Atomic AGD

[oss003]

Hi guys,

I started a new project: Atomic AGD and want to share my findings with you.

Jonathan Cauldwell has written a program called Arcade Game Designer for the ZX Spectrum. It's a game creation program with build-in editors. You can define a font, objects, sprites, blocks, screens and add event based script commands describing how the game runs. You can start the game from the menu for testing. The program is written for a ZX Spectrum 48/128KB. It's a nice program with a good engine and a lot of games are written with AGD. If the game is ready, you can save it in a tapefile as stand alone game which runs without the editors. AGD can be downloade here: https://jonathan-cauldwell.itch.io/arcade-game-designer

Jonathan has also written an editorless compiler running in Windows. This compiler needs 2 files to generate a game: a game engine and a script file. The script file is compiled to an assembler file and merged with the game engine file. A Z80 assembler source file is generated which can be compiled to Z80 code. This is very interesting for porting AGD to the Atom because if I can convert the game engine to Atomic 6502 code and change the compiler to generate 6502 assembler from the script file, you have an assembler file which can be compiled to 6502 code.

The big advantage is that the engine file is system dependent but the game itself, written as a script file, can be easily ported.
So the AGD system consists of:

- an general game engine file with routines to manage the game
- a game specific script file with font-, objects-, sprites-, blocks- and screen data and commands describing the game

The engine file can't be 100% compatible with all ported systems due to resolution-, colour- and memory differences. The script commands however can be 100% compatible so this means to run a game, you have to change the data and compile it for your system.

These are the steps I want to follow:

1. Convert the Z80 game engine to 6502 monochrome engine
2. Change the compiler to generate 6502 code
3. Add editors to easily change font-, objects-, sprites-, blocks- and screens data
4. Add colourmode

For now I am working on step 1 and this is for 70% completed. You can see some examples here:

ZX Spectrum version Terrapins: https://www.youtube.com/watch?v=ea_dsXmjL90
Atom version Terrapins: https://www.youtube.com/watch?v=Vh4ykI0H0iE&t=53s

ZX Spectrum version Dodgy: https://youtu.be/pyMmxWgJLTo
Atom version Dodgy: https://youtu.be/P9hZtE3y31M

Greetings
Kees

[kieranhj]

Great to see your progress on this Kees! Really excited about this one. You know I’m a big supporter of the project, just a little tied up right now. I look forward to hearing what everyone else has to say and what the appetite might be for a BBC version...

On a tangential note, I see that Jonathan has started writing an editor for Windows as using text for sprites & screen layouts wasn’t so intuitive. I wonder if you could combine the editor, the compiler and an emulator together for the ultimate 8-bit game designer...

[SimonSideburns]

Great progress on this Kees.

I've been playing with AGD on the Speccy for a while myself, and while I've not designed anything of release quality, I've often wondered about the possibilities of the system for other machines.

I'd love something similar for the Beeb, and if it could be developed on Windows (or any similar OS with a GUI based system) that would be fantastic.

A bit beyond my skill levels however.

[oss003]

Just a quick test with Dodgy Geezer in 4 colour mode.
It's the same game engine, only different sprite/block data.

Greetings
Kees

[oss003]

And here is the demo video: https://youtu.be/HTGFHAPp5eU

Greetings
Kees

[oss003]

Ok guys, an update on Atomic AGD: jumping, falling and fall too far is working now so platform games will work.

Jump: https://youtu.be/SA5be2REgkg

Greetings
Kees

[oss003]

Getting closer ......... some more demo's are running now:
The compiler is working now and the engine is improved.

Alien: https://www.youtube.com/watch?v=GqzbrLY1-Mg
Baloons: https://www.youtube.com/watch?v=dUa8H39J890

Greetings
Kees

[oss003]

Another update ...... object handling is working fine now.

https://www.youtube.com/watch?v=Fzu30RzXWcc

Greetings
Kees

[oss003]

And getting closer to a working AGD engine ......

