Java-Gaming.org Java4K winners: [ by our judges | by the community ]         
Featured games (67)
games approved by the League of Dukes 		Games in Showcase (∞)
games submitted by our members



News: Read the Java Gaming Resources, or peek at the official Java tutorials
 
    Home     Help   Search   Login   Register   
Pages: [1] 2
  Print  
  HDR Fireworks + Particle Engine benchmark  (Read 2875 times)
0 Members and 3 Guests are viewing this topic.
Online theagentd

JGO Wizard
****

Posts: 1392
Medals: 88
« on: 2011-11-06 06:40:57 »
Hey, guys.
I just put together a small executable version of my little fireworks test.
MediaFire link >_>


Features:

HDR:
 - HDR rendering to a 16-bit floating point render target.
 - Bloom effect to make it look really awesome.
 - Advanced luminance based tone mapping.

Particle engine:
 - Just basic smoothed points.
 - Batched together to a single draw call.
 - A special data buffer that keeps the particles compacted.
 - Uses MappedObject for better performance.
 - Performance is excellent: 575 000 particles at 63 FPS.


Requires:
 - A OpenGL 3.0 compatible video card and OS.
 - Java 1.6.


Any feedback or questions?
There is no god.
Offline pitbuller

Sr. Member
**

Posts: 340
Medals: 9
« Reply #1 on: 2011-11-06 07:40:59 »
About 250k with 30fps.
Effect is really cool.
https://market.android.com/details?id=boxtrix.android
http://gamejolt.com/online/games/arcade/kers/4944/
http://gamejolt.com/online/games/arcade/shapetronic/541
Online theagentd

JGO Wizard
****

Posts: 1392
Medals: 88
« Reply #2 on: 2011-11-06 08:58:48 »
Quote from: pitbuller on 2011-11-06 07:40:59
About 250k with 30fps.
Effect is really cool.
Hehe, isn't it? Thanks for testing it!
There is no god.
Games published by our own members! Go get 'em!
Offline ReBirth

JGO Wizard
****

Posts: 1272
Medals: 19
« Reply #3 on: 2011-11-06 09:24:07 »
"Can't find server JVM" on mine. Win 7 Java 7.
Twitter @drabiter
Offline CommanderKeith

JGO Wizard
****

Posts: 1455
Medals: 9
« Reply #4 on: 2011-11-06 10:13:30 »
Hi, looked cool at the start, then the VM died after 2s of fireworks. Looks like my computer exploded after all lol.

Btw try setting default values in the menus when they pop up.

Looked cool, good job

1  
2  
3  
4  
5  
6  
7  
8  
9  
10  
11  
12  
13  
14  
15  
16  
17  
18  
19  
20  
21  
22  
23  
24  
25  
26  
27  
28  
29  
30  
31  
32  
33  
34  
35  
36  
37  
38  
39  
40  
41  
42  
43  
44  
45  
46  
47  
48  
49  
50  
#
# A fatal error has been detected by the Java Runtime Environment:
#
#  EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x000000006da33a4b, pid=3356, tid=756
#
# JRE version: 6.0_26-b03
# Java VM: Java HotSpot(TM) 64-Bit Server VM (20.1-b02 mixed mode windows-amd64 compressed oops)
# Problematic frame:
# V  [jvm.dll+0x243a4b]
#
# If you would like to submit a bug report, please visit:
#   http://java.sun.com/webapps/bugreport/crash.jsp
#

---------------  T H R E A D  ---------------

Current thread (0x000000000023b000):  JavaThread "main" [_thread_in_vm, id=756, stack(0x0000000002470000,0x0000000002570000)]

siginfo: ExceptionCode=0xc0000005, writing address 0x00000000177f4000

Registers:
RAX=0x3f8000003c23d70a, RBX=0x000000000023b000, RCX=0x0000000002aa1000, RDX=0x0000000016110000
RSP=0x000000000256ee70, RBP=0x000000000256efa0, RSI=0x0000000018bb1000, RDI=0x0000000000000000
R8 =0x0000000000002aab, R9 =0x00000000177f4008, R10=0x0000000002581359, R11=0x000000006d9f2630
R12=0x0000000000000000, R13=0x000000077bec9f00, R14=0x000000000256f000, R15=0x000000000023b000
RIP=0x000000006da33a4b, EFLAGS=0x0000000000010202

