Do; Remove unused methods and classes; Increase statically bound calls; Cut dead code and unnecessary instructions, including checks for null; Use computationally cheaper alternatives (strength reduction); Replace runtime computations with compiled results; Remove unused fields; Remove unnecessary parts of compiled files; Reduce necessary parts of compiled files; Alter access control to speed up invocations; Inline calls; Remove dynamic type checks; Unroll loops; Code motion; Eliminate common subexpressions

Eliminate unnecessary assignmentsRename classes, fields, and methods; Reorder or change bytecodes; Generate information to help a VM; Managing Compilers; Sun's Compiler and Runtime Optimizations; Optimizations You Get for Free; Literal constants are folded; String concatenation is sometimes folded; Constant fields are inlined; Dead code branches are eliminated; Optimizations Performed When Using the -O Option; Performance Effects From Runtime Options; Compile to Native Machine Code; Native Method Calls; Uncompressed ZIP/JAR Files; Performance Checklist; Chapter 4. Object Creation

Why Is It Slow? --

2 --

Tuning Game --

3 --

System Limitations and What to Tune --

3 --

A Tuning Strategy --

5 --

Perceived Performance --

6 --

Starting to Tune --

10 --

What to Measure --

15 --

Don't Tune What You Don't Need to Tune --

17 --

2. Profiling Tools --

20 --

Measurements and Timings --

21 --

Garbage Collection --

23 --

Method Calls --

26 --

Object-Creation Profiling --

43 --

Monitoring Gross Memory Usage --

50 --

Client/Server Communications --

56 --

3. Underlying JDK Improvements --

64 --

Garbage Collection --

64 --

Replacing JDK Classes --

66 --

Faster VMs --

68 --

Better Optimizing Compilers --

73 --

Sun's Compiler and Runtime Optimizations --

82 --

Compile to Native Machine Code --

89 --

Native Method Calls --

90 --

Uncompressed ZIP/JAR Files --

91 --

4. Object Creation --

94 --

Object-Creation Statistics --

96 --

Object Reuse --

97 --

Avoiding Garbage Collection --

112 --

Initialization --

115 --

Early and Late Initialization

117 --

5. Strings --

122 --

Performance Effects of Strings --

122 --

Compile-Time Versus Runtime Resolution of Strings --

125 --

Conversions to Strings --

126 --

Strings Versus char Arrays --

142 --

String Comparisons and Searches --

150 --

Sorting Internationalized Strings --

153 --

6. Exceptions, Casts, and Variables --

161 --

Exceptions --

161 --

Casts --

167 --

Variables --

169 --

Method Parameters --

172 --

7. Loops and Switches --

174 --

Java.io. Reader Converter --

178 --

Exception-Terminated Loops --

184 --

Switches --

189 --

Recursion --

195 --

Recursion and Stacks --

200 --

8. I/O, Logging, and Console Output --

204 --

Replacing System.out --

206 --

Logging --

208 --

From Raw I/O to Smokin' I/O --

209 --

Serialization --

217 --

Clustering Objects and Counting I/O Operations --

230 --

Compression --

232 --

9. Sorting --

237 --

Avoiding Unnecessary Sorting Overhead --

237 --

An Efficient Sorting Framework --

241

Better Than O(nlogn) Sorting --

249 --

Performance Checklist --

255 --

10. Threading --

256 --

User-Interface Thread and Other Threads --

258 --

Race Conditions --

259 --

Deadlocks --

261 --

Synchronization Overheads --

265 --

Timing Multithreaded Tests --

275 --

Atomic Access and Assignment --

276 --

Thread Pools --

278 --

Load Balancing --

279 --

Threaded Problem-Solving Strategies --

291 --

11. Appropriate Data Structures and Algorithms --

293 --

Collections --

294 --

Java 2 Collections --

296 --

Hashtables and HashMaps --

298 --

Cached Access --

302 --

Caching Example I --

303 --

Caching Example II --

306 --

Finding the Index for Partially Matched Strings --

310 --

Search Trees --

314 --

12. Distributed Computing --

335 --

Tools --

337 --

Message Reduction --

339 --

Comparing Communication Layers --

342 --

Caching --

344 --

Batching I --

346 --

Application Partitioning --

347 --

Batching II --

348 --

Low-Level Communication Optimizations

349 --

Distributed Garbage Collection --

354 --

Databases --

355 --

13. When to Optimize --

357 --

When Not to Optimize --

358 --

Tuning Class Libraries and Beans --

359 --

Analysis --

362 --

Design and Architecture --

366 --

Tuning After Deployment --

382 --

More Factors That Affect Performance --

383 --

14. Underlying Operating System and Network Improvements --

389 --

Hard Disks --

390 --

CPU --

395 --

RAM --

397 --

Network I/O --

398.