Java Performance Myths.

[In response to the comment about the original notes being a little too scant, I've decided to edit this entry. Sun will provide audio & slides, which I will link to once available.]

. I heard (or googled) that making fields or methods final (or private) will help performance (or it will allow more inlining).
…Wrong. With or without ‘final’ every inlinable methods is inlined by the runtime compilers.

. I heard (or googled) that try/catch block are free (or very expensive).
…The reality is — it depends, in general try to avoid try/catch blocks in tight loops. Don’t use exceptions to end loops, or for null checking on say list traversal. You’d be defeating JIT optimizations, and duplicating the automatic range check.

. I heard (or googled) that using RTTI is better than instance_of (and/or better than v-call (virtual call))
…RTTI (Runtime Type Information –google for samples) is an ugly hack from c++, don’t do it unless you need to squeeze the last 10% of perf. improvement. RTTI wins but it’s too ugly (in the OO sense). Use v-call if you can.
…v-call is more expensive in Hotspot than other VMs, Hotspot implements better subtype checking (efficent switch).
…The bottom line is runtime compilers are optimized for the common patterns of coding out there, so stick to clean design/coding, use OO principles.

v-call : v_call(); // dynamic dispatch here
instance_of:
if( this instanceof Child1 )
((Child1)this).non_v_call();
RTTI: switch( _rtti ) { case 1: // hand-inline Child1 specific ...


. Should I avoid synchronization at all cost?
…The average system today spends 55-110 ns doing uncontented lock/unlock operations, so it’s not free but not terribly expensive either.
…Hotspot’s synchronization operations on Xeons are apparently slow (~275 cycles), IBM/BEA/Azul are much better at it.
…For light contention situations, BEA outperforms the other VMs, Sun performs poorly.
…In general, synchronization is better than bugs! (especially now that’s we’re in the multi-core era). So beware of the costs (and profile if you can) and try to use the new concurrency API in 1.5 but don’t avoid them as threading bugs like race conditions are notoriously hard to fix. Think more about your algorithm.

. I heard (or googled) that I should use object pools say to help the garbage collector, or should I reuse objects or create new ones and assume that the GC is smart enough to efficently take care of the cleanup?
…It depends on cost of initialization of the object and the turnover rates. Don’t do it for small objects, but it may be a possible win for large ones or those with heavy initalization cost (JPanel?). As always, profile — use JConsole or VisualGC.
…Don’t pool objects like Hashtables.

. How much of a performance impact the 5.0 features have on Java code?
…The foreach construct and autoboxing are FREE! (no additional cost incurred). They’re syntactic sugar.
…Note that enums on Xeon with Sun’s Hotspot has issues (more on this in a later post, but it’s when iterating over enums).

.Other predictions/advice
…Pause times are going down, GCs are getting more efficient. Concurrent GC will be the default in most VMs in the coming years, so if possible use them now!
…Escape analysis (optimization that can be performed to improve storage allocation and reclamation of objects) is being integrated into VMs, one less reason to pool objects.
…Locking is getting cheaper but multi-cores will make it more expensive, a CAS instruction on x86 takes ~200+ cycles.

Technorati Tags: JavaOne | CodePerformance

Interesting talk by Hans Boehm on Finalization and the JMM. It was mostly a repeat of his presentation at POPL03, that you can find here.

Cramer mentioned he was working with Kaffe folks in Brazil for them to be able to deploy apps from within Netbeans. Coyote was mentioned also as another example of how not just Netbeans but also SE is opening up to other non-Java languages. Hamilton said they’ll be enhancing the Java bytecode in Dolphin (aka Java 7) to ease support for scripting languages.

Sun seems impressed by the turnout, rooms were pretty crowded…

CodePerformance | JavaOne

BEA — 26 talks: 13 BOFs, 13 Sessions
Google — 2 talks: 1 BOF, 1 Session
IBM — No talks???
Sun — 144 Talks: 71 BOFs, 73 Sessions
Oracle — 16 Talks: 5 BOFs, 11 Sessions
Nokia — 11 Talks: 7 BOFs, 4 Sessions
HP — 2 talks: 1 BOF, 2 Sessions
SAP — 2 talks: 1 BOF, 1 Session
Motorola — 5 talks: 2 BOFs, 3 Sessions
JBoss — 6 Talks: 1 BOF, 5 Sessions
Interface21 — 2 talks: 2 Sessions

15 Talks about or related to NetBeans: 10 Sessions, 5 BOFs
6 Talks about or related to Eclipse: 2 BOFs, 4 Sessions

Other interesting tidbits:

Talks–
. Hans Bohem: Finalization, Threads, and the Javaâ„¢ Technology Memory Model
. Arthur Van Hoff: Tahiti: Designing for the TiVo Platform, Interactive Media Applications Around the Home
. Azul Systems will be giving 3 Sessions.
. Performance Myths Exposed.
. “…We compare tweaked and untweaked code on a wide range of modern Java Virtual Machines (JVM machines) to see what makes Java technology run faster…. nearly all performance tweaks from five years ago are now strictly counterproductive. “
. Speculative Locking: Breaking the Scale Barrier
. “… Speculative Locking allows many threads to speculatively execute at the same time into a locked region. Only threads that actually hit data contention will need to roll back and retry.”
. Network Attached Processing: Tapping 384-Way SMP, 256GB Javaâ„¢ Technology Compute Bricks
. “… network attached processing, a fundamentally new model for delivering massive amounts of compute power to Javaâ„¢ technology-based applications running on existing server hardware, operating systems, and middleware platforms”

Technorati Tags: JavaOne | CodePerformance

2005 will mark the return of Microsoft to JavaOne though they did send a few employees at attend last year, this year as Sun CEO Scott McNealy announced on Friday (13th of May) not only are they a “major” sponsor but they’re also presenting sessions and BOFs.

Their talks:

BOF-9095 On the Couch With Sun and Microsoft
Ray Lai, Sun Microsystems; Simon Guest, Microsoft Corporation; Marina Fisher, Sun Microsystems; Laurence Moroney, Philotic Software Solutions; Dino Chiesa, Microsoft Corporation; Doug Purdy, Microsoft Corporation

BOF-9854 Systems Management Using Web Services and WS-Management
Akhil Arora, Sun Microsystems; Josh Cohen, Microsoft Corporation

TS-3367 Interoperability Between Javaâ„¢ Technology and .NET: More Than Just Web Services
Kevin Wittkopf, Microsoft; Wayne Citrin, JNBridge, LLC

TS-3552 Javaâ„¢ Technology and .NET Interoperability Using WS-* Web Services Architecture
Robert Brewin, Sun Microsystems; Mark Hapner, Sun Microsystems, Inc.; Andrew Layman, Microsoft Corporation

TS-3553 Javaâ„¢ and .NET: Interoperability Challenges and Rewards
Mark Nottingham, BEA Systems, Inc.; Eduardo Pelegri-Llopart, Sun Microsystems, Inc; Dino Chiesa, Microsoft Corporation; Anne Thomas Manes, Burton Group; Simon Fell, Salesforce.com; William Henry, IONA Technologies

TS-9866 Advanced Web Services Interoperability
Simon Guest, Microsoft Corporation; Raghavan Srinivas, Sun Microsystems

Technorati Tags: JavaOne | CodePerformance