Top of Stack: (sp=0x000000000256ee70)
0x000000000256ee70:   000000000000000e 0000000000000000
0x000000000256ee80:   000000000023b000 0000000000000000
0x000000000256ee90:   000000000023b000 000000006d9bdf57


etc etc




---------------  S Y S T E M  ---------------

OS: Windows 7 Build 7601 Service Pack 1

CPU:total 8 (4 cores per cpu, 2 threads per core) family 6 model 30 stepping 5, cmov, cx8, fxsr, mmx, sse, sse2, sse3, ssse3, sse4.1, sse4.2, popcnt, ht

Memory: 4k page, physical 8320880k(5640404k free), swap 16639912k(13493832k free)

vm_info: Java HotSpot(TM) 64-Bit Server VM (20.1-b02) for windows-amd64 JRE (1.6.0_26-b03), built on May  4 2011 07:15:24 by "java_re" with MS VC++ 8.0 (VS2005)

time: Mon Nov 07 02:11:31 2011
elapsed time: 34 seconds
StraightEdge 2D polygon project

SydneyEngine multiplayer top-down view shooter
Online theagentd

JGO Wizard
****

Posts: 1392
Medals: 88
« Reply #5 on: 2011-11-06 11:10:09 »
Quote from: ReBirth on 2011-11-06 09:24:07
"Can't find server JVM" on mine. Win 7 Java 7.
The bat file starts it with the -server VM command. No idea why it doesn't work. Try to just double click the jar.

Quote from: CommanderKeith on 2011-11-06 10:13:30
Hi, looked cool at the start, then the VM died after 2s of fireworks. Looks like my computer exploded after all lol.

Btw try setting default values in the menus when they pop up.

Looked cool, good job

1  
snip...


Because you just had to create more than one million particles. Sigh. I'm currently just allocating data for 1 000 000 particles on the GPU, so if you go over that it'll write outside the buffer bounds and explode. The problem is that 1 million particles is already 24MBs, so it's pretty much starting to get out of hand. BTW, do you have any performance numbers from your shiny quad core? =D
There is no god.
Offline EgonOlsen

JGO Strike Force
***

Posts: 821
Medals: 6
« Reply #6 on: 2011-11-06 11:54:07 »
Runs fine @77fps with ~680,000 particles (Core i7 2600K @ 4Ghz, Radeon HD 5870, Cat 11.9). However, clicking the mouse to "turn off the epic bloom filter" results in a black and white rendering of the fireworks. Turning it on again results in a white screen.
jPCT, a 3D engine for Java -
jPCT-AE, a 3D engine for Android -
Robombs - a multiplayer game inspired by Bomberman
Offline Cero

JGO Neuromancer
****

Posts: 1050
Medals: 18
« Reply #7 on: 2011-11-06 14:14:41 »
Quote from: theagentd on 2011-11-06 11:10:09
Quote from: ReBirth on 2011-11-06 09:24:07
"Can't find server JVM" on mine. Win 7 Java 7.
The bat file starts it with the -server VM command. No idea why it doesn't work. Try to just double click the jar.

server jvm isnt included in the default jre. only in the jdk one
so usually what people do is copy the server folder from the jdk jre to the normal jre
Alchemic-Tempest.com  Google+  Twitter  Facebook
Offline Rejechted

Full Member
**

Posts: 119
Medals: 1


Just a guy making some indie games :D
« Reply #8 on: 2011-11-06 15:21:20 »
I5 2500k @ 4.3ghz, radeon 6950.