Keyhandling: https://www.youtube.com/watch?v=jQVzN6gHfO0
Inventory menu: https://www.youtube.com/watch?v=X5ND3kcyNQM
Ladders: https://www.youtube.com/watch?v=_hnHqsoLKuk
Multi screen: https://www.youtube.com/watch?v=iJCGFGjHs7c

Greetings
Kees

[hoglet]

Hi Kees,

The last two videos look amazing! Is this the performance at 1MHz?

Do you thing there is scope for further optimization to speed things up? Is this the drawing code:
https://github.com/oss003/Atomic-AGD/bl ... .inc#L4772

I was wondering whether it would be possible to use the 6502 bus sniffer as a profiling tool. Do you have a way of profiling the engine?

I'm also not at all familiar with AGD. Are the game definitions platform independent?

Anyway, really nice.....

Dave

[oss003]

Hi Dave,

yes, these videos are at 1 mHz.

There certainly is need for further optimalization, for now I have converted the Z80 code to 6502 code and that's not optimal.
Especially the sprite routine can be improved. The sprites are 16x16 pixels and for every line the old sprite (2 bytes) are removed and the new sprite (2 bytes) are placed. Z80 does this with swapping registers but this is slow in 6502 code so this can be optimized.

The latest code can be found here: https://github.com/oss003/Atomic-AGD-Script-Compiler

The AGD compiler works like this:

- You have an AGD file with generic graphic definitions and script commands defining the game, system independent
- You need an engine file for the Atom with all game depending routines in 6502 code
- The compiler converts the AGD file to a 6502 assembler file and merges it with the engine file to a game source file
- The game source file can be assembled to 6502 code with CA65 and LD65.

I'm a member of the AGD development team and someone is developing a Windows AGDX Studio program to design the graphics and export them as AGD file. The goal is the create a system independent AGD file which can be compiled on several systems, with some small changes.

Greetings
Kees

[oss003]

Dave, I don't understand what you mean with profiling the engine, do you mean debugging?
It would be helpfull if we have an option in Atomulator for real time memory monitoring, maybe use the Memory Viewer window because I don't think anyone is using this.

PS I have added the AGD file of Atic to give you an impression how it works.

Greetings
Kees

[hoglet]

Dave, I don't understand what you mean with profiling the engine, do you mean debugging?

Very often when trying to speed up a program people just guess where the time is going. For example, with AGD we might guess that the sprite drawing is expensive. If you are wrong, then you have wasted a day re-writing code that was already fast enough.

Profiling lets you understand exactly where the time is spent when a programme executes. With this information, you can spend time optimizing the code that actually takes 50% of the time, rather than 0.5% of the time.

For the Atom, I'm we can hook up the 6502 Bus Sniffer to the the tube port, and that will let us capture every instruction that executes:

With the 6502decoder, it's very easy to turn this back into an instruction stream. The stream length is pretty much unlimited. i.e. you could capture severl minutes of operation if you wanted.

I've just added simple profiling to the 6502decoder:
https://github.com/hoglet67/6502Decoder ... 0e6234d1f2

Here's a very small example:

Code: Select all

 0800          :LL0
 0800 A2 00     LDX @0
 0802          :LL1
 0802 A0 00     LDY @0
 0804          :LL2
 0804 20 07 08  JSR LL3
 0807          :LL3
 0807 68        PLA
 0808 68        PLA
 0809 88        DEY
 080A D0 F8     BNE LL2
 080C CA        DEX
 080D D0 F3     BNE LL1
 080F EA        NOP
 0810 EA        NOP
 0811 EA        NOP
 0812 4C 00 08  JMP LL0

And the profile I have just captured of it executing for about 10 seconds:

Code: Select all

$ decode6502 --profile=0800,08FF --quiet --byte atom_demo_capture.bin
0800 :   157998 (  1.477851%)
0802 :   473991 (  4.433525%)
0804 :  2843928 ( 26.600979%)
0807 :  1895952 ( 17.733986%)
0808 :  1895952 ( 17.733986%)
0809 :   947976 (  8.866993%)
080a :  1263968 ( 11.822658%)
080c :   315992 (  2.955664%)
080d :   421323 (  3.940889%)
080f :   105330 (  0.985215%)
0810 :   105330 (  0.985215%)
0811 :   105330 (  0.985215%)
0812 :   157995 (  1.477823%)
     : 10691065 (100.000000%)

The total at the bottom shows 10,691,065 bus cycles were captured, which is ~10.7s at 1MHz. The single most expensive instruction was the JSR.

If you can generate me an ATM file and an assembler log for your LADDERS example, I'll have a quick try at profiling it. It would be useful to have a non-trivial example to work with.

Dave

P.S. Actually, you can only hook up the 6502 Bus Snooper to the Tube port on the Atom if you bypass the data bus buffer (IC4). This is quite easy to do though. I initially forgot about IC4 and spent 10 minutes being very confused.

P.P.S. It would also be possible to add this sort of profiling to Atomulator.

[oss003]

Thanks, I understand it now.
I have added the Atictest demo and a disas list.

mloop at #0FBF is the main gameloop.
If you can time mloop, you will have a good idea about the timing.
The routine dspr handles displaying the sprites.

I suspect that the sprite routine sspric at #1E6E is the most time consuming routine.
Can you give it a try?

Greetings
Kees

[oss003]

And here's the LADDER demo.

Greetings
Kees

[hoglet]

I suspect that the sprite routine sspric at #1E6E is the most time consuming routine.
Can you give it a try?

Here's the profile of aticdemo (I've removed any line that was less that 0.1% of the total.

Code: Select all

10f0 :    22656 (  0.192498%)
1109 :    22656 (  0.192498%)
110c :    22656 (  0.192498%)
1110 :    14160 (  0.120311%)
162a :    19360 (  0.164493%)
162c :    15488 (  0.131595%)
162f :    23232 (  0.197392%)
1631 :    19360 (  0.164493%)
1a6a :    13776 (  0.117049%)
1a8e :    14532 (  0.123472%)
1a91 :    14532 (  0.123472%)
1c68 :    33405 (  0.283827%)
1c6a :    20043 (  0.170296%)
1c6b :    33405 (  0.283827%)
1c6d :    40086 (  0.340593%)
1c6f :    26724 (  0.227062%)
1c70 :    40086 (  0.340593%)
1c72 :    40086 (  0.340593%)
1c75 :    20039 (  0.170263%)
1c7a :    40086 (  0.340593%)
1c7d :    20014 (  0.170050%)
1c82 :    40086 (  0.340593%)
1c85 :    19650 (  0.166957%)
1cb0 :    12456 (  0.105833%)
1d46 :    12528 (  0.106445%)
1d4e :    12528 (  0.106445%)
1d56 :    12528 (  0.106445%)
1d5e :    12528 (  0.106445%)
1d66 :    12528 (  0.106445%)
1d79 :    12528 (  0.106445%)
1e79 :    77442 (  0.657990%)
1e7c :    77442 (  0.657990%)
1e7f :    77442 (  0.657990%)
1e82 :    77436 (  0.657939%)
1e85 :    25812 (  0.219313%)
1e86 :    37858 (  0.321663%)
1e8e :    55420 (  0.470879%)
1e90 :   138550 (  1.177198%)
1e92 :   110840 (  0.941758%)
1e95 :   138550 (  1.177198%)
1e97 :   166260 (  1.412638%)
1e99 :   138550 (  1.177198%)
1e9b :   138545 (  1.177156%)
1e9d :   138545 (  1.177156%)
1e9f :   110836 (  0.941724%)
1ea2 :   138545 (  1.177156%)
1ea4 :   166254 (  1.412587%)
1ea6 :   166254 (  1.412587%)
1ea9 :    82210 (  0.698502%)
1eae :   138545 (  1.177156%)
1eb0 :   138545 (  1.177156%)
1eb2 :   138545 (  1.177156%)
1eb4 :   166254 (  1.412587%)
1eb6 :   166254 (  1.412587%)
1eb9 :    83127 (  0.706293%)
1ebe :   138545 (  1.177156%)
1ec0 :    55418 (  0.470862%)
1ec1 :    83127 (  0.706293%)
1ec3 :    55418 (  0.470862%)
1ec5 :    83127 (  0.706293%)
1ec7 :    79663 (  0.676861%)
1ec9 :    17320 (  0.147160%)
1ecb :   166254 (  1.412587%)
49c6 :    82602 (  0.701833%)
49c8 :    82602 (  0.701833%)
49ca :    82602 (  0.701833%)
49cc :    82602 (  0.701833%)
49ce :    82602 (  0.701833%)
49d0 :    82602 (  0.701833%)
49d2 :    82602 (  0.701833%)
49d4 :    82602 (  0.701833%)
49d6 :    82602 (  0.701833%)
49d8 :    82602 (  0.701833%)
49da :    82602 (  0.701833%)
49dc :    82602 (  0.701833%)
49de :    82602 (  0.701833%)
49e0 :    82602 (  0.701833%)
49e2 :    82602 (  0.701833%)
49e4 :    82602 (  0.701833%)
49e6 :    82602 (  0.701833%)
49e8 :    82602 (  0.701833%)
49ea :    82602 (  0.701833%)
49ec :    82602 (  0.701833%)
49ee :    82602 (  0.701833%)
49f0 :    82602 (  0.701833%)
49f2 :    82602 (  0.701833%)
49f4 :    82602 (  0.701833%)
49f6 :   165204 (  1.403665%)
fe6b :  2129056 ( 18.089646%)
fe6e :  1596446 ( 13.564295%)

