Hi,

I have a couple of questions about the latest incarnation of
devkitadv.

On our project, we're still using devkitadv 3(i think? gcc 3.0.2),
and it has served well up untli now.
I have got to a stage where the problem with pointers to member
functions and inteworking is getting very troublesome:
dereferencing a pointer to member function fails if the function is
compiled as thumb.

Q. is this fixed in the latest version?

If so :) is there a list anywhere of the changes - particularly to
system variables that the linkscript/crt0 uses...  I tried to get
version5 to compile our project, but when i eventually sorted out all
the mingw environment problems, i got a stream of errors:

f:\devkitadv-r5-beta-3\BIN\..\lib\gcc-lib\arm-agb-elf\3.2.2
\..\..\..\..\arm-agb-elf\bin\ld.exe: section .iwram [08070af4
  -> 080710f7] overlaps section .footer [08070af4 -> 08070aff]
f:\devkitadv-r5-beta-3\BIN\..\lib\gcc-lib\arm-agb-elf\3.2.2
\..\..\..\..\arm-agb-elf\lib\interwork\crt0.o: In function `_
_crt0_multiboot_start':
f:\devkitadv-r5-beta-3\BIN\..\lib\gcc-lib\arm-agb-elf\3.2.2
\..\..\..\..\arm-agb-elf\lib\interwork\crt0.o(.text+0xe0): un
defined reference to `__load_start_footer'
f:\devkitadv-r5-beta-3\BIN\..\lib\gcc-lib\arm-agb-elf\3.2.2
\..\..\..\..\arm-agb-elf\lib\interwork\crt0.o: In function `_
_crt0_reset':
f:\devkitadv-r5-beta-3\BIN\..\lib\gcc-lib\arm-agb-elf\3.2.2
\..\..\..\..\arm-agb-elf\lib\interwork\crt0.o(.text+0x1c0): u
ndefined reference to `__memmap_rom_start'
f:\devkitadv-r5-beta-3\BIN\..\lib\gcc-lib\arm-agb-elf\3.2.2
\..\..\..\..\arm-agb-elf\lib\interwork\crt0.o(.text+0x1d8): u
ndefined reference to `__appended_start'
f:\devkitadv-r5-beta-3\BIN\..\lib\gcc-lib\arm-agb-elf\3.2.2
\..\..\..\..\arm-agb-elf\lib\interwork\crt0.o(.text+0x1e4): u
ndefined reference to `__load_start_data'
f:\devkitadv-r5-beta-3\BIN\..\lib\gcc-lib\arm-agb-elf\3.2.2
\..\..\..\..\arm-agb-elf\lib\interwork\crt0.o(.text+0x1e8): u
ndefined reference to `__load_stop_data'
f:\devkitadv-r5-beta-3\BIN\..\lib\gcc-lib\arm-agb-elf\3.2.2
\..\..\..\..\arm-agb-elf\lib\interwork\crt0.o(.text+0x1f8): u
ndefined reference to `__load_start_ewram'
f:\devkitadv-r5-beta-3\BIN\..\lib\gcc-lib\arm-agb-elf\3.2.2
\..\..\..\..\arm-agb-elf\lib\interwork\crt0.o(.text+0x204): u
ndefined reference to `__load_start_iwram'
f:\devkitadv-r5-beta-3\BIN\..\lib\gcc-lib\arm-agb-elf\3.2.2
\..\..\..\..\arm-agb-elf\lib\interwork\crt0.o(.footer+0x0): u
ndefined reference to `__load_stop_appended'
f:\devkitadv-r5-beta-3\BIN\..\lib\gcc-lib\arm-agb-elf\3.2.2
\..\..\..\..\arm-agb-elf\lib\interwork\crt0.o(.footer+0x4): u
ndefined reference to `__load_start_appended'
f:\devkitadv-r5-beta-3\BIN\..\lib\gcc-lib\arm-agb-elf\3.2.2
\..\..\..\..\arm-agb-elf\lib\interwork\crt0.o(.footer+0x8): u
ndefined reference to `__load_stop_appended'
collect2: ld returned 1 exit status
make: *** [EOD_demo.elf] Error 1

cheers

Col

Hi,

Sorry to send this here, but I received the following e-mail today:

>**************** eManager Notification *****************
>
>The following message was blocked because it contains sensitive
content.
>
>Source mailbox: francis_lillie@...
>Destination mailbox(es): gbadev@yahoogroups.com
>Rule/Policy: Sexual Discrimination
>Action: Quarantine to
D:\Trend\Quarantine\2003-08-30\10-45->28.57692\DFImessagebody3f507238e15
c.tmp
>
>Content filter has detected a sensitive e-mail.
>
>******************* End of message *********************

The problem is that I haven't sent an e-mail to gbadev for months now,
so either someone's trying to have a laugh at my expense by spoofing my
mail address, or more seriously (for someone) there's someone out there
who has my mail address and may well have a virus that's shot off a
spoofed mail.  Just thought I'd bring this to your attention.  Sorry to
have troubled you.

FGL

There resonable support for OS X on the software side of things.
http://forum.gbadev.org has some recent posts on the subject; just
search for 'Mac'.