Roughly 575,000 particles, 103 fps.
Blog for our project (Codenamed Lead Crystal): http://silvergoblet.tumblr.com
Offline ra4king

JGO Kernel
*****

Posts: 3155
Medals: 196


I'm the King!
« Reply #9 on: 2011-11-06 15:34:42 »
1  
2  
3  
4  
5  
6  
7  
8  
9  
10  
11  
12  
13  
14  
15  
16  
17  
18  
19  
20  
21  
#
# A fatal error has been detected by the Java Runtime Environment:
#
#  EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x000000006bf1e5bb, pid=4704, t
id=9728
#
# JRE version: 7.0_01-b08
# Java VM: Java HotSpot(TM) 64-Bit Server VM (21.1-b02 mixed mode windows-amd64
compressed oops)
# Problematic frame:
# V  [jvm.dll+0x1ee5bb]
#
# Failed to write core dump. Minidumps are not enabled by default on client vers
ions of Windows
#
# An error report file with more information is saved as:
# E:\Roi Atalla\Downloads\fireworks\hs_err_pid4704.log
#
# If you would like to submit a bug report, please visit:
#   http://bugreport.sun.com/bugreport/crash.jsp
#

i7 2600k, GTX 570
I used 1000 fireworks with a brightness of 300

EDIT:
Lowered it to 25 fireworks with a brightness of 15:
about 850K at ~70FPS
-Roi
Doodle Jump online!
Follow me on Twitter!
Games published by our own members! Go get 'em!
Offline Nate

JGO Neuromancer
****

Posts: 1062
Medals: 30


mooooo
« Reply #10 on: 2011-11-06 17:54:31 »
Quote
Exception in thread "main" java.lang.IllegalStateException: Function is not supported
        at org.lwjgl.BufferChecks.checkFunctionAddress(BufferChecks.java:58)
        at org.lwjgl.opengl.GL33.glGenSamplers(GL33.java:141)
        at fireworks.gfx.obj.Sampler.<init>(Sampler.java:11)
Next time I reboot I'll switch to the ATI cards instead of the crap Intel and try again.
projects
Online theagentd

JGO Wizard
****

Posts: 1392
Medals: 88
« Reply #11 on: 2011-11-06 19:54:43 »
Quote from: EgonOlsen on 2011-11-06 11:54:07
Runs fine @77fps with ~680,000 particles (Core i7 2600K @ 4Ghz, Radeon HD 5870, Cat 11.9). However, clicking the mouse to "turn off the epic bloom filter" results in a black and white rendering of the fireworks. Turning it on again results in a white screen.
It's black and white because all color channels are over 1, being capped to white. That's the whole point of the bloom filter. However, the brightness of 680,000 particles is on par with a nuclear bomb, so yeah, no wonder that it's white. Reduce the color strength of each star a little and/or scroll the mouse wheel up (a lot) to change reduce exposure of the tone mapper and make everything darker.

Quote from: Rejechted on 2011-11-06 15:21:20
I5 2500k @ 4.3ghz, radeon 6950.

Roughly 575,000 particles, 103 fps.
Really nice FPS for both you and Cero! Now think about what I could do with multithreading! =D

Quote from: ra4king on 2011-11-06 15:34:42
1  
snip...

i7 2600k, GTX 570
I used 1000 fireworks with a brightness of 300

EDIT:
Lowered it to 25 fireworks with a brightness of 15:
about 850K at ~70FPS
1000 fireworks per update? As in 60 000 fireworks per second, each spewing out a fire trail of 4 particles per update, and then finally exploding into 50-300 stars? The current cap is 1 million particles, which translates to about 30 fireworks per update and it crashes if you go over that (Buffer overflow on the mapped buffer xD). See the bottom of this post

Quote from: Nate on 2011-11-06 17:54:31
Quote
Exception in thread "main" java.lang.IllegalStateException: Function is not supported
        at org.lwjgl.BufferChecks.checkFunctionAddress(BufferChecks.java:58)
        at org.lwjgl.opengl.GL33.glGenSamplers(GL33.java:141)
        at fireworks.gfx.obj.Sampler.<init>(Sampler.java:11)
Next time I reboot I'll switch to the ATI cards instead of the crap Intel and try again.
Only the newest Intel cards support OpenGL 3 if I recall correctly, but I doubt that they support OpenGL 3.3 anyway.