I clearly need to find a way to generate some kind of a call graph, so it's possible to view this hierarchically.

Dave

[oss003]

Can you try the mloop routine, hopefully it fits within 1/50 sec?

Maybe you can do somthing like this:

Code: Select all

10f0 :    22656 (  0.192498%) +++++++++
1109 :    22656 (  0.192498%) +++++++++
110c :    22656 (  0.192498%) +++++++++
1110 :    14160 (  0.120311%) ++++++
162a :    19360 (  0.164493%) ++++++++
162c :    15488 (  0.131595%) ++++++
162f :    23232 (  0.197392%) ++++++++++
1631 :    19360 (  0.164493%) ++++++
1a6a :    13776 (  0.117049%) +++++
1a8e :    14532 (  0.123472%) ++++++
1a91 :    14532 (  0.254321%) ++++++++++++

Greetings
Kees

[BigEd]

(A thought: we don't want to know which instructions are executed most often, so much as where in the code the most time is spent. So, we need cycles consumed rather than times executed - if we don't already have that [and it seems we do, oops] - and perhaps it's worth binning the results into 8 byte or 16 byte paragraphs, in the absence of any idea of function boundaries or loop extents or basic blocks...??)

[hoglet]

(A thought: we don't want to know which instructions are executed most often, so much as where in the code the most time is spent. So, we need cycles consumed rather than times executed - if we don't already have that - and perhaps it's worth binning the results into 8 byte or 16 byte paragraphs, in the absence of any idea of function boundaries or loop extents or basic blocks...??)

The counts are already 6502 cycle counts:
https://github.com/hoglet67/6502Decoder ... ain.c#L361

I'd really like to get to the call graph though.

[BigEd]

Yes, sorry about that! Do you think aggregating by paragraphs would help? The idea is to compress the information and make the hot code stand out more, and with no flow analysis needed.