There is also (somewhat sketchy) support for the Flash Advance Xtreme
USB on OS X. See http://www.cs.helsinki.fi/u/jikorhon/condev/gba/

> Hi,
> I have not kept up on this stuff enough I guess. I didn't realize
there was a USB linker
> available. I had kind of given up on OS X, but now maybe there is
hope.
>
> I am off to order a linker and I would be interested in the software
once the linker
> arrives.
>
> Is there a good place to find out about programming on OS X for the
gameboy - like a
> step by stop guide (tutorial)?
>
> Thanks,
> John

--- In gbadev@yahoogroups.com, eli curtz <spam1@n...> wrote:
> I've been porting Ulrich Hecht's Linux F2A linker code to OS X, and am
> about to the point where I need some other people to look at it.
>
> If you've got a USB F2A linker and a machine running OS X with the
> development tools installed and would be willing to test this out
> please drop me a line. Anybody who also has a PC they'd be willing to
> compare things on would be especially helpful. BTW this is NOT a fully
> featured tool, it only writes and reads from the card, but I'd be happy
> to help (given my limited time) if anybody wants to do a complete
> cartridge management tool.
>
> Thanks,
> eli
> gbastuff@n...

Hi,
I have not kept up on this stuff enough I guess. I didn't realize there was a
USB linker
available. I had kind of given up on OS X, but now maybe there is hope.

I am off to order a linker and I would be interested in the software once the
linker
arrives.

Is there a good place to find out about programming on OS X for the gameboy -
like a
step by stop guide (tutorial)?

Thanks,
John

Hi Tom,

> Any change of getting the source or a linux binary for Conv2AAS.exe ?

Sure. I'll put together a Linux binary ASAP.
--
Cheers,
James.

     /\  apex                                 James Daniels
    //\\  designs            james.daniels@...
   //__\\                       http://www.apex-designs.net

On Thursday 21 Aug 2003 17:09, James Daniels wrote:
> http://www.apex-designs.net/tools_aas.html
>
> I've also created an AAS discussion list which you can join from the
> page above.

Any change of getting the source or a linux binary for Conv2AAS.exe ?

Thanks

Tom

--
  ^__^               Tom Badran
  (oo)\______        Imperial College
(__)\       )\/\
     ||----w |
     ||     ||       Using Debian SID

Hi Martin,

> Yeah I was comparing the quality of "Cream Of The Earth" with your demo and
> a quickly hacked together version with our internal engine on GBA. Those are
> the same settings I have used to.

Ok, I'll investigate and see what I can do to increase the sound quality
to Argonaut standards in future versions.
--
Cheers,
James.

     /\  apex                                 James Daniels
    //\\  designs            james.daniels@...
   //__\\                       http://www.apex-designs.net

Hi,

If I'm not mistaken sprites with the same priorities as layers are always
drawn above the corresponding layer. Shouldn't be that hard actually ;)

Good luck!

Collin

----- Original Message -----
From: "gunterpete" <reno@...>
To: <gbadev@yahoogroups.com>
Sent: Thursday, August 21, 2003 10:23 PM
Subject: [gbadev] Background Priority Woes


> Hi all,
>
> I'm new to GBA programming, and have gone through all the tutorials on
> the PERN project. I'm trying to write a RPG demo, and am having some
> difficulty with layering backgrounds and sprites.
>
> I'm Using mode 0, with backgrounds 0 and 1 enabled. I want the GB to
> draw background 0 first, then my sprite, then background 1. I thought
> this would be possible with the priorities set to 3, 3 and 2
> respectively like so:
>
>    REG_BG0CNT = ( BG_COLOR256 | TEXTBG_SIZE_256x256 | (31 <<
> SCREEN_SHIFT) | WRAPAROUND | PRIORITY(3) );
>    REG_BG1CNT = ( BG_COLOR256 | TEXTBG_SIZE_256x256 | (30 <<
> SCREEN_SHIFT) | WRAPAROUND | PRIORITY(2) );
> ...
>    //Initialise the player sprite
>    sprites[playerName].attribute0 = ( COLOR_16 | TALL | 56 );
>    sprites[playerName].attribute1 = ( SIZE_32| 120 );
>    sprites[playerName].attribute2 = ( 8 | PRIORITY(3) );
>
> Unfortunately with these settings the sprite seems to be drawn under
> both layers. I've tried other priorities, including 3, 2,1, with which
> the sprite is drawn on top of both layers.
>
> Can anyone help me with this please?
>
> -Regards,
> Pete Gunter
>
>
>
>
>
>
>
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>

Yeah I was comparing the quality of "Cream Of The Earth" with your demo and
a quickly hacked together version with our internal engine on GBA. Those are
the same settings I have used to.

-----Original Message-----
From: James Daniels [mailto:james.daniels@...]
Sent: 21 August 2003 17:02
To: gbadev@yahoogroups.com
Subject: Re: [gbadev] Apex Audio System


Hi Martin,

> Forgive me for saying but your downloadable demo really does sound ike it
> distorts and crackles a lot. So much so I had to check my installation!
This
> is even when using the high mixer rates. The hi-hat and snare in "Cream of
> the Earth" particularly stand out as being "nasty".

Are you sure that's not the samples in the MOD that are at fault rather
than the mixer? I just played CreamOfTheEarth.mod in both AAS and Winamp
and the only significant difference I could notice was that Winamp was
louder. I was using the following settings:

VBA:
44 KHz sound output and 1x volume.

AASDemo.gba:
32 KHz, 4 channel, stereo, dynamic off, MOD volume 256.

If you try a MOD with higher quality samples (like Style Blend, the
first song in the demo) it should sound a lot better.

> I would also suggest adding XM support for up to 32 channels, it really
> knocks the socks off publishers when they hear some of the 32 channel XMs
> playing on a GBA through headphones.

Ok, I've added that to my "To Do" list.
--
Cheers,
James.

     /\  apex                                 James Daniels
    //\\  designs            james.daniels@...
   //__\\                       http://www.apex-designs.net







Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

At 00:16 2003-08-21 -0600, you wrote:

> > >I think I found the bug when compiling the bzip sources, which compiled
> > >fine on 3.2.2...
> >
> >   Buggy where? With the compiler output? Or did the compiler crash?
>
>gbadev archives, message 14266
>
>arm-thumb-elf-gcc -O3 -mthumb -c blocksort.c
>blocksort.c: In function `mainSort':
>blocksort.c:805: internal compiler error: Segmentation fault
>Please submit a full bug report,
>with preprocessed source if appropriate.
>See <URL:<http://gcc.gnu.org/bugs.html>http://gcc.gnu.org/bugs.html> for
>instructions.

   Yep.

   Tried with gcc-3.3.1 and it is still failing for thumb mode with
optimization levels >= 2. It fails  in line 1058, though:

----8<----
$ arm-thumb-elf-gcc -v -O2 -mthumb -c blocksort.c
Reading specs from d:\gba\lib\gcc-lib\arm-thumb-elf\3.3.1\specs
Configured with: ../gcc-3.3.1/configure --prefix=d:/gba
--exec-prefix=d:/gba --host=i686-pc-mingw32 --target=arm-thumb-elf
--enable-targets=arm-elf --with-cpu=arm7tdmi --enable-interwork
--enable-multilib --with-newlib --disable-threads --disable-shared
--disable-win32-registry --with-gcc --with-gnu-ld --with-gnu-as
--with-stabs --with-headers=../newlib-1.11.0/newlib/libc/include
--enable-languages=c,c++
Thread model: single
gcc version 3.3.1
   d:\gba\lib\gcc-lib\arm-thumb-elf\3.3.1\cc1.exe -quiet -v -iprefix
/cygdrive/d/gba/bin/../lib/gcc-lib/arm-thumb-elf\3.3.1\ -D__GNUC__=3
-D__GNUC_MINOR__=3 -D__GNUC_PATCHLEVEL__=1 -D__ARM_ARCH_4T__ -D__ELF__
-D__USES_INITFINI__ blocksort.c -quiet -dumpbase blocksort.c -mthumb
-auxbase blocksort -O2 -version -o c:/windows/temp/ccmCNFlp.s
   GNU C version 3.3.1 (arm-thumb-elf)
          compiled by GNU C version 3.2 20020927 (prerelease).
GGC heuristics: --param ggc-min-expand=51 --param ggc-min-heapsize=40891
ignoring duplicate directory "d:/gba/lib/gcc-lib/arm-thumb-elf/3.3.1/include"
ignoring duplicate directory "d:/gba/arm-thumb-elf/sys-include"
ignoring duplicate directory "d:/gba/arm-thumb-elf/include"
#include "..." search starts here:
#include <...> search starts here:
   /cygdrive/d/gba/lib/gcc-lib/arm-thumb-elf/3.3.1/include
    /cygdrive/d/gba/arm-thumb-elf/sys-include
   /cygdrive/d/gba/arm-thumb-elf/include
   d:/gba/include
End of search list.
blocksort.c: In function `mainSort':
blocksort.c:1058: internal compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.
---->8----

arm mode compiles ok.


--
PabloBleyerKocik /"...I didn't want to be kissing Kevin Spacey.
   pbleyer        / Come on! Lying there naked with rose petals?"
    @embedded.cl /- Kirsten Dunst on turning down American Beauty

Hi all,

I'm new to GBA programming, and have gone through all the tutorials on
the PERN project. I'm trying to write a RPG demo, and am having some
difficulty with layering backgrounds and sprites.

I'm Using mode 0, with backgrounds 0 and 1 enabled. I want the GB to
draw background 0 first, then my sprite, then background 1. I thought
this would be possible with the priorities set to 3, 3 and 2
respectively like so:

    REG_BG0CNT = ( BG_COLOR256 | TEXTBG_SIZE_256x256 | (31 <<
SCREEN_SHIFT) | WRAPAROUND | PRIORITY(3) );
    REG_BG1CNT = ( BG_COLOR256 | TEXTBG_SIZE_256x256 | (30 <<
SCREEN_SHIFT) | WRAPAROUND | PRIORITY(2) );
...
    //Initialise the player sprite
    sprites[playerName].attribute0 = ( COLOR_16 | TALL | 56 );
    sprites[playerName].attribute1 = ( SIZE_32| 120 );
    sprites[playerName].attribute2 = ( 8 | PRIORITY(3) );

Unfortunately with these settings the sprite seems to be drawn under
both layers. I've tried other priorities, including 3, 2,1, with which
the sprite is drawn on top of both layers.

Can anyone help me with this please?

-Regards,
Pete Gunter

Hi Martin,

> Forgive me for saying but your downloadable demo really does sound ike it
> distorts and crackles a lot. So much so I had to check my installation! This
> is even when using the high mixer rates. The hi-hat and snare in "Cream of
> the Earth" particularly stand out as being "nasty".

Are you sure that's not the samples in the MOD that are at fault rather
than the mixer? I just played CreamOfTheEarth.mod in both AAS and Winamp
and the only significant difference I could notice was that Winamp was
louder. I was using the following settings:

VBA:
44 KHz sound output and 1x volume.

AASDemo.gba:
32 KHz, 4 channel, stereo, dynamic off, MOD volume 256.

If you try a MOD with higher quality samples (like Style Blend, the
first song in the demo) it should sound a lot better.

> I would also suggest adding XM support for up to 32 channels, it really
> knocks the socks off publishers when they hear some of the 32 channel XMs
> playing on a GBA through headphones.

Ok, I've added that to my "To Do" list.
--
Cheers,
James.

     /\  apex                                 James Daniels
    //\\  designs            james.daniels@...
   //__\\                       http://www.apex-designs.net

Hi,

>- The AAS example program seems to have serious stability
>  problems. It eventually crashed on just about every file I tried.
>  I wonder if I didn't mess up an interrupt or something.

I believe I have now fixed this issue. It was caused by r13 being used
by the mixer. This doesn't cause any problems with Single and Fast
Interrupts (r13 is restored when the mixer returns), but can cause
problems with Multiple Interrupts. An updated version of AAS is now
available here:

http://www.apex-designs.net/tools_aas.html

I've also created an AAS discussion list which you can join from the
page above.
--
Cheers,
James.

     /\  apex                                 James Daniels
    //\\  designs            james.daniels@...
   //__\\                       http://www.apex-designs.net

> >Seriously though, I was able to build 3.2 successfully under mingw.
> >However,
> >it seemed to run slightly slower and so I've stuck with cygwin ever since.
> >I would be very interested in someone with actual compile time benchmarks
> >backing up the "mingw is faster" claims.
> >
> >Cheers,
> >-Pete

I can tell you that using cygwin to build DevKitAdvance itself takes about
90 minutes while using MinGW to build DevKitAdvance itself takes just over
60 minutes.

Even if it's the x86 backend and not the ARM backend, the only variable
here is MinGW vs Cygwin, so I think its fair.

Forgive me for saying but your downloadable demo really does sound ike it
distorts and crackles a lot. So much so I had to check my installation! This
is even when using the high mixer rates. The hi-hat and snare in "Cream of
the Earth" particularly stand out as being "nasty".
I wish I could send you the demo of the 32 channel engine we developed
internally here for Bionicle to show you a good comparison. Unfortunately it
is internal only.
I would also suggest adding XM support for up to 32 channels, it really
knocks the socks off publishers when they hear some of the 32 channel XMs
playing on a GBA through headphones.

-----Original Message-----
From: James Daniels [mailto:james.daniels@...]
Sent: 20 August 2003 22:31
To: gbadev@yahoogroups.com
Subject: Re: [gbadev] Apex Audio System


Hi Pete,

> - The AAS library supports only four-channel MOD's, while Krawall
>   supports many-channel .xm and .s3ms with lots of nice effects.

I can't argue with that, although if there's demand then xm and s3m
support can be added in future.

> - The AAS library lacks support for even basic MOD effects.  I
>   actually had to look pretty hard to find a MOD file that even
>   played correctly at all.  Here's a typical output:
>
>       Adding MOD CALIBRA.MOD...Done!
>       Unsupported effects:
>       0: Arpeggio (1)
>       6: Vibrato + Volume Slide (Vibrato ignored) (33)
>       B: Position Jump (1)
>       ED: Note Delay (2)
>       song_length:52 last_pattern:28

You're right. I will expand the list of supported effects considerably.

> - The AAS example program seems to have serious stability
>   problems. It eventually crashed on just about every file I tried.
>   I wonder if I didn't mess up an interrupt or something.

I haven't had any problems with it - did you make any changes to it
before compilation? (You mentioned switching to "multiple interrupts" so
would it be possible to have a look at the modified code?)

If there do turn out to be problems with the library then I'll obviously
take that very seriously and fix them ASAP.

> - For ROCKIT.mod at least, the AAS_CONFIG_DYNAMIC_ON setting
>   produces output which is noticeably distorted

Yes, I don't recommend using dynamic mixing yet (although it does work
very well in some cases, which I why it's been included), it's still
somewhat experimental. This is mentioned in the docs, although perhaps
it needs to be more prominent. I've got some ideas on how to improve it
in the future though.

> P.S. Is there some kind of inside joke going on with names
> like "krapPlay" and "AAS"?  I've never seen so much crap
> and ass in one place!  ;-D

LOL! :)
--
Cheers,
James.

     /\  apex                                 James Daniels
    //\\  designs            james.daniels@...
   //__\\                       http://www.apex-designs.net







Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

Hi Pete,

> - The AAS library supports only four-channel MOD's, while Krawall
>   supports many-channel .xm and .s3ms with lots of nice effects.

I can't argue with that, although if there's demand then xm and s3m
support can be added in future.

> - The AAS library lacks support for even basic MOD effects.  I
>   actually had to look pretty hard to find a MOD file that even
>   played correctly at all.  Here's a typical output:
>
>       Adding MOD CALIBRA.MOD...Done!
>       Unsupported effects:
>       0: Arpeggio (1)
>       6: Vibrato + Volume Slide (Vibrato ignored) (33)
>       B: Position Jump (1)
>       ED: Note Delay (2)
>       song_length:52 last_pattern:28

You're right. I will expand the list of supported effects considerably.

> - The AAS example program seems to have serious stability
>   problems. It eventually crashed on just about every file I tried.
>   I wonder if I didn't mess up an interrupt or something.

I haven't had any problems with it - did you make any changes to it
before compilation? (You mentioned switching to "multiple interrupts" so
would it be possible to have a look at the modified code?)

If there do turn out to be problems with the library then I'll obviously
take that very seriously and fix them ASAP.

> - For ROCKIT.mod at least, the AAS_CONFIG_DYNAMIC_ON setting
>   produces output which is noticeably distorted

Yes, I don't recommend using dynamic mixing yet (although it does work
very well in some cases, which I why it's been included), it's still
somewhat experimental. This is mentioned in the docs, although perhaps
it needs to be more prominent. I've got some ideas on how to improve it
in the future though.

> P.S. Is there some kind of inside joke going on with names
> like "krapPlay" and "AAS"?  I've never seen so much crap
> and ass in one place!  ;-D

LOL! :)
--
Cheers,
James.

     /\  apex                                 James Daniels
    //\\  designs            james.daniels@...
   //__\\                       http://www.apex-designs.net

> >I think I found the bug when compiling the bzip sources, which compiled
> >fine on 3.2.2...
>
>   Buggy where? With the compiler output? Or did the compiler crash?

gbadev archives, message 14266

arm-thumb-elf-gcc -O3 -mthumb -c blocksort.c
blocksort.c: In function `mainSort':
blocksort.c:805: internal compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:<http://gcc.gnu.org/bugs.html>http://gcc.gnu.org/bugs.html> for
instructions.

binutils 2.13.2.1
gcc-3.2.2
newlib 1.11.0
mingw 2.0.0-3 msys 1.0.8
Athlon 1.7mhz 512mb windows 2000
downloaded cygwin tonight (2003-08-20).
I also ran this on Linux 2.4.20 (Slackware 9), same computer.

I have a piece of code that tends to take a lot of crunch time.  I ran gcc
three times each per optimization:

timer
arm-thumb-elf-gcc -O0 -c game.c
timer
arm-thumb-elf-gcc -O0 -c game.c
timer
arm-thumb-elf-gcc -O0 -c game.c
timer

timer
arm-thumb-elf-gcc -O3 -c game.c
timer
arm-thumb-elf-gcc -O3 -c game.c
timer
arm-thumb-elf-gcc -O3 -c game.c
timer

In this forum we have seen the second run go faster than the first, I didnt
see that this time.

-O0 option
MingW  6 seconds
cygwin  5-6 seconds
Linux 5 seconds

-O3 option
Mingw 186 seconds
cygwin 256 seconds
Linux  181 seconds

David

At 02:14 AM 8/20/2003 -0400, you wrote:
>David Welch wrote:
>
> >Is anyone else having problems building gcc 3.2.3 under mingw (for arm)?
> >
>Dude, get with the times!  The latest release is 3.3.1.  :-D
>
>Seriously though, I was able to build 3.2 successfully under mingw.
>However,
>it seemed to run slightly slower and so I've stuck with cygwin ever since.
>I would be very interested in someone with actual compile time benchmarks
>backing up the "mingw is faster" claims.
>
>Cheers,
>-Pete
>
>
>
>
>
>
>
>Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

At 23:13 2003-08-19 -0600, you wrote:

>Hmmm, back when 3.3 came out, I tried it and posted info showing that gcc
>3.3 was pretty buggy, so I reverted back to 3.2.2, and wanted to try 3.2.3
>not 3.3 (there is  a 3.3.1 out btw)...

   IIRC, 3.2 had some issues with some ARM targets that were fixed in 3.3,
although I don't know if they affected arm7tdmi.

>I think I found the bug when compiling the bzip sources, which compiled
>fine on 3.2.2...

   Buggy where? With the compiler output? Or did the compiler crash?

>with mingw 2.0.0 -3 binutils fails to complete the build (I didnt record
>the exact error messages) with mingw 3.whatever rc4, binutils builds fine
>but gcc fails pretty early in the build...

   I used GCC "mingw special 20020817-1", which I guess is the same you have
(latest distro from SF). All built smoothly (GCC, G++, etc.), though I
remember some issues with newlib headers that were easy to fix (perhaps it
was my own mingw installation). GDB compiled fine with the latest cygwin
(~1 month ago).

   Regards.


--
PabloBleyerKocik / "Personally I don`t understand the motivation
   pbleyer        / to build robots in human form since humans are
    @embedded.cl / so available and inexpensive." -- Lou Boyd

James Daniels wrote:

>The Apex Audio System (AAS) is an extremely efficient audio mixer and
>Protracker MOD player for the GBA. Its main features are a low memory
>footprint and record breaking performance. It is designed to allow
>developers to easily make full use of the GBA's sound hardware with
>little overhead.
>
Attached is a benchmark comparison against Krawall which appears to
support these speed claims.  I ran the test on VisualBoyAdvance 1.5.1.
I started with the example programs included with Krawall and AAS,
and replaced the main loop with a fixed loop that increments a global
variable, logging the amount of work done every 64 vblanks to the console.
To compute the CPU usage, I divided these numbers by the measurements
with audio disabled.  I also aode a few other minor changes:

- I changed crt0.s so both apps use "multiple interrupts"
- I moved the Krawall kramWorker() call to a vcount interrupt
- I disabled Krawall's waitstate reprogramming (for fairness)
- I tried a few different combinations of the AAS_CONFIG_MIX_
   and AAS_CONFIG_DYNAMIC_ parameters

The SDK versions are:

- http://mind.riot.org/krawall/  Krawall Private Version 23032003
- http://www.apex-designs.net/   Apex Audio System v1.01

The tune I used is an old Amiga file ROCKIT.mod.  Since Krawall does
not support mod directly, I used ModPlug to convert to .xm and .s3m
and then used .s3m because it was slightly faster.

The attached benchmarks seem to indicate that AAS is significantly
faster than Krawall.  I should qualify these conclusions with a few
comments in Krawall's favor:

- The AAS library supports only four-channel MOD's, while Krawall
   supports many-channel .xm and .s3ms with lots of nice effects.

- The AAS library lacks support for even basic MOD effects.  I
   actually had to look pretty hard to find a MOD file that even
   played correctly at all.  Here's a typical output:

       Adding MOD CALIBRA.MOD...Done!
       Unsupported effects:
       0: Arpeggio (1)
       6: Vibrato + Volume Slide (Vibrato ignored) (33)
       B: Position Jump (1)
       ED: Note Delay (2)
       song_length:52 last_pattern:28

- The AAS example program seems to have serious stability
   problems. It eventually crashed on just about every file I tried.
   I wonder if I didn't mess up an interrupt or something.

- For ROCKIT.mod at least, the AAS_CONFIG_DYNAMIC_ON setting
   produces output which is noticeably distorted

- The AAS API is somewhat limited.  When the mod is playing, you get
   only 4 sfx channels which must be managed manually.  In contrast,
   Krawall supports many channels and allocates them automatically.

Even so, the speed increase could more than compensate for these
limitations.  I will be very interested to hear other people's
results.

Cheers,
-Pete

P.S. Is there some kind of inside joke going on with names
like "krapPlay" and "AAS"?  I've never seen so much crap
and ass in one place!  ;-D

I posted compile times a while back, mingw was quite a bit faster.  I have
a project/program that really works gcc, I will try to do another benchmark...

Are you using the stock mingw 3.x.x and msys 1.9.x or have you also added
other packages?

David


At 02:14 AM 8/20/2003 -0400, you wrote:
>David Welch wrote:
>
> >Is anyone else having problems building gcc 3.2.3 under mingw (for arm)?
> >
>Dude, get with the times!  The latest release is 3.3.1.  :-D
>
>Seriously though, I was able to build 3.2 successfully under mingw.
>However,
>it seemed to run slightly slower and so I've stuck with cygwin ever since.
>I would be very interested in someone with actual compile time benchmarks
>backing up the "mingw is faster" claims.
>
>Cheers,
>-Pete
>
>
>
>
>
>
>
>Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

Hmmm, back when 3.3 came out, I tried it and posted info showing that gcc
3.3 was pretty buggy, so I reverted back to 3.2.2, and wanted to try 3.2.3
not 3.3 (there is  a 3.3.1 out btw)...

I think I found the bug when compiling the bzip sources, which compiled
fine on 3.2.2...

with mingw 2.0.0 -3 binutils fails to complete the build (I didnt record
the exact error messages) with mingw 3.whatever rc4, binutils builds fine
but gcc fails pretty early in the build...

David




At 09:54 PM 8/19/2003 -0400, you wrote:

>   gcc-3.3. Here.
>
>   Builds and works perfectly ;^)
>
>   Which problems did you have?
>
>   Regards.
>
>At 10:38 2003-08-19 -0600, you wrote:
>
> >Is anyone else having problems building gcc 3.2.3 under mingw (for arm)?
> >
> >I tried with ming 2.0.0 -3 and the latest 3.x rc4  and msys 1.9 I think it
> >is...binutils 2.14 wouldnt build under 2.0.0 -3 but does under the new one,
> >gcc appeared to not build under the new mingw...
> >
> >binutils 2.14 and gcc 3.2.3 and newlib 1.11.0 DOES build on the latest
> >Slackware...
> >
> >David
>
>--
>PabloBleyerKocik /"...I didn't want to be kissing Kevin Spacey.
>   pbleyer        / Come on! Lying there naked with rose petals?"
>    @embedded.cl /- Kirsten Dunst on turning down American Beauty
>
>
>
>
>
>
>Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

David Welch wrote:

>Is anyone else having problems building gcc 3.2.3 under mingw (for arm)?
>
Dude, get with the times!  The latest release is 3.3.1.  :-D

Seriously though, I was able to build 3.2 successfully under mingw.
However,
it seemed to run slightly slower and so I've stuck with cygwin ever since.
I would be very interested in someone with actual compile time benchmarks
backing up the "mingw is faster" claims.