Okay, new version: http://www.mediafire.com/?c1iszts14vackoe
Changes:
 - Dynamically allocates more GPU buffer memory for particles.
 - Soft capped at 10 000 000 particles to avoid filling up RAM and freezing.
 - Default values (the nice looking settings).
 - Writes out the number of fireworks to the console too.
There is no god.
Offline ra4king

JGO Kernel
*****

Posts: 3155
Medals: 196


I'm the King!
« Reply #12 on: 2011-11-06 20:04:45 »
Ok with fireworks2 and 30 fireworks at brightness of 15:
~1.02million particles
~65FPS

I <3 my computer Cheesy
-Roi
Doodle Jump online!
Follow me on Twitter!
Offline EgonOlsen

JGO Strike Force
***

Posts: 821
Medals: 6
« Reply #13 on: 2011-11-07 00:57:52 »
Quote from: theagentd on 2011-11-06 19:54:43
It's black and white because all color channels are over 1, being capped to white. That's the whole point of the bloom filter. However, the brightness of 680,000 particles is on par with a nuclear bomb, so yeah, no wonder that it's white. Reduce the color strength of each star a little and/or scroll the mouse wheel up (a lot) to change reduce exposure of the tone mapper and make everything darker.
No, i think you misunderstood...it's fine after startup (but pretty bright as you've said) without pressing the mouse button (i.e. with bloom(?)). Clicking the mouse turns it into a black/white-picture, i.e. everything is either deep black or plain white. That includes rockets and such. Clicking it again turns the whole screen white, i.e. there's no other color visible than 255,255,255.
jPCT, a 3D engine for Java -
jPCT-AE, a 3D engine for Android -
Robombs - a multiplayer game inspired by Bomberman
Online theagentd

JGO Wizard
****

Posts: 1392
Medals: 88
« Reply #14 on: 2011-11-07 05:03:39 »
Quote from: ra4king on 2011-11-06 20:04:45
Ok with fireworks2 and 30 fireworks at brightness of 15:
~1.02million particles
~65FPS

I <3 my computer Cheesy
I hate you. My stationary computer is on the other side of the globe at the moment. Damn this laptop! xD
I have another test that is just a particle sprayer optimized for multicores. Due to the fact that it is very memory limited, it also gets pretty big boosts from hyperthreading! I can get 1 million particles on my laptop's hyperthreaded dual core i5. Think about how many particles it could handle with a hyperthreaded quad-core running at a higher clock than my processor! Would you mind testing that other program out? It's a much more boring demo than this one though...

Quote from: EgonOlsen on 2011-11-07 00:57:52
Quote from: theagentd on 2011-11-06 19:54:43
It's black and white because all color channels are over 1, being capped to white. That's the whole point of the bloom filter. However, the brightness of 680,000 particles is on par with a nuclear bomb, so yeah, no wonder that it's white. Reduce the color strength of each star a little and/or scroll the mouse wheel up (a lot) to change reduce exposure of the tone mapper and make everything darker.
No, i think you misunderstood...it's fine after startup (but pretty bright as you've said) without pressing the mouse button (i.e. with bloom(?)). Clicking the mouse turns it into a black/white-picture, i.e. everything is either deep black or plain white. That includes rockets and such. Clicking it again turns the whole screen white, i.e. there's no other color visible than 255,255,255.
That sounds really weird... You mean the graphics break after you turn off and on the bloom filter?
Concerning the black and white: The particles have a brightness much higher than the screen can display. Therefore, even if the color of a particle is "red" as in (10, 1, 1) after tone mapping, it will still look like white because they are all clamped to 1. Just when the particles fade away you should be able to see their actual color. Sorry if I'm explaining the same thing again, but I just want to be sure you didn't misunderstand me. xD
You said the fireworks become white too? That's obviously not supposed to happen. Do you have time to post what graphics card you have, and maybe even screenshots showing the problems?

Thanks again to everyone who tested it!!!
There is no god.
Offline princec
« League of Dukes »

JGO Kernel
*****

Posts: 8086
Medals: 94


Eh? Who? What? ... Me?
« Reply #15 on: 2011-11-07 05:40:15 »
80fps with 350,000 particles or so with bloom on (Core i7, Vista 64, Java 7u01, Nvidia somethingorother) Impressively massive particle count. Makes my piddly sprite engine system look lame.