[hoglet]

Yes, sorry about that! Do you think aggregating by paragraphs would help? The idea is to compress the information and make the hot code stand out more, and with no flow analysis needed.

Here's an example with 16-byte buckets:

Code: Select all

0640 :     2076 (  0.025815%) *
0650 :     1038 (  0.012907%) *
0e30 :        6 (  0.000075%) *
0e40 :      560 (  0.006964%) *
0e50 :      994 (  0.012360%) *
0e60 :        6 (  0.000075%) *
0f80 :       24 (  0.000298%) *
0f90 :       26 (  0.000323%) *
0fa0 :       24 (  0.000298%) *
0fb0 :     1056 (  0.013131%) *
0fc0 :     4170 (  0.051853%) *
0fd0 :     3460 (  0.043025%) *
0fe0 :     4152 (  0.051630%) *
0ff0 :     4498 (  0.055932%) *
1000 :     2941 (  0.036571%) *
1010 :     2415 (  0.030030%) *
1070 :       12 (  0.000149%) *
1080 :        3 (  0.000037%) *
10d0 :     7552 (  0.093908%) *
10e0 :     9440 (  0.117385%) **
10f0 :    68000 (  0.845573%) *********
1100 :    71764 (  0.892378%) *********
1110 :    16992 (  0.211294%) ***
1120 :        2 (  0.000025%) *
1130 :       85 (  0.001057%) *
1140 :      190 (  0.002363%) *
1150 :      155 (  0.001927%) *
12f0 :      112 (  0.001393%) *
1300 :      694 (  0.008630%) *
1310 :      306 (  0.003805%) *
1320 :      184 (  0.002288%) *
1330 :      224 (  0.002785%) *
1340 :      224 (  0.002785%) *
1350 :      240 (  0.002984%) *
1360 :      264 (  0.003283%) *
1370 :      304 (  0.003780%) *
1380 :      176 (  0.002189%) *
1390 :      240 (  0.002984%) *
13a0 :       48 (  0.000597%) *
1550 :     8712 (  0.108333%) **
1560 :    12584 (  0.156481%) **
1570 :    17424 (  0.216666%) ***
1580 :    13552 (  0.168518%) **
1590 :    10648 (  0.132407%) **
15e0 :     9680 (  0.120370%) **
15f0 :    12584 (  0.156481%) **
1600 :    14520 (  0.180555%) **
1610 :    11132 (  0.138425%) **
1620 :    71148 (  0.884718%) *********
1630 :    55660 (  0.692126%) *******
1640 :     2904 (  0.036111%) *
16e0 :        8 (  0.000099%) *
16f0 :       26 (  0.000323%) *
1700 :       22 (  0.000274%) *
1710 :       23 (  0.000286%) *
1720 :       30 (  0.000373%) *
1730 :      102 (  0.001268%) *
1740 :     8976 (  0.111616%) **
1750 :     3701 (  0.046022%) *
1760 :     3267 (  0.040625%) *
1770 :     1691 (  0.021027%) *
1780 :     4523 (  0.056243%) *
1790 :     4000 (  0.049740%) *
1850 :      280 (  0.003482%) *
1860 :      880 (  0.010943%) *
1870 :     2548 (  0.031684%) *
1880 :     2760 (  0.034320%) *
1890 :     6900 (  0.085801%) *
18a0 :     7590 (  0.094381%) *
18b0 :     4416 (  0.054912%) *
19c0 :     3864 (  0.048048%) *
19d0 :     5520 (  0.068641%) *
19e0 :     3450 (  0.042900%) *
19f0 :     2208 (  0.027456%) *
1a20 :     7552 (  0.093908%) *
1a30 :     7552 (  0.093908%) *
1a40 :     9688 (  0.120469%) **
1a50 :    48440 (  0.602346%) *******
1a60 :    48566 (  0.603913%) *******
1a70 :     4844 (  0.060235%) *
1a80 :    26988 (  0.335593%) ****
1a90 :    27334 (  0.339895%) ****
1aa0 :     1384 (  0.017210%) *
1ab0 :     3460 (  0.043025%) *
1be0 :      692 (  0.008605%) *
1bf0 :    23707 (  0.294794%) ***
1c00 :    19412 (  0.241386%) ***
1c10 :    31169 (  0.387583%) ****
1c20 :    18657 (  0.231998%) ***
1c30 :    13749 (  0.170967%) **
1c40 :     7860 (  0.097738%) *
1c50 :     7487 (  0.093100%) *
