playsoft

I don't know whether it was this version or the VCS one as they play about the same, but back in the day I used to be unbeatable at this (well... against the computer, my brother and my uncle anyway). This was one of the first A8 games and while the graphics look a little crude you have to admire that it only uses half the cartridge space - it fits in 4K. Also that each basketballer is a single player graphic and coloured by tracking their positions on screen.

Thank you but I must say that it does not take much to do these 16K conversions (i.e. games which run on a 16K A8). They are still 99% the work of the original A8 author, my 1% is just making the 5200 look like it's an A8.

Yes I'd added the offset and compared to what I was doing before I'm saving on average 2.5 cycles on those scan lines which is fantastic thank you.

Thank you so much, I did not know about this... and it's not often you get cycles for free!

The keypad and joystick reading code you see in phobos.s is pretty much what I do each time. For example, with the basketball conversion I had to read all 4 joysticks and then combine them into fake PORTA,B registers - but the actual reading was done in exactly the same way. Regarding the keypad, in the first couple of conversions I used an IRQ interrupt but I subsequently switched to key held checks in the vertical blank (I had to do this in ATARI BLAST! because I'm using a Pokey timer for the sprites). The other bits of operating system and hardware I do as required by the program. For example, Savage Pond was using the o/s count down timers. It is the controllers which are most likely not to be emulated correctly but if you end up doing a conversion before you have real h/w you would have no shortage of testers here. The first release of GTIABlast! (as it was then) on the 5200 was developed completely blind as I did not have any h/w at the time and the Ultimate SD is not supported by any of the emulators.

In case this is of any interest, here is the build directory for this conversion. I use ca65 and have the bin sub-directory in my path. The .001-.006 source code files were generated from the A8 executable using the dis6502 interactive disassembler (downloaded from the atarimax website). For these 16K games I do not worry too much about dotting the i's and crossing the t's - I just try to make sure that I have identified the code and data sections correctly. With the code sections identified, most of the time it is enough to modify the h/w register equates (I pull in atari5200.inc) although this can miss some h/w locations if accesses have been done through zero page. I put hooks into the code to cater for the A8/5200 differences without disturbing the footprint of the code which is what lets me get away without doing a 100% disassemble as everything remains at its original location. phobos.zip

To fully complete the Caverns saga... Start = begin game Pause = pause game Reset = select level There is a cheat in the A8 original where shift + control + tab takes you to the next sub-level (A..P). I have remapped this to the * key. phobos.bin

I've managed to get through to the frog level by concentrating on getting the dropped eggs. Should you miss one then you really need the next worm as that makes you invincible for a little while and you can pick up the egg. By the way if anyone else makes it this far you control the frog (or rather its tongue) with the * key.

I was not familiar with Lady Bug but that looks like where the revolving doors may have come from, although Drelbs does something different with them. As one of my favourite games it is a shame more is not known about it. I think it's a very original game (despite the Lady Bug doors!) with a unique atmosphere and excellently programmed.

I do not use xasm but the manual indicates it is possible using the opt directive: OPT - set assembler options Five options are available: F - fill the space between memory areas with $FF G - Atari 5200 mode for hardware register abbreviations H - generate Atari executable headers L - write to the listing O - write to the object file You can turn any of these on or off. The default (if no OPT specified) is opt f-g-h+l+o+. Examples: opt l- listing off opt l+o- listing on, object file off opt f+g+h- useful for Atari 5200 cartridges - raw output, 5200 hw regs

One of the very first A8 games. Instructions as per the original (below) except that it is the port 1 controller start to begin and reset to select the number of players. http://www.atarimania.com/game-atari-400-800-xl-xe-basketball_529.html I only have 2 controllers but I am told 4 work ok. Interesting backwards shooting quirk under Altirra but it does this with the original A8 version too (which goes away if I set port 2 input to none). cxl4004.bin

Thank you both, I have added the fixed version back to my original post.

For those having problems, is this any better? I do not have any issues running the original from my USD - but random characters may suggest it is not initialising it's variables and relying on RAM being cleared. Perhaps the USD or the particular version of firmware in mine clears out RAM when launching a cart where others don't. Anyway, in this version RAM is cleared during initialisation. ant.bin

I only have a USD to test with and have not had any issues. I am sorry if it has wiped your cart clean and I will remove the file in case it causes problems for anyone else. Paul

Sorry I don't know. With these simple conversions (A8 games which run in 16K) I do not have to understand what the code does. All I do is disassemble the program making sure I have identified all the code and data sections, change the h/w register accesses and patch in the changes which are necessary for the 5200. So this version is still 99.9% the same code as the original A8 version.

It should be single player by default - but you can toggle between the number of players with *

I did the conversion for Aking and I had missed the cup/bomb - I'd not played the game before and wasn't sure what they were supposed to do. The reason they didn't work was that the collision detection register was read through a zero page location and was missed when modifying the GTIA addresses. The attached version has this fixed - if there is anything else wrong please let me know. oil_fixed.bin

There are instructions at http://www.atarimania.com/game-atari-400-800-xl-xe-drelbs_1737.html

Well it is somewhere around 450K at the moment although there is some wasted space in there. It will require a MaxFlash 8Mbit cart for the 8-bit and a Ultimate SD cart on the 5200 where it runs as a 512K hybrid image. There are 4 tracks - the Xevious jingle, two from Gradius and one from another game (I forget). I picked the midi files up from The Midi Shrine website which I converted to a format suitable for the 8-bit. The plan was for the music to be a homage to other shooters but now I hope to have some original music instead.

Is any of the history behind Drelbs known? It was published by Synapse but was Kelly Jones employed by them to write it or had he already developed the program and they simply published it?

I have tested on Altirra and on real h/w using the Ultimate SD cart, I've not tried kat5200 but here is a version with the cartridge header if it helps. drelbs.zip

Drelbs is one of my favourite 8-bit games but I couldn't find a 5200 version. This is a port of what I think must have been the cassette version as it ran in 16K. There was no title screen so I've added one with a speedy rendition of Wilder Reiter. I've changed nothing in the game itself other than to remap the keyboard controls: START - restart game (at any time) # - press this for difficulty selection, then press * repeatedly to select difficulty, then press START PAUSE - pause game, press RESET or 0..9 to resume drelbs.bin

Thanks Steve. The original intent was for it to be a single load .xex file, GTIA 9 only with a couple of levels. Exactly where it will end up I don't know as it seems to have taken on a life of its own. As long as it continues to be fun to work on is the main thing - I'm not going to beat myself up saying it must do this or that - it has already exceeded what I was expecting to do.

Steve kindly provided me with details of the hybrid 8K write/8K read mode supported by the Atarimax 5200 Ultimate SD cart. I have been using this mode primarily for the 8K read bank but now that I've experimented with the 8K write bank I thought I'd pass along the details in case they are useful. To use hybrid mode the image must be 512K in length with a ".hyb" extension. The cartridge is mapped: $4000..$7FFF banked area $8000..$BFFF fixed read area, top 16K of image The banked area can be either a 16K read bank or a 8K write/8K read bank. The cartridge has 512K of onboard SRAM and when you select an image it is copied into SRAM. The 8K write/8K read mode allows your 5200 program to write to the onboard SRAM. You select the bