LWJGL 0.96 Released!

Started by Matzon, March 30, 2005, 19:59:41

Previous topic - Next topic
Print
Go Down Pages 1 2 3
User actions

princec

#30
April 06, 2005, 07:45:32
Aye sorry I was going to chip in with the solution we talked about on IRC the other night - sleep(1) works perfectly, even with my wierdy hyperthreaded speedstepped laptop! Death to yield(), it no longer cuts the mustard.

And also - we are going to replace sync() and sync2() with a single sync() method based on sync2(), that uses sleep(1). So it should work for everyone, all the time.

Cas :)

tomb

#31
April 06, 2005, 20:54:13
Anyone done any research on how sleep(1) acts on different system, and different loads?

I had some weird fps in my q3 map viewer. With sleep(1) I got roughly 60 fps, without it I got 2-300.

I modified the sphere bouncing test that I posted above so that I could change the number of times the sircly is rendered. Also changed it so I could print out how long the sleep took, mesured with the sun.misc.perf. Here is the result:
less than 10 balls, sleep took 1.7ms
10 to 30 balls, inconsistant results
more than 30 balls, sleep took 14ms

You could clearly see that the movent got choppy. The fps dropped from hundreds to 55, just as in my viewer.

Let me know if anyone wan't to see the code, and I'll post it.

If this is correct then I'm not convinced yet that the new timer can be used for all games.

tomb

#32
April 06, 2005, 23:31:40
Quote from: "oNyx">Sorry if they removed the Perf timer in java 5.

Yep, it's gone. It's System.nanoTime() now.

I know about nanoTime. Just because they added that don't mean they had to remove Perf. nonoTime don't have to use the performance counter, although it probably does.

It's just that I devlop using 1.4 and was not sure if it still worked on 1.5, since it is in the sun package. Anyway, I tested sun.misc.Perf on 1.5.0_02-b09 runtime and it still works.

blue_tomato

#33
April 16, 2005, 07:24:14
Sys.VERSION gives me 0.95 after I installed 0.96. Did I download the wrong files somehow, or got the installation wrong? Or is the string simply not updated?  :?:

blue_tomato

hmm
#34
April 16, 2005, 07:34:33
Never mind, after a reset of the system it reported the right version number. Odd...  :oops:

Nova_C

#35
April 19, 2005, 14:25:15
I get a bug report output every time I try to use anything with devIL in 0.96. It worked in 0.95. I noticed that in 0.96 there is no ilInit() anymore. Am I totally missing something?

Matzon

#36
April 19, 2005, 15:55:09
ilInit was removed coz it's done automatically now - due to some issues with mac.
I am not aware of any issues with devil using 0.96

EgonOlsen

#37
April 19, 2005, 19:22:46
Nice work. I've modified jPCT to use the AWTGLCanvas as an option and it works just fine even when rendering into multiple Canvas'. Well done!

MickeyB

#38
April 28, 2005, 20:35:57
Understanding the timing issues above, performance doenst seem to be the same.  I moved from 092 to 096 and seems very choppy.  I set vsync2(60) and a sleep(1).

also (cross post apologies) the getEventCharacter on Keyboard does not return anything, jsut a space as far as I can tell.  I have tried removing and replacing Keyboard.poll() to no avail.  I understand the keyboard buffer is noe on by default and the method was removed.  thoguhts?
M

elias

#39
April 29, 2005, 09:36:14
Quote from: "MickeyB"
also (cross post apologies) the getEventCharacter on Keyboard does not return anything, jsut a space as far as I can tell.  I have tried removing and replacing Keyboard.poll() to no avail.  I understand the keyboard buffer is noe on by default and the method was removed.  thoguhts?

OS? Does org.lwjgl.test.input.KeyboardTest work for you?

- elias
Print
Go Up Pages 1 2 3
User actions