using logcat to output you fps will reduce your fps.
On Jan 22, 6:37 am, Andre andranik.abra...@gmail.com wrote:
Hi Jason,
Thank you very much for test and code example
Best Regards, Andre
On 22 янв, 01:19, Jason Arora jaso...@gmail.com wrote:
Hi Andre,
I got around 60 fps without
Link to my .apk file:
http://rapidshare.com/files/338665093/Asteroids.apk.html
You can use following command to see output from the app:
adb logcat dalvikvm:D Asteroids:D *:S
There is a little issue with gc somewhere during the app init, not
solved yet..., and after this I have no gc calls
On Jan 21, 9:22 am, Andre andranik.abra...@gmail.com wrote:
Link to my .apk file:
http://rapidshare.com/files/338665093/Asteroids.apk.html
You can use following command to see output from the app:
adb logcat dalvikvm:D Asteroids:D *:S
There is a little issue with gc somewhere during the
Hi skink, thanks for test.
On what device you tested? What FPS you got, when not shooting
bullets?
Yes, I think I should avoid any rotation operation on Canvas...
Best Regards, Andre
On 21 янв, 12:41, skink psk...@gmail.com wrote:
On Jan 21, 9:22 am, Andre andranik.abra...@gmail.com wrote:
On Jan 21, 11:54 am, Andre andranik.abra...@gmail.com wrote:
Hi skink, thanks for test.
On what device you tested? What FPS you got, when not shooting
bullets?
Yes, I think I should avoid any rotation operation on Canvas...
Best Regards, Andre
a friend of mine tested your apk on his
Hi all, interesting thread, here's my tuppenceworth.
My first attempt at a simple game took the same approach, things
looked favourable when I first threw a few dozen sprites onto the
canvas along with some simple game logic to try and get a handle on
fps and timing. I was getting around 40-45fps
Hi TonyDoc,
I agree with you, recently I've started to study Open GL ES...
But I think it worth to know, what limitations exist for Canvas? And
what kind of games can be implemented with Canvas? As we know, thanks
to google, simple 2D games like Snake, Tetris and so on, can be
created without
Hi Andre,
I got around 60 fps without bullets on my Google Ion device. When
touching the screen it would go around 53 and when shooting it would
dip just under 40. Before doing my test, I closed all apps with Task
Killer.
You can remove the status bar by calling the following in your
Question.. I havent ran your app yet.. going to on my moto droid. Do you
show the fps, or do I need to run it attached via USB in debug mode to see
this info?
On Thu, Jan 21, 2010 at 2:19 PM, Jason Arora jaso...@gmail.com wrote:
Hi Andre,
I got around 60 fps without bullets on my Google Ion
Kevin, you have to use logcat while its connected via USB (debug
mode):
adb logcat dalvikvm:D Asteroids:D *:S
On Jan 21, 2:32 pm, Kevin Duffey andjar...@gmail.com wrote:
Question.. I havent ran your app yet.. going to on my moto droid. Do you
show the fps, or do I need to run it attached via
Hi Jason,
Thank you very much for test and code example
Best Regards, Andre
On 22 янв, 01:19, Jason Arora jaso...@gmail.com wrote:
Hi Andre,
I got around 60 fps without bullets on my Google Ion device. When
touching the screen it would go around 53 and when shooting it would
dip just under
Hi, thanks for reply
Yes, I can share my .apk, if you tell, how to upload something here?
I agree, that emulator should be much slower than real device, but let
me quote a snippet from the document Designing for Performance at
http://developer.android.com/guide/practices/design/performance.html:
After I wrote you, I realized that issue too. I remember reading that and
after I responded.. wondering if my response was inaccurate. I wonder how
much the speed/cores of your cpu play into the emulator performance? I know
that my small test apps run faster on my moto droid than they do my
I'm not running any sort of beastly machine, quad-core 2.8ghz, with an
nvidia 9800gtx, and almost all the time, emulator performance is lower than
physical device (especially with openGL, but even with canvas). Might have
something to do with the emulator specifically on linux, so your results
I'm running a Core 2 Duo T9500 (2.5Ghz, 6MB cache, etc) w/8GB RAM on
Win7 x64 and G1 still runs a little faster than Canvas and a lot
faster on OpenGL.
