Author Topic: jPCT benchmark?  (Read 51557 times)

Offline EgonOlsen

  • Administrator
  • quad
  • *****
  • Posts: 12295
    • View Profile
    • http://www.jpct.net
Re: jPCT benchmark?
« Reply #60 on: April 09, 2009, 10:50:31 pm »
An update, triggered by this thread: http://www.javagaming.org/index.php/topic,20248.0.html

There's a switch in JME to lock the meshes, which improves performance by a huge amount and i managed to load it into Java3D.

Taken from the thread on JGO:
Quote
So on the Core2 Quad@3Ghz, ATI Radeon HD 4870, Java6, Vista Ultimate, we now have:

jPCT: 580fps
JME: 520fps
xith3D: 220fps
Java3D: 120fps (with some visual glitches, but that could be a bug in the loader)

Offline paulscode

  • double
  • *****
  • Posts: 863
    • View Profile
    • PaulsCode.Com
Re: jPCT benchmark?
« Reply #61 on: April 09, 2009, 11:21:37 pm »
Hehe, I read those other forums - sounds like there might be some bruised egos.  I love how people are quick to say the test was rigged without actually trying it for themselves..

Offline JavaMan

  • long
  • ***
  • Posts: 231
    • View Profile
Re: jPCT benchmark?
« Reply #62 on: April 09, 2009, 11:28:15 pm »
I was watching the talk on the jME forum earlier today. You've stirred the pot a little Egon. :) Of course, with speed increase like that, how could it not?

I am happy jpct is still faster than jME. With jME1.0 considered outdated, and jME considered a "pre-release" and now some talk of jME3.0. Its a little hard to decide what to use.
« Last Edit: April 10, 2009, 03:49:21 am by JavaMan »

Offline .jayderyu

  • long
  • ***
  • Posts: 116
    • View Profile
Re: jPCT benchmark?
« Reply #63 on: April 09, 2009, 11:39:10 pm »
Well finally. I'm glad that my suggestion that the tests get out their came about. I'm also glad that the various other groups aren't living in an isolated bubble including this one ;).

The new tests performance seem to be much better. Glad to see JME and JPCT running pretty darn well. Though I feel sorry for the guy who is doing DzzD with no benchmark with him.

I agree the other guy seems there are some bruised egos. Who knows maybe this can be an eye opener for various Java 3d engines to grow and be a serious comipitor for PC games :)

In the end seems like my design decisions are about the same. JPCT for my Applets and smaller stuff. Hurray for the upcoming Shader support :)

Offline zammbi

  • float
  • ****
  • Posts: 361
    • View Profile
Re: jPCT benchmark?
« Reply #64 on: April 10, 2009, 06:53:17 am »
Yeah seemed that post did stir a little....
Sorry about that EgonOlsen.
But competition causes improvements  :)
« Last Edit: April 10, 2009, 06:56:04 am by zammbi »

Offline .jayderyu

  • long
  • ***
  • Posts: 116
    • View Profile
Re: jPCT benchmark?
« Reply #65 on: April 10, 2009, 07:27:13 am »
Hey the soup being stirred was awesome. Though I think some of the members could have handled it better. Rather than jumping to accusations of either bias results, code just opening a line of dialog of what are "you" doing... would have been better.

Offline zammbi

  • float
  • ****
  • Posts: 361
    • View Profile
Re: jPCT benchmark?
« Reply #66 on: April 10, 2009, 07:44:49 am »
Yeah it was great. They could of handled it better.
Nice to see Java3D also being compared in the last test, we most likely going see many tests of people trying different ways to increase there fps :)

Offline EgonOlsen

  • Administrator
  • quad
  • *****
  • Posts: 12295
    • View Profile
    • http://www.jpct.net
Re: jPCT benchmark?
« Reply #67 on: April 10, 2009, 03:43:03 pm »
Yep, it's great how i got blamed for being biased and stuff but only one person actually tried it for himself (and came up with the lockMeshes()-call of which i wasn't aware). They all claim to be "best performers", but they never actually try and compare.

Offline fireside

  • double
  • *****
  • Posts: 607
    • View Profile
Re: jPCT benchmark?
« Reply #68 on: April 10, 2009, 04:29:16 pm »
It's hard to do comparisons.  Like with the lock-mesh, you'd have to be a regular Jme user to know about it.  There was that other test a while back that insisted on bsp collision even though it's not being used that much anymore.  Also, the engines fill different niches.  I use Jpct because of it's size.  If I were really intent on features I would think about Jme because it has a lot, but mostly I want an applet to boot up in reasonable amount of time and still provide a game that's entertaining.  Now I'll be able to experiment with a hardware game after my current software game is finished.  Which is the other thing I like about Jpct, it has a meaningful software engine that can keep the game really small. 
click here->Fireside 7 Games<-

Offline zammbi

  • float
  • ****
  • Posts: 361
    • View Profile
Re: jPCT benchmark?
« Reply #69 on: April 10, 2009, 05:26:26 pm »
Really I don't see much point of JME if you already know JPCT. There are many physics libraries which can work just fine for JPCT.
I guess it has a particle engine, which there isn't many libraries for that. Maybe something JPCT could add in the future?
EgonOlsen do you have a plan of what you going to add/update for the future? Or do you just add/update when needed?

Offline JavaMan

  • long
  • ***
  • Posts: 231
    • View Profile
Re: jPCT benchmark?
« Reply #70 on: April 10, 2009, 06:15:02 pm »
I guess it has a particle engine, which there isn't many libraries for that. Maybe something JPCT could add in the future?

There is a demo for a particle system on the demos page. Currently it throws an Exception, but it worked before.

Offline EgonOlsen

  • Administrator
  • quad
  • *****
  • Posts: 12295
    • View Profile
    • http://www.jpct.net
Re: jPCT benchmark?
« Reply #71 on: April 10, 2009, 07:27:34 pm »
There is a demo for a particle system on the demos page. Currently it throws an Exception, but it worked before.
I throws an exception, if your system is too fast...yeah, stupid bug. I would fix it, but i can't find the sources. Maybe i'll write a new one, it's not much work at all.

Offline EgonOlsen

  • Administrator
  • quad
  • *****
  • Posts: 12295
    • View Profile
    • http://www.jpct.net
Re: jPCT benchmark?
« Reply #72 on: April 10, 2009, 08:56:17 pm »
I did another test with the tank model. Results can be found in the JGO thread here: http://www.javagaming.org/index.php/topic,20248.msg163263.html#msg163263

I'm ready to be accused of being biased and testing wrong again, but i did the best i could with all tested engines.

Offline .jayderyu

  • long
  • ***
  • Posts: 116
    • View Profile
Re: jPCT benchmark?
« Reply #73 on: April 11, 2009, 07:47:36 am »
Personally with their hurt feelings they should go about to create their own test that does the same thing. You know your engine. They know there engine. It is only fair that you can setup your engine under the most basic principles. If they feel that their engine is underperforming and your being bias they can create their test benchmark code themselves.

It's not nice accusing others until talking about it first The blatantly ignoring your own efforts to be fair under what you understand of their engine.

oh well getting kicked in the pants does this to people. I'm sure it will get sorted out and the various teams will do what they feel to improve their engine.

I do appreciate 3DzzD though. He's handling the entire thing very well and mature.

Offline paulscode

  • double
  • *****
  • Posts: 863
    • View Profile
    • PaulsCode.Com
Re: jPCT benchmark?
« Reply #74 on: April 11, 2009, 12:55:19 pm »
I'm sure it will get sorted out and the various teams will do what they feel to improve their engine.
...
I do appreciate 3DzzD though. He's handling the entire thing very well and mature.
Good point.  Rather than getting involved in the mud slinging, DzzD took advantage of the oportunity to discover and fix some bugs with his engine (and in the process he jumped up two spots to outperform both JME and xith3D in the tank test ;D).