Cheers,
-Pete

Helo.

Its gone a while that the cartlib.c was modificated. Since we had published a
small how-to Tutorial about using the catlib.c on www.devgba.de we have futurer
problems with people that uses F2A flash modules and cannot test it.

Thats why we ask around on (r. ziegler, dady, jeff, ex. visolys) - can someone
send infomations about using the flash features of the f2a carts?
dady, maybe can u resend me that visoly extreme mod. from a year ago?

we glad about any help or infos, thx.

dummy NewBee

   ----- Original Message -----
   From: Pablo Bleyer Kocik
   To: gbadev@yahoogroups.com
   Sent: Wednesday, August 20, 2003 3:54 AM
   Subject: Re: [gbadev] gcc 3.2.3



     gcc-3.3. Here.

     Builds and works perfectly ;^)

     Which problems did you have?

     Regards.

   At 10:38 2003-08-19 -0600, you wrote:

   >Is anyone else having problems building gcc 3.2.3 under mingw (for arm)?
   >
   >I tried with ming 2.0.0 -3 and the latest 3.x rc4  and msys 1.9 I think it
   >is...binutils 2.14 wouldnt build under 2.0.0 -3 but does under the new one,
   >gcc appeared to not build under the new mingw...
   >
   >binutils 2.14 and gcc 3.2.3 and newlib 1.11.0 DOES build on the latest
   >Slackware...
   >
   >David

   --
   PabloBleyerKocik /"...I didn't want to be kissing Kevin Spacey.
     pbleyer        / Come on! Lying there naked with rose petals?"
      @embedded.cl /- Kirsten Dunst on turning down American Beauty



         Yahoo! Groups Sponsor
               ADVERTISEMENT




   Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.

gcc-3.3. Here.

   Builds and works perfectly ;^)

   Which problems did you have?

   Regards.

At 10:38 2003-08-19 -0600, you wrote:

>Is anyone else having problems building gcc 3.2.3 under mingw (for arm)?
>
>I tried with ming 2.0.0 -3 and the latest 3.x rc4  and msys 1.9 I think it
>is...binutils 2.14 wouldnt build under 2.0.0 -3 but does under the new one,
>gcc appeared to not build under the new mingw...
>
>binutils 2.14 and gcc 3.2.3 and newlib 1.11.0 DOES build on the latest
>Slackware...
>
>David

--
PabloBleyerKocik /"...I didn't want to be kissing Kevin Spacey.
   pbleyer        / Come on! Lying there naked with rose petals?"
    @embedded.cl /- Kirsten Dunst on turning down American Beauty

Is anyone else having problems building gcc 3.2.3 under mingw (for arm)?

I tried with ming 2.0.0 -3 and the latest 3.x rc4  and msys 1.9 I think it
is...binutils 2.14 wouldnt build under 2.0.0 -3 but does under the new one,
gcc appeared to not build under the new mingw...

binutils 2.14 and gcc 3.2.3 and newlib 1.11.0 DOES build on the latest
Slackware...

David

wow :)
i guess that if you have a good enough compiler, that kind of approach
will always work well...

though it's totally OT, did working in lisp help the development? and
why lisp? sorry- i'm just curious :)


alex:)


Eddie Edwards wrote:
> That's true (I worked on J&D) but the LISP dialect, GOAL, was actually
> compiled into R5900 native code, rather than being a scripting language.
> Even the graphics engine was written in LISP!!!
>
> Eddie
>
> -----Original Message-----
> From: lx [mailto:nal@...]
> Sent: 16 August 2003 21:00
> To: gbadev@yahoogroups.com
> Subject: Re: [gbadev] Re: Scripting on the GBA
>
>
> apparently jak&daxter (on the ps2) was mostly written in lisp
>
> ...well, i thought that was cool, anyway :)
>
>
>
> Julian Squires wrote:
>
>>Hi.  My two cents.
>>
>>On Tue, Aug 12, 2003 at 04:55:19AM -0000, Damian Yerrick wrote:
>>
>>
>>>What is the fundamental difference between scripts and C anyway?
>>
>>
>>In theory, not too much.  In practice, overhead -- scripts are generally
>>interpreted, usually at a bytecode level; scripting languages usually
>>have all the storage issues hidden from view, usually with automatic
>>variable creation (on a stack) and a garbage-collected memory allocator;
>>scripting languages almost always have a larger runtime support system
>>than C, especially if they support lisp-ish eval and similar.
>>
>>IMO, the real issue is that if you really want to push the boundries on
>>the GBA, you can't always afford those overhead issues.  A game like
>>Maniac Mansion is the perfect example of the style of game where
>>scripting is great... you're unlikely to be doing all that much in
>>realtime except some decompression, there's lots of special case game
>>logic for different areas, and the game's entities don't have much to do
>>in terms of AI.
>>
>>Also, scripting languages are often used with the consideration that the
>>designers can write in them, and test with them, more easily than they
>>could in C.  Of course, this only applies with a certain set of games.
>>
>>There will always be people who frown on any language that isn't C,
>>especially languages with the overhead of the typical scripting
>>language;  but on the GBA, for many games, they're right.
>>
>>(What I don't get is why no one considers any of the nice modern
>>languages for their game logic code, like modula3, or even ocaml or
>>sather)
>>
>>Cheers.
>>
>
>
>
>
> ________________________________________________________________________
> This email has been scanned for all viruses by the MessageLabs Email
> Security System. For more information on a proactive email security
> service working around the clock, around the globe, visit
> http://www.messagelabs.com
> ________________________________________________________________________
>
>
>
>
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>
>
>
>
>
>
>
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>
>
>