1c60 :   156807 (  1.949878%) ********************
1c70 :   160509 (  1.995912%) ********************
1c80 :    64170 (  0.797947%) ********
1c90 :    30285 (  0.376591%) ****
1ca0 :    32005 (  0.397979%) ****
1cb0 :    29953 (  0.372462%) ****
1cc0 :    22542 (  0.280307%) ***
1cd0 :    21675 (  0.269526%) ***
1ce0 :    19074 (  0.237183%) ***
1cf0 :    20808 (  0.258745%) ***
1d00 :    24276 (  0.301869%) ****
1d10 :     7803 (  0.097029%) *
1d20 :     2094 (  0.026039%) *
1d30 :    37329 (  0.464182%) *****
1d40 :    62640 (  0.778922%) ********
1d50 :    62640 (  0.778922%) ********
1d60 :    56376 (  0.701029%) ********
1d70 :    43138 (  0.536416%) ******
1d80 :    14673 (  0.182457%) **
1d90 :    19105 (  0.237569%) ***
1da0 :    11713 (  0.145650%) **
1db0 :     7755 (  0.096433%) *
1dc0 :    18964 (  0.235815%) ***
1dd0 :    24998 (  0.310847%) ****
1de0 :    21748 (  0.270434%) ***
1df0 :    20904 (  0.259939%) ***
1e00 :    39015 (  0.485147%) *****
1e10 :    34660 (  0.430993%) *****
1e20 :    43105 (  0.536006%) ******
1e30 :    35980 (  0.447407%) *****
1e40 :    42067 (  0.523099%) ******
1e50 :    15679 (  0.194967%) **
1e60 :     7166 (  0.089108%) *
1e70 :   244380 (  3.038839%) *******************************
1e80 :   206846 (  2.572107%) **************************
1e90 :  1080676 ( 13.438088%) ***************************************************************************************************************************************
1ea0 :   697310 (  8.670974%) ***************************************************************************************
1eb0 :   831270 ( 10.336752%) ********************************************************************************************************
1ec0 :   540327 (  6.718907%) ********************************************************************
1f20 :     1456 (  0.018105%) *
1f30 :     4576 (  0.056902%) *
1f40 :     5897 (  0.073329%) *
1f70 :     1872 (  0.023278%) *
1f80 :    27462 (  0.341487%) ****
1f90 :    26266 (  0.326615%) ****
1fa0 :    18290 (  0.227434%) ***
1fb0 :     2160 (  0.026859%) *
1fc0 :     2619 (  0.032567%) *
1fd0 :     2732 (  0.033972%) *
1fe0 :     1219 (  0.015158%) *
1ff0 :      641 (  0.007971%) *
2000 :      164 (  0.002039%) *
2010 :       81 (  0.001007%) *
2050 :       18 (  0.000224%) *
2060 :       46 (  0.000572%) *
2070 :       19 (  0.000236%) *
2080 :       26 (  0.000323%) *
2090 :        3 (  0.000037%) *
21c0 :    34938 (  0.434450%) *****
21d0 :    19426 (  0.241560%) ***
21e0 :       45 (  0.000560%) *
21f0 :       30 (  0.000373%) *
2200 :       21 (  0.000261%) *
2210 :       53 (  0.000659%) *
2220 :       84 (  0.001045%) *
2230 :       40 (  0.000497%) *
2240 :       21 (  0.000261%) *
2250 :      153 (  0.001903%) *
2260 :      173 (  0.002151%) *
2270 :       29 (  0.000361%) *
22b0 :       20 (  0.000249%) *
22c0 :       50 (  0.000622%) *
22d0 :       19 (  0.000236%) *
22e0 :       16 (  0.000199%) *
22f0 :        6 (  0.000075%) *
2300 :       27 (  0.000336%) *
2310 :       24 (  0.000298%) *
2320 :       24 (  0.000298%) *
2330 :       21 (  0.000261%) *
2340 :       24 (  0.000298%) *
2350 :       32 (  0.000398%) *
2360 :       28 (  0.000348%) *
2370 :        9 (  0.000112%) *
2c80 :     1211 (  0.015059%) *
2c90 :     4029 (  0.050100%) *