It's not going to be a direct comparison because the emulator is
emulating ARM so there's a huge loss in speed there from the
emulation.
You
I wouldn't expect a emulated code to run faster than on the target hardware.
Perhaps you are assuming that the emulated code is using the graphics
directly. Question is, does it?
On Jan 20, 2010 9:42 AM, Dan Sherman impact...@gmail.com wrote:
I'm not running any sort of beastly machine,
I was previously involved with game emulator development. No way that
the speed of the emulator for a 500+ Mhz processor will be faster on a
2GHz computer than the real device.
It doesn't matter if it's dual-core or not -- processor emulation is
pretty much single threaded.
On Jan 21, 2:15 am,
Hi Andre,
Everyone here has brought up great points:
1) Matrix manipulation is slow. Instead, rotate and store bitmaps
ahead of time (note: initial loading time will be longer).
2) Don't use getter/setters, instead make commonly used attributes
public.
3) Don't call System.currentTimeMillis() or
if you must, use width 1.
You gotta be kidding here -- are you saying that the Android compiler
won't do this for you automatically? Even compilers 20 years ago
could so such basic optimizations.
--
You received this message because you are subscribed to the Google
Groups Android
Hi to all,
Thanks a lot for many helpfull posts!
Agreed, I think that the main purpose of the emulator is emulate not
emulate as fast as possible. My dev machine is much slower than
yours (I have old laptop with Athlon 2.4 Ghz, 768MB RAM and Ubuntu
9.04 Janty). Now, with your help, I got ~25 FPS
Hello, thank you all for many helpful posts.
I tried to replace all getters and setters with member variables and
asteroids[i], bullets[i]... with corresponding asteroid,
bullet... variables.
Though this is very important issue, as I supposed, this doesn't
improve FPS in my case.
I agree, that
Hello Robert.
I want to thank you for sharing your experience with beginners, your
dev journal is extremely helpful!
I will change my code to work with Matrix as you suggest and post the
result here...
If I correctly understood your question about surface opacity - Yes my
surface is opaque, I
On Jan 19, 9:41 am, Andre andranik.abra...@gmail.com wrote:
Hello Robert.
I want to thank you for sharing your experience with beginners, your
dev journal is extremely helpful!
I will change my code to work with Matrix as you suggest and post the
result here...
If I correctly understood
On Jan 19, 10:33 am, skink psk...@gmail.com wrote:
On Jan 19, 9:41 am, Andre andranik.abra...@gmail.com wrote:
Hello Robert.
I want to thank you for sharing your experience with beginners, your
dev journal is extremely helpful!
I will change my code to work with Matrix as you suggest
That's OK, I've already tried use RGB_565 and other bitmap configs,
this makes sence if you using background image (wich should be RGB_565
for faster draw).
On 19 янв, 12:37, skink psk...@gmail.com wrote:
On Jan 19, 10:33 am, skink psk...@gmail.com wrote:
On Jan 19, 9:41 am, Andre
On Jan 19, 10:57 am, Andre andranik.abra...@gmail.com wrote:
That's OK, I've already tried use RGB_565 and other bitmap configs,
this makes sence if you using background image (wich should be RGB_565
for faster draw).
and you are sure that your bitmaps (asteroids, bullets, space ship)
are
Currently I use:
Bitmap bitmap = BitmapFactory.decodeResource(getResources(),
R.drawable.spaceship);
in that case:
bitmap.getConfig() == null;
as I mention earlier, I checked all kinds of bitmap configs with
little difference in performance
On 19 янв, 14:01, skink psk...@gmail.com wrote:
On
On Jan 19, 1:56 pm, Andre andranik.abra...@gmail.com wrote:
Currently I use:
Bitmap bitmap = BitmapFactory.decodeResource(getResources(),
R.drawable.spaceship);
in that case:
bitmap.getConfig() == null;
hmmm, null is a bit weird. it means that bitmaps config is not any
known public
Hi,
Yesterday, I've changed my code, to work with the matrix more
effectively...
My GameEntity class now looks like this:
--GAME ENTITY--
public class GameEntity {
public Bitmap mBitmap;
private Rect mBounds;
private Matrix mMatrix;
public int
I run my app on emulator, and currently have
not tested it on real device.
