[maemo-developers] [maemo-developers] first impressions and some reports

[From: Gustavo Barbieri]

On 7/7/05, Ruda Moura <ruda.moura at indt.org.br> wrote:
> Gustavo Barbieri wrote:
> 
> > Hello,
> >
> > Now that I have the maemo environment set up, I started to love this
> > thing!!! And python!
> >
> > Thanks to Osvaldo and Ruda, I can run PythonCAD without any changes!
> >
> > Also, my game tutorial (pt_BR) runs perfectly (just need to check the
> > perfomance on a real device, since it's side scroll and redraws the
> > screen every time):
> 
> 
> Hello Gustavo, thank you for your feedback!
> 
> > http://www.gustavobarbieri.com.br/jogos/jogo.tar.gz
> >    Could someone run "python2.4 jogo/a06/jogo.py -f -r 800x480" on a
> > real device and check if performance is an issue... if it can do at
> > least 30fps it's good enough. If it's ok I'll develop a game using
> > that engine. Also, it can be optimized a bit, I wrote it as an
> > educative project.
> 
> SHORT ANSWER
> 
> It run slow, about 15 ~ 20 fps and crashed with Segmentation fault at
> the end (when ships goes to zero).

the fps is expected, but the segfault isn't :-(

 
> LONG ANSWER
> 
> It was visible slower than what was printed on screen (15 ~ 20 fps) but...
> You should think not in terms of PC, for a small device a game should
> take modification in order to run fast. First I notice some calculations
> with division ( like "something" / 16 ) and I change it to shifts ( >> 4
> ). Why? Because I know 770 lacks division opcodes in CPU. well, it as
> not a good improvement at all :)
> 
> The screen size *is* big, 800x480 is much data. People in this case
> should consider resize the rolling image to a small size, put some
> static pictures around, etc.
> 
> Maybe your code is good to teach OO but it isn't the best case for game
> speed. For your information, here is what PyGame show me about my
> display in Nokia 770.
> 
> > <VideoInfo(hw = 0, wm = 1,video_mem = 0
> >              blit_hw = 0, blit_hw_CC = 0, blit_hw_A = 0,
> >              blit_sw = 0, blit_sw_CC = 0, blit_sw_A = 0,
> >              bitsize  = 16, bytesize = 2,
> >              masks =  (63488, 2016, 31, 0),
> >              shifts = (11, 5, 0, 0),
> >              losses =  (3, 2, 3, 8)>
> 
> Again, thank you for your feedback and if you want to do more tests in
> "the real thing" I'm ready to help you.

I know, actually many of the classes could go away if the game is as
simpe as this one.
But it can give us some feedback if it's feasible to run a
sidescroller with this hardware. It'll not unless there is some
optimizations.
   Anyway, much games doesn't need to be side scrolling or that kind
of scroll... I repaint the whole screen every time, what could be done
is scroll the screen a bit more once and keep it fixed for 2-4
seconds, then use dirty rectangles to make it fast!

Anyway, I always planned a Sokoban pygame, maybe it's time to code it
and Battleship.
   But I don't promise anything soon since my job is demanding some
work these days and I'm also trying to run Opie(Qt-based pda
environment) on the emulator (will need some feedback on how good it
runs with the real device) and also the MAS
(www.mediaapplicationserver.net) development which I hope to help
soon.

-- 
Gustavo Sverzut Barbieri
---------------------------------------
Computer Engineer 2001 - UNICAMP
GPSL - Grupo Pro Software Livre
Cell..: +55 (19) 9165 8010
Jabber: gsbarbieri at jabber.org
  ICQ#: 17249123
   MSN: barbieri at gmail.com
 Skype: gsbarbieri
   GPG: 0xB640E1A2 @ wwwkeys.pgp.net