2ca0 :     1169 (  0.014536%) *
2cb0 :     1118 (  0.013902%) *
2cc0 :     2379 (  0.029583%) *
2cd0 :      240 (  0.002984%) *
2ce0 :      220 (  0.002736%) *
2cf0 :     2023 (  0.025156%) *
2d20 :     1903 (  0.023664%) *
2d50 :     2422 (  0.030117%) *
2d60 :     2420 (  0.030092%) *
2d70 :     1573 (  0.019560%) *
2d80 :     1211 (  0.015059%) *
2d90 :     2076 (  0.025815%) *
2da0 :     1923 (  0.023912%) *
2db0 :     3222 (  0.040065%) *
2dc0 :     1478 (  0.018379%) *
2dd0 :       21 (  0.000261%) *
2de0 :       24 (  0.000298%) *
2df0 :       21 (  0.000261%) *
2e00 :       23 (  0.000286%) *
2e10 :       22 (  0.000274%) *
2e20 :       19 (  0.000236%) *
2e30 :       15 (  0.000187%) *
2e70 :      870 (  0.010818%) *
2e80 :     1092 (  0.013579%) *
2e90 :      429 (  0.005335%) *
2ea0 :      464 (  0.005770%) *
2eb0 :      191 (  0.002375%) *
2ec0 :      560 (  0.006964%) *
2ed0 :      552 (  0.006864%) *
2ee0 :      504 (  0.006267%) *
2ef0 :      419 (  0.005210%) *
2f00 :      628 (  0.007809%) *
2f10 :      960 (  0.011937%) *
2f20 :      840 (  0.010445%) *
2f30 :      604 (  0.007511%) *
2f60 :      532 (  0.006615%) *
2f90 :      392 (  0.004874%) *
2fa0 :     6471 (  0.080466%) *
2fb0 :     8310 (  0.103334%) **
2fc0 :    10696 (  0.133004%) **
2fd0 :     4208 (  0.052326%) *
2fe0 :       78 (  0.000970%) *
2ff0 :       78 (  0.000970%) *
3000 :       60 (  0.000746%) *
3010 :       75 (  0.000933%) *
3020 :       90 (  0.001119%) *
3030 :       66 (  0.000821%) *
3040 :       69 (  0.000858%) *
3050 :       84 (  0.001045%) *
3060 :       41 (  0.000510%) *
3070 :     3073 (  0.038212%) *
3080 :     1872 (  0.023278%) *
3090 :     1560 (  0.019398%) *
30a0 :     1638 (  0.020368%) *
30b0 :     1560 (  0.019398%) *
30c0 :     1509 (  0.018764%) *
30d0 :      449 (  0.005583%) *
30e0 :       44 (  0.000547%) *
30f0 :       56 (  0.000696%) *
3100 :       48 (  0.000597%) *
3110 :       44 (  0.000547%) *
3120 :       54 (  0.000671%) *
3130 :     1269 (  0.015780%) *
3140 :     1295 (  0.016103%) *
3150 :     1037 (  0.012895%) *
3190 :     1159 (  0.014412%) *
31a0 :     1403 (  0.017446%) *
31b0 :     2650 (  0.032952%) *
31d0 :     1918 (  0.023850%) *
31f0 :     1918 (  0.023850%) *
3200 :     1141 (  0.014188%) *
3210 :      803 (  0.009985%) *
3220 :      373 (  0.004638%) *
3230 :      400 (  0.004974%) *
3240 :      756 (  0.009401%) *
3250 :      926 (  0.011515%) *
3260 :      781 (  0.009712%) *
3270 :      228 (  0.002835%) *
3280 :     3252 (  0.040438%) *
3290 :       36 (  0.000448%) *
32a0 :       40 (  0.000497%) *
32b0 :       44 (  0.000547%) *
32c0 :       74 (  0.000920%) *
32d0 :      124 (  0.001542%) *
32e0 :      134 (  0.001666%) *
32f0 :      156 (  0.001940%) *
3300 :     2218 (  0.027581%) *
3310 :     2131 (  0.026499%) *
3320 :     2076 (  0.025815%) *
3330 :        6 (  0.000075%) *
49c0 :   413010 (  5.135734%) ****************************************************
49d0 :   660816 (  8.217175%) ***********************************************************************************
49e0 :   660816 (  8.217175%) ***********************************************************************************
49f0 :   413010 (  5.135734%) ****************************************************
     :  8041888 (100.000000%)