The emulator is not, in any way, going to run the same speed as a real
device.
Run it on a real device and see what FPS you get. Or if you want, publish
an APK and someone here will run it I'm sure (I'll do it =P)
-
Was just watching the video from back in May by Chris Pruett. He had a
performance slide comparing canvas to opengl and as he noted there, canvas
is much slower. I am sure you know this, but getting 30fps on a Canvas with
more than a dozen or so sprites might be difficult when you factor in game
Hi Andre,
you should definitely read the Designing for Performance document
from:
http://developer.android.com/guide/practices/design/performance.html
I think you might speed up your onDraw() method a little bit this way,
though I think you might not reach more than 30 or 35 FPS.
At first you
On Jan 18, 10:16 am, Andre andranik.abra...@gmail.com wrote:
From this output I can assume, that my main problem here is draw game
stuff wich includes:
- draw 1 spaceship
- draw 5 asteroids
- draw from 1 to 30 bullets
in drawAsteroids() i would assign GameEntity asteroid =
Hello Marc, thanks for quick reply.
I have read this document very carefully, though this is not my
production code,
I try to use some principles described there (local variables instead
of class fields).
Especially in those methods where I have loop: drawAsteroids,
drawBullets,
updateAsteroids,
Hello skink, thanks for advice
I agree, my array manipulations is not quite well.
And I think you right about getters too...(this is
one of the key principles in designing for performance
document), I don't yet replace them with members,
just becouse my updatePhysics is take ~1ms to execute...
On Jan 18, 1:01 pm, Andre andranik.abra...@gmail.com wrote:
Hello skink, thanks for advice
I agree, my array manipulations is not quite well.
And I think you right about getters too...(this is
one of the key principles in designing for performance
document), I don't yet replace them with
Yes, now I can do following:
private void drawAsteroids(Canvas canvas) {
GameEntity[] asteroids = mAsteroids;
GameEntity asteroid;
for(int i = 0; i MAX_ASTEROIDS; i++) {
asteroid = asteroids[i];
if(asteroid.isAlive()) {
canvas.setMatrix(null);
On Jan 18, 2:03 pm, Andre andranik.abra...@gmail.com wrote:
Yes, now I can do following:
private void drawAsteroids(Canvas canvas) {
GameEntity[] asteroids = mAsteroids;
GameEntity asteroid;
for(int i = 0; i MAX_ASTEROIDS; i++) {
asteroid = asteroids[i];
Let me clarify...
Lets look at two methods:
- updateAsteroidsPhysics()
- drawAsteroids(Canvas canvas)
They both use getters instead of members, and asteroids[i] instead of
asteroid variable, the only difference between them, that drawAsteroids
(Canvas canvas) is also use canvas calls, and I
On Jan 18, 2:49 pm, Andre andranik.abra...@gmail.com wrote:
Let me clarify...
Lets look at two methods:
- updateAsteroidsPhysics()
- drawAsteroids(Canvas canvas)
They both use getters instead of members, and asteroids[i] instead of
asteroid variable, the only difference between them,
The little stuff definitely adds up and helps but when talking about
such a small number of objects, I don't think it will double his FPS
to fix.
First of all, thanks Andre for actually posting your code so that we
have some context to work with and I'm also glad to hear that you've
read my dev
Hi,
1. Get rid of all get/set methods.
2. Eliminate some of the System.currentTimeMillis (two consecutive
calls leads to the same number)
3. Save a reference to the current object inside the array instead of
accessing an array lots of times (e.g.: asteroids[i])
4. Most important: don't call
On Jan 18, 7:19 pm, Robert Green rbgrn@gmail.com wrote:
The little stuff definitely adds up and helps but when talking about
such a small number of objects, I don't think it will double his FPS
to fix.
agreed. i haven't noticed that number of objects is that small...
pskink
--
You
Another thought.. you have the asteroid[i] array.. and then you do the if
(asterod.isAlive()) method. First.. eliminate the method call as you've been
told.
But for this particular example.. why are you keeping an array of objects
that may not be displayed? By this I mean.. if you have 30
44 matches
Mail list logo
