Jump to content

Talk:HotSpot (virtual machine)

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

License

[edit]

I have tried to figure out how the license terms for Hotspot relates to the license terms of the binary distribution of Java SE from Oracle. If I read correctly (but I'm genuinely uncertain!) the source is GPL and can be downloaded and built and used as you please (under FOSS constraints) while the binary distribution of the same thing is licensed under terms that exclude embedded applications. This is a trick that if it is really the case ought to be explained in the License section. Or?

See http://www.oracle.com/technetwork/java/javase/terms/license/index.html — Preceding unsigned comment added by Daniel.wengelin (talkcontribs) 09:55, 27 September 2011 (UTC)[reply]

Performance

[edit]

Does anyone have a reference for the statement, "In theory, though rarely in practice, it is possible for adaptive optimization of a JVM to exceed hand coded C++ or assembly language?" I don't doubt this might be true, but I'm intrigued and would like to read some of the research on the topic.

I added a link to an article about this. Hervegirod 10:14, 25 November 2006 (UTC)[reply]
The link doesn't say that; it says "is comparable to that of C++, " I think a better wording would be "In theory, though rarely in practice, adaptive optimization of the JVM can be comparable in speed to fully optimized coded C++ or assembly language. As always, benchmarks can vary" Jgwinner (talk) 19:35, 27 October 2010 (UTC)[reply]
The article also says : Five composite benchmarks listed below show that modern Java has acceptable performance, being nearly equal to (and in many cases faster than) C/C++ across a number of benchmarks. So I changed the existing sentence by: In some cases, it is possible for adaptive optimization of a JVM to exceed the performance of hand-coded C or C++ Hervegirod (talk) 22:16, 28 October 2010 (UTC)[reply]
The reference you have provided does not actually say that the difference in their benchmarks was due to adaptive optimization. The benchmarks do not seem reliable - one numerical computation result ("LUP matrix inversion") claims that Java was 3.9 times faster than C! Their tests were on a single core Pentium, but the Computer Language Benchmarks Game shows that no Java program beats C on that platform (link) - at best they are equal, at worst Java is 4 times slower. 77.97.210.38 (talk) 21:33, 7 December 2014 (UTC)[reply]
This statement really needs to be clarified with some analysis of proper benchmarks, and some proof that adaptive optimisation was the cause of the performance increase. Simply pointing to one or two badly written C benchmarks and then claiming that "the JVM can exceed the performance of hand-coded C or C++" is not valid science - and what does "hand coded" mean anyway? The text seems to be trying to imply that naive Java code can beat manually optimised C, which is not true for the vast majority of benchmarks. 77.97.210.38 (talk) 21:41, 7 December 2014 (UTC)[reply]

-client and -server being phased-out?

[edit]

Judging by http://developers.sun.com/learning/javaoneonline/j1sessn.jsp?sessn=TS-3412&yr=2006&track=coreplatform page 52, Sun are phasing out the different VMs and in favour of a single, combined VM. Is this true? Would be worth a mention. Wootery (talk) 15:02, 2 August 2008 (UTC)[reply]

Version 17 is out.

[edit]

JDK 1.6 update 21 release notes mention Hotspot VM17.0. --Xerces8 (talk) 16:08, 5 August 2010 (UTC)[reply]

[[|default|default]] — Preceding unsigned comment added by 176.58.69.184 (talk) 10:18, 19 September 2012 (UTC)[reply]

Click on Click

[edit]

The statement that singles out the hiring of Cliff Click relies on a single interview that Click did with Werner Schuster about his work at another company entitled "Cliff Click on Azul's Pauseless GC, Zing, JVM Languages". That's hardly a reliable source on his activity and specifically, it fails WP:SELFSOURCE specifically, #1 and 4. The material is self-serving and an exceptional claim.

To single him out based on his own clams without any WP:SECONDARY sources that support it is inappropriate and fails the principle of verifiablility. I had removed the specific reference to him and left the other claims from that interview in the article, but an IP (likely Tarcieri) has reverted it and re-inserted what looks like promotion of Click. Tarcieri (talk · contribs) left me a note on my talk page indignant that I had removed Click's name and implying that I had to have personal knowledge of the history of HotSpot to edit the page.

This smells suspiciously like someone trying to promote Click's involvement. Unless, reliable, secondary sources that don't use Click as the source for this info and the context are provided, this material should be removed. Toddst1 (talk) 15:11, 14 December 2012 (UTC)[reply]

Thanks. Toddst1 (talk) 04:28, 15 December 2012 (UTC)[reply]
"an IP (likely Tarcieri)" umm no, that wasn't me, but thank you for assuming the worst in people. I posted the revision where the references to Cliff I added were deleted to Twitter, and can only assume that one of my followers made the edit. From your comments, apparently I'm part of some sort of conspiracy that seeks to promote Cliff Click! Or I'm simply someone who recognizes his obvious involvement in the early days of the JVM. It seems your problem is I'm citing Cliff on Cliff. Okay, that's a reasonable complaint, I suppose, although a complaint anyone reasonably versed in the history of the JVM probably wouldn't make. Nevertheless I will prepare ample sources before I add Cliff back to the article. Once again I must voice my frustration with the way Wikipedia operates and once again ask you to consider why participation in Wikipedia is declining. But contrary to what you would immediately assume about me giving some random IP made an edit, I am no sock puppet, and before I touch the article again I will try to (painstakingly) conform to Wikipedia's policies, once again in protest of how onerous and discouraging of participation as they may be. Tarcieri (talk) 05:42, 20 December 2012 (UTC)[reply]
Fair enough. I retract my assumption. Toddst1 (talk) 21:06, 28 December 2012 (UTC)[reply]
Hi after a long time. I'm an anonymous contributor. I've added back info on Cliff Click's involvement in development of highly optimizing JIT compiler in HotSpot. This time it's backed by highly cited paper https://dl.acm.org/doi/10.5555/1267847.1267848 - it's highly cited also by other HotSpot developers. The relevant change is: https://wiki.riteme.site/w/index.php?title=HotSpot_(virtual_machine)&diff=prev&oldid=1218437115 193.108.73.47 (talk) 18:20, 11 April 2024 (UTC)[reply]