(probably we are a bit off-topic for the AGD thread)

[BigEd]

(ooh, that looks relatively digestible to me! Will be interesting to see what Kees thinks.)

[oss003]

This confirms my suspicion, it is the sprite routine.
The routine at #1E8E displays 2 bytes of the sprite and is called 32 times per sprite.
To swap the old- and new spriteaddress, the routine #49C6 is also called 32 times per sprite.
So these are indeed time consuming routines.

Nice job Dave ....

Greetings
Kees

[oss003]

And here's a complete game this time: Mike the Guitar written by Sebastian Braunert and Uwen Geiken:

mike.png (7.36 KiB) Viewed 9590 times

https://www.youtube.com/watch?v=9ErOVp3itLI

Greetings
Kees

[oss003]

And some more progress .......... (looking familiar John .......... )
Does however need some work ....

https://youtu.be/SWbh4SkU4o0

Greetings
Kees

[sbadger]

And some more progress .......... (looking familiar John .......... )
Does however need some work ....

https://youtu.be/SWbh4SkU4o0

Greetings
Kees

oh my, that's fantastic, One of deathmonks best...I was thinking about this on a beeb, but this is pretty close.

[hoglet]

Hi Kees,

Again, some very nice work here.

I'd like to try to build the "Foggy" demo (again as a good test of the profiler I'm working on).

Are the files for this checked into your github?

Also, I've spotted a bug in the script compiler that causes a core dump on Linux. In main() you have:

Code: Select all

	char *cChar;

and then later:

Code: Select all

	lSize = fread( cChar, 1, 1, pEngine );			/* read first character of engine source. */

That's wrong, because cChar is not initialized to point to anything. So the read ends up writing to location 0.

A fix would be to change the definition to:

Code: Select all

	char cbuf;
	char *cChar = &cbuf;

This fixes the core dump for me.

Dave

[oss003]

Ok, I've updated Github and added your update ...thanks.
The compiler has some more updates like adding a Font is nUseFont=1 and a Createmessage problem is solved with multiline messages.

Foggy is too big to fit in $0300-$07fff so I removed the font.
This means that FONTFOG.ATM has to be loaded at $9800 before Foggy is executed.
It's still buggy but the titlesscreen is ok and the first screen is playable.

Greetings
Kees

[oss003]

Another update ...... a part of the partical engine is working now ... trailing

https://www.youtube.com/watch?v=M42u5ewiwbE

Greetings
Kees

[oss003]

Stars in the particle engine are also working now:

https://youtu.be/ngzxr0XFF_A

Greetings
Kees