________________________________________________________________________
This email has been scanned for all viruses by the MessageLabs Email
Security System. For more information on a proactive email security
service working around the clock, around the globe, visit
http://www.messagelabs.com
________________________________________________________________________

Hi all,

The Apex Audio System (AAS) is an extremely efficient audio mixer and
Protracker MOD player for the GBA. Its main features are a low memory
footprint and record breaking performance. It is designed to allow
developers to easily make full use of the GBA's sound hardware with
little overhead.

It supports up to 8 channels and takes an average of just 2.8% CPU to
play a 4 channel MOD at 16 KHz. It's free for non-commercial use and can
be downloaded here:

http://www.apex-designs.net/tools_aas.html

Feedback welcome!
--
Cheers,
James.

     /\  apex                                 James Daniels
    //\\  designs            james.daniels@...
   //__\\                       http://www.apex-designs.net

That's true (I worked on J&D) but the LISP dialect, GOAL, was actually
compiled into R5900 native code, rather than being a scripting language.
Even the graphics engine was written in LISP!!!

Eddie

-----Original Message-----
From: lx [mailto:nal@...]
Sent: 16 August 2003 21:00
To: gbadev@yahoogroups.com
Subject: Re: [gbadev] Re: Scripting on the GBA


apparently jak&daxter (on the ps2) was mostly written in lisp

...well, i thought that was cool, anyway :)



Julian Squires wrote:
> Hi.  My two cents.
>
> On Tue, Aug 12, 2003 at 04:55:19AM -0000, Damian Yerrick wrote:
>
>>What is the fundamental difference between scripts and C anyway?
>
>
> In theory, not too much.  In practice, overhead -- scripts are generally
> interpreted, usually at a bytecode level; scripting languages usually
> have all the storage issues hidden from view, usually with automatic
> variable creation (on a stack) and a garbage-collected memory allocator;
> scripting languages almost always have a larger runtime support system
> than C, especially if they support lisp-ish eval and similar.
>
> IMO, the real issue is that if you really want to push the boundries on
> the GBA, you can't always afford those overhead issues.  A game like
> Maniac Mansion is the perfect example of the style of game where
> scripting is great... you're unlikely to be doing all that much in
> realtime except some decompression, there's lots of special case game
> logic for different areas, and the game's entities don't have much to do
> in terms of AI.
>
> Also, scripting languages are often used with the consideration that the
> designers can write in them, and test with them, more easily than they
> could in C.  Of course, this only applies with a certain set of games.
>
> There will always be people who frown on any language that isn't C,
> especially languages with the overhead of the typical scripting
> language;  but on the GBA, for many games, they're right.
>
> (What I don't get is why no one considers any of the nice modern
> languages for their game logic code, like modula3, or even ocaml or
> sather)
>
> Cheers.
>



________________________________________________________________________
This email has been scanned for all viruses by the MessageLabs Email
Security System. For more information on a proactive email security
service working around the clock, around the globe, visit
http://www.messagelabs.com
________________________________________________________________________





Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

Hey, nice demo. The link in your message doesn't work, so I just went
through the files on scene.org and found it at
http://www.scene.org/file_dl.php?
url=ftp://ftp.scene.org/pub/parties/2003/assembly03/mobile_demo/counting_clouds_\
by_brainwave_team.zip&id=193021 (for anybody else that
wants to see it). Great background music, do you have a
MOD/XM/whatever of it I could download?
Thanks,
~Dennis

--- In gbadev@yahoogroups.com, "Aleksey N. Malov" <vivid@b...> wrote:
> Hello gbadev,
>
>   We released our first GBA demo at Assembly'2003.
> It took the last (9th) place but this is pretty nice demo.
> Just download and enjoy.
>
> http://www.scene.org/file.php?file=%2Fparties%2F2003%2Fassembly03%
2Fmobile_demo%2Fcounting_clouds_by_brainwave_team.zip&fileinfo
>
> np: Fat Boy Slim - Soul surfing
> -- [ZX][GBA][Scene][Anime][Ot.A.Ma.][B.A.K.A. S.T.][Powered by
Jenechka]
> Best regards,
>  Vivid / Brainwave Team                    mailto:vivid@b...