Cas Smiley
Play Revenge of the Titans - Play Titan Attacks - Play Droid Assault
Offline EgonOlsen

JGO Strike Force
***

Posts: 821
Medals: 6
« Reply #16 on: 2011-11-07 06:15:50 »
Quote from: theagentd on 2011-11-07 05:03:39
That sounds really weird... You mean the graphics break after you turn off and on the bloom filter?
Concerning the black and white: The particles have a brightness much higher than the screen can display. Therefore, even if the color of a particle is "red" as in (10, 1, 1) after tone mapping, it will still look like white because they are all clamped to 1. Just when the particles fade away you should be able to see their actual color. Sorry if I'm explaining the same thing again, but I just want to be sure you didn't misunderstand me. xD
You said the fireworks become white too? That's obviously not supposed to happen. Do you have time to post what graphics card you have, and maybe even screenshots showing the problems?
There's no fading and such...it's either bright white for all particles or black (in areas without particles). Just like as if you have converted the image to 1 bpp. After disabling, the image is all white (as said). I've already posted my specs in my first post.
jPCT, a 3D engine for Java -
jPCT-AE, a 3D engine for Android -
Robombs - a multiplayer game inspired by Bomberman
Offline ReBirth

JGO Wizard
****

Posts: 1272
Medals: 19
« Reply #17 on: 2011-11-07 07:55:32 »
Okay I can start it now. AMD Brazos E450.

Try 1000 + 200 + yes = explode
Try 1000 + 15 + yes = explode
Try 10 + 15 + no = work slowly

now, where can I see the FPS? I started from jar. Also I pressed ESC but it wont exit.

@Cero
actually I have Java 7u1 SDK here, no normal jre only installation.
Twitter @drabiter
Offline Cero

JGO Neuromancer
****

Posts: 1050
Medals: 18
« Reply #18 on: 2011-11-07 09:00:36 »
Quote from: ReBirth on 2011-11-07 07:55:32
@Cero
actually I have Java 7u1 SDK here, no normal jre only installation.

doesn matter, when you install jdk it installs jre too and it doesnt have the server folder by default

Program Files\Java\jre7\bin\
client folder but no server folder here

Program Files\Java\jdk1.7.0_01\jre\bin\
there is a server folder

so copy it. and there you go

sometimes I guess the server folder is there / or the system uses the jre inside the jdk folder by default
but yeah just copy it and thats it
Alchemic-Tempest.com  Google+  Twitter  Facebook
Offline EgonOlsen

JGO Strike Force
***

Posts: 821
Medals: 6
« Reply #19 on: 2011-11-07 11:27:01 »
A screen shot of the b/w-mode: http://jpct.de/pix//bwfire.png
jPCT, a 3D engine for Java -
jPCT-AE, a 3D engine for Android -
Robombs - a multiplayer game inspired by Bomberman
Offline ra4king

JGO Kernel
*****

Posts: 3155
Medals: 196


I'm the King!
« Reply #20 on: 2011-11-07 18:13:18 »
Quote from: theagentd on 2011-11-07 05:03:39
Quote from: ra4king on 2011-11-06 20:04:45
Ok with fireworks2 and 30 fireworks at brightness of 15:
~1.02million particles
~65FPS

I <3 my computer Cheesy
I hate you. My stationary computer is on the other side of the globe at the moment. Damn this laptop! xD
I have another test that is just a particle sprayer optimized for multicores. Due to the fact that it is very memory limited, it also gets pretty big boosts from hyperthreading! I can get 1 million particles on my laptop's hyperthreaded dual core i5. Think about how many particles it could handle with a hyperthreaded quad-core running at a higher clock than my processor! Would you mind testing that other program out? It's a much more boring demo than this one though...
I'd love to Smiley
I'm running an i7 2600k overclocked to 4GHz. That's 8 threads running at 4GHz......the power is too immense to grasp........
-Roi
Doodle Jump online!
Follow me on Twitter!
Online theagentd

