Closed GoogleCodeExporter closed 9 years ago
Yes i had iPad 1, and a 3g iPhone (awful performace even with private apis like
core surface). Now I have a iPad 2 (that runs mame perfectly) and iPhone 4 that
runs mame well for mame classics but it doesn't work well with neo games
although they are playable with frameskip.
When i had the iPad 1 I think that the performance was quite well, with some
few execeptions. Maybe with the new iOS releases or the jailbreak stuff, the
iPad 1 performance is not so well rigth now. I think like you. With such a
high renewal rate (the future iPhone 5 is almost here) working on doing some
costly effort optimizations that could jeopardize future builds is something
that deserves thought to be studied although it seems very interesting your
CADisplayLink develop and i thought to implement it at some point at the past.
Stuart Cartney (the developer behind iAmiga) has doing some work on
implementing a asm arm iOS 6800 CPU and he has offer me with a free license. I
have now working on an Android port of imame4all since i have gift my wife with
a Samsung Galaxy Tab 10.1 that i like very much so I'm very entertained when my
two babies leave me time to develop.. ahaha... i prefer to work on that
directoion so more people could enjoy imame4all on other devices but i doint
mind to add your improvements if you find relevant.
Said that, not much people has complain about iPad 1 performance. Your are an
advanced user who know when something is not right at all, but most people are
happy with current iPad 1 performance and when they switch to a more modern
device will be delighted with how well it works with the extra performance.
Original comment by seleuco....@gmail.com
on 19 Aug 2011 at 5:22
Yes, I think you're right about the current state of iMAME4all, it's pretty
much iOS compliant and hack free for the performance with the "safe drawing"
code and should work fine on future versions of iOS. It's just a matter of
letting the hardware catch up with the performance as the iPad 2 performance
proves your code is correct. I would consider using 3GS as a baseline for the
compiler as the ARMv7 CPU (from 3GS onwards) has some nice automatic CPU
optimisations.
My CADisplayLink code was just experimental so isn't fit to give to you, it was
quite simple but I had problems keeping the two threads (MAME core and iOS
drawing) in sync as the iOS was triggering the display draw (using the
CADisplayLink timer). Although looking at the iPad 2 I don't think it will be
needed as it's 99% in sync (no stutter).
I have a working XCode 4 project for the jailbroken version if you're
interested, it should just slot into the base directory of the source code and
will enable people to easily build with XCode 4. The XCode environment is
rather nice to develop and debug in.
I would like to tweak the game selection input, i.e. put a delay on the
joystick movement when selecting games, it can fly off down the list as there's
currently no delay. I also would like to see a favourites filter - I can help
with that if you want. I implemented a simple filter in ZodMAME that I think
would work OK.
Original comment by squid123...@gmail.com
on 22 Aug 2011 at 10:24
[deleted comment]
It would be nice to add your xcode project in a mame4all branch. I think that
the game filter is a good idea... it is one of the things i would like to add.
I really like a gp2x frontend that let you filter by game type, year,
manufacturer and so on... all info is in the drivers and it could be parsed in
some way. i really be very grateful if you add some kind of game filter and i
publish a version with your changes.
As i said, i not working in the iOS version now but an android version that i
have already running on my devices :). A honycomb tablet and a htc phone.
Android is like a kick in he ass.... far better iOS. At first instance i think
that i couldnt do a such smooth draw as iPad 2 without move to GL ES that i
dont like since i want to target all kind of hw out there... i have currenly 3
render paths, synched (legacy phones), threaded (two core phones) and finally a
hw accelerated for the honycomb tablet. sadly they havent added 2d hw
acceleration till android 3.0. incredible ahaha. But finally i have a tablet
render way nearly as good as the iPad version... <the phone version is really
good without filters... 60fps synched.
As tou see i am very entertained... so work in the base MAME code... it will
add to the android version also.
Original comment by seleuco....@gmail.com
on 22 Aug 2011 at 11:44
Seleuco, just to update you on what I'm up to. Been wrestling with the XCode
debug environment again but have finally got that sorted - I can write a Dev
document to show how to set it up. So far I've implemented the following code
changes:
* Use old style iCade controls if using an iPad, otherwise use the new
icontrolpad layout.
* Added delay on game selection list to make it easier to navigate through the
games and page up/down.
* Fixes to sound initialisation code.
* Changed hardcoded /var/mobile/Media/ROMs paths in source to preprocessor
defines (makes it easier to develop on XCode)
* Added ARMv6 and ARMv7 architectures to build, an iOS standard requirement but
code size is much bigger. Better performance on ARMv7 devices though.
Please let me know if you update the iOS source so I don't get too out of sync
with the SVN.
Next I'll look into a filter, I have some ideas that should work quite nicely.
Original comment by squid123...@gmail.com
on 27 Aug 2011 at 4:24
Fine. :)
can you implement a switch buttons in global options to select iPad/iCade
mapping? it should be easy. Maybe some user with an ipad have an iCade.
I'm n ot working on iOS source. I have made a folder copy named android with OS
dependant files.
keep the good work.
Original comment by seleuco....@gmail.com
on 29 Aug 2011 at 12:12
Yes, I was assuming someone with an iControlPad would only use it with an
iPhone and an iCade only with an iPad but people apparently use the iCP with
iPads too. I'll add an option for it.
Original comment by squid123...@gmail.com
on 31 Aug 2011 at 8:22
Just to update on progress:
* added option to choose iCade or iControlPad layout
* sleep processor between inputs on game select/settings screens, previously
CPU ran at constantly at 100%, now idles so better battery life.
Original comment by squid123...@gmail.com
on 2 Sep 2011 at 2:45
Good to know. I hope release the first android beta in one week. Twins let me
few time :)
Original comment by seleuco....@gmail.com
on 3 Sep 2011 at 12:34
Here's a version that looks ready for release. It has the following changes
which I think are significant enough for a release. I've attached the compiled
version. This version is bigger as it contains both the ARMv7 and ARMv6
compilations. I haven't done the ROM filtering yet, that's next on the list
but will take some time.
• key delay on game selection - eases game selection. (Currently only works
on safe render).
• added option to set Sound defaults
• added option to choose iCade or iControlPad layout
• sleep processor between inputs on game select screen, previously CPU ran at
constantly at 100%, now idles so better battery life.
• Fixed multiple key presses problem
• sound initialisation fixes
• Added ARMv7 exe for 3GS and later devices, speed boost for some games.
It maybe easier if I can commit the changes directly to the SVN source code,
but that is up to you.
Original comment by squid123...@gmail.com
on 3 Sep 2011 at 2:43
Attachments:
Nice work. This is a huge executable :)
is it possible to speed up a litle the game selection delay...?
I have added you as commiter so you can add your changes. Let me know if not
work.
cheers!
Original comment by seleuco....@gmail.com
on 4 Sep 2011 at 10:25
forget speed up... i think i was a little laggy on iPhone 4... but now i think
is OK. good work.
Original comment by seleuco....@gmail.com
on 4 Sep 2011 at 10:54
Yes, it's a large exe as it has both code compilations in there. Up to you if
you only want ARMv6 but including ARMv7 is an easy speed boost. The game
selection delay may need tweaking, i.e. quickly pressing keys is slow but I
think it's OK for now. Code for that is simple but a little tricky due to the
strange output from iCade with quick presses.
I've committed to the SVN most of my changes, still tidying things up before
confirming it looks OK. I've put the XCode project in there, which uses
relative paths, so it can be easily compiled from XCode 4. XCode appears to
require the CFBundleExecutable setting to match the name of the project, i.e.
"iMAME4all" instead of "mame". Seems to run OK for me if I use "iMAME4all" as
the exe name.
Original comment by squid123...@gmail.com
on 5 Sep 2011 at 12:39
i cant ldid (fake Pseudo-Signing for 3f arm6) with the universal build.
iPad-2-de-David:/Applications/iMAME4all.app root# ldid -S mame
codesign_allocate: for architecture armv6 object: mame malformed object
(unknown load command 4)
util/ldid.cpp(582): _assert(0:WEXITSTATUS(status) == 0)
iPad-2-de-David:/Applications/iMAME4all.app root#
i going to package it anyway and post it in touch arcade to see if some 3g user
can test it.
Thanks for your good work. i'm waiting for the game filtering. :) can i tag the
svn version?
Original comment by seleuco....@gmail.com
on 5 Sep 2011 at 6:04
I don't use ldid, just XCode to fake sign it.
OK, you can tag the SVN now, I've finished updating it.
Original comment by squid123...@gmail.com
on 5 Sep 2011 at 7:48
It's not the same... Ldid it is a jailbreak tool needed to run 3g not signed
apps as I know... I think Xcode not do the trick... But we'll see...
Original comment by seleuco....@gmail.com
on 5 Sep 2011 at 8:40
Ok. It seems the uunicersal build works well. I have uploaded the 1.9 package
to Cydia... but i have seen that you have fixed the readme. Sorry. Next version
then. I have taged 1.9 since i have begun beta testing MAME4droid and i will
commit changes in a few days... all changes are isolated anyway from core MAME
and iOS. Lot of work on MAME4droid ;)
Original comment by seleuco....@gmail.com
on 7 Sep 2011 at 11:38
That's good news. Good luck with MAME4droid, I'm sure it will be a big success.
Original comment by squid123...@gmail.com
on 8 Sep 2011 at 7:25
Closed.
Original comment by squid123...@gmail.com
on 22 Sep 2011 at 12:10
Original comment by squid123...@gmail.com
on 22 Sep 2011 at 1:06
Original issue reported on code.google.com by
squid123...@gmail.com
on 19 Aug 2011 at 1:51