JGO Wizard
****

Posts: 1392
Medals: 88
« Reply #21 on: 2011-11-07 20:47:15 »
@Princec
Well, it's on a computer, not a phone. The particles are also just colored points, so it's lighting fast. The bloom filter is terribly slow even though it looks awesome. It takes over 10 ms GPU time, but at least it's done in parallel to the CPU calculations... It obviously wouldn't work out in a real game though. You can't spend 13 ms blurring the screen. xD

@EgonOlsen
I think I've actually seen this problem before! A loooong time ago when I was doing 2D lighting for a RPG Maker style game I had an FBO which I drew the lighting and shadows to. Now, on our really old Radeon card in my family's computer, this turned out black and white. I never really checked into it (it was a really old computer), but it could very well have been the same problem.
I have no idea what's causing it though. The only thing I do for mouse clicks is change a boolean which controls if the bloom is rendered and added to the screen after the particle rendering or not. I'm not even unloading and reloading anything. I'm sadly gonna have to blame it on the Radeon drivers... T___T Sorry... Thanks for the screenshot!

@Rebirth
Why do you have to fire off 1000 fireworks per update?! xD That's 60 000 fireworks per second (at 60 FPS), each producing a trail of 4 particles per update (240 particles per second). You'll reach 10 000 000 particles long before they even explode from just the trail particles. And it doesn't "crash", it shuts itself down to prevent using excessive amounts of RAM. The engine is "stable" as in running but very slow up to about 16 000 000 particles. Sadly, I only have 4GB of RAM so that's pretty much the limit for me.
The FPS can be seen in the console window if you start it with the bat-file. If you don't have the server VM (which is faster) you can edit the bat-file with Notepad and remove the -server parameter. It also shows the number of particles.

@Cero
Thanks, good to know.

@Ra4king
Check your PMs. =)
There is no god.
Offline Mads

JGO Ninja
***

Posts: 674
Medals: 16


Directly directional
« Reply #22 on: 2011-11-07 20:51:02 »
1  
2  
3  
4  
5  
6  
7  
8  
9  
10  
11  
12  
13  
14  
15  
16  
17  
18  
Exception in thread "main" java.lang.IllegalStateException: Function is not supp
orted
        at org.lwjgl.BufferChecks.checkFunctionAddress(BufferChecks.java:58)
        at org.lwjgl.opengl.GL33.glGenSamplers(GL33.java:141)
        at fireworks.gfx.obj.Sampler.<init>(Sampler.java:11)
        at fireworks.HDRCore.init(HDRCore.java:32)
        at fireworks.tests.FireworkTest.init(FireworkTest.java:57)
        at fireworks.Core.<init>(Core.java:28)
        at fireworks.HDRCore.<init>(HDRCore.java:20)
        at fireworks.tests.FireworkTest.<init>(FireworkTest.java:53)
        at fireworks.tests.FireworkTest.main(FireworkTest.java:160)
        at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
        at sun.reflect.NativeMethodAccessorImpl.invoke(Unknown Source)
        at sun.reflect.DelegatingMethodAccessorImpl.invoke(Unknown Source)
        at java.lang.reflect.Method.invoke(Unknown Source)
        at org.lwjgl.util.mapped.MappedObjectClassLoader.fork(MappedObjectClassL
oader.java:76)
        at fireworks.tests.FireworkTest.main(FireworkTest.java:155)


Didn't run.
http://www.JavaDaemon.com/
Offline ra4king

JGO Kernel
*****

Posts: 3155
Medals: 196


I'm the King!
« Reply #23 on: 2011-11-07 22:14:55 »
@Mads, you're obviously running an Intel integrated gpu, amirite?  Pointing
-Roi
Doodle Jump online!
Follow me on Twitter!
Offline Mads

JGO Ninja
***

Posts: 674
Medals: 16


Directly directional
« Reply #24 on: 2011-11-07 22:16:59 »
Quote from: ra4king on 2011-11-07 22:14:55
@Mads, you're obviously running an Intel integrated gpu, amirite?  Pointing

I'm not. I'm running AMD Radeon HD  Cool
http://www.JavaDaemon.com/
Offline ra4king

JGO Kernel
*****

Posts: 3155
Medals: 196


I'm the King!
« Reply #25 on: 2011-11-07 22:19:42 »
Wow.....crappy drivers then? Try to update your drivers since the drivers you currently have don't even support OpenGL 3.3 >.>
-Roi
Doodle Jump online!
Follow me on Twitter!
Online theagentd

JGO Wizard
****

Posts: 1392
Medals: 88
« Reply #26 on: 2011-11-08 05:54:33 »
Quote from: ra4king on 2011-11-07 22:19:42
Wow.....crappy drivers then? Try to update your drivers since the drivers you currently have don't even support OpenGL 3.3 >.>
But it's using OpenGL 3.3 and its Sampler objects! Maybe an extension saved the day... If you can get some newer drivers, it would be nice to know if it fixed the problem.

Concerning the multithreaded particle test that Ra4king tested: it ran at almost 60 FPS with 3 million particles!!! Mind = blown. The next thing I try will be GPU particles, so I can avoid the single-threaded copying of the data to OpenGL. Eliminating my slow little CPU will be awesome!
There is no god.
Offline princec
« League of Dukes »

JGO Kernel
*****

Posts: 8086
Medals: 94


Eh? Who? What? ... Me?
« Reply #27 on: 2011-11-08 06:02:47 »
Blurring the whole screen shouldn't take 13ms! Render to a texture, then render that whole texture in one go to the screen with a bloom shader - should give similar results but be 10x faster. I think.

And the sprite engine I was referring to is the one I use on the desktop! (Well, it's the same as the Android one)

Cas Smiley
Play Revenge of the Titans - Play Titan Attacks - Play Droid Assault
Online theagentd

JGO Wizard
****

Posts: 1392
Medals: 88
« Reply #28 on: 2011-11-08 06:09:46 »
Quote from: princec on 2011-11-08 06:02:47
Blurring the whole screen shouldn't take 13ms! Render to a texture, then render that whole texture in one go to the screen with a bloom shader - should give similar results but be 10x faster. I think.

And the sprite engine I was referring to is the one I use on the desktop! (Well, it's the same as the Android one)

Cas Smiley
I am doing something like that, but I have a separable Gaussian blur so it's two passes to blur it. However, I want the bloom to be able to make the whole screen white from a single extremely bright pixel (like if you're looking into the sun), so I need a blur-kernel that is as big as the screen. To achieve this I basically have 8 levels of bloom, with the first level being at screen resolution, the second level being half as big, e.t.c. The problem is that I'm simply doing too many fullscreen passes with lots of texture lookups. Obviously in this particle example I can have this high quality bloom, but it is possible to scale it back to under 3 ms by reducing the number of levels and skipping the first full-sized level. It looks a lot worse for particles though.
There is no god.
Offline Cero

JGO Neuromancer
****

Posts: 1050
Medals: 18
« Reply #29 on: 2011-11-08 10:37:54 »
I would never program something that requires OpenGL 3, only if its really just experimental, just saying - but we talked about this before

If I would want to use new fancy pancy OpenGL, I would at least check what version is available and write code that works on machines until GL1.1

and I what theagentd would say, but the bottom line is - if you don't already have an audience /major publisher, then, filtering your audience like this will not help
Alchemic-Tempest.com  Google+  Twitter  Facebook
Pages: [1] 2
  Print  
 
 
Jump to:  

Featured Game
Latest Featured
Add yours!

Add your game by posting it in the showcase section.

The first screenshot will be displayed as a thumbnail.
Latest pastebins
obsidian_golem 2012-05-23 10:14:50
Danny02 2012-05-21 17:10:34
Danny02 2012-05-21 17:07:10
Danny02 2012-05-21 16:56:12
davedes 2012-05-21 13:59:23
obsidian_golem 2012-05-20 20:28:41
darkjava55 2012-05-12 16:14:40
Ultroman 2012-05-12 09:36:05
Ultroman 2012-05-11 22:49:53
Ultroman 2012-05-11 22:20:01
Powered by MySQL Powered by PHP Powered by SMF 1.1.16 | SMF © 2011, Simple Machines Valid XHTML 1.0! Valid CSS!
Page created in 0.158 seconds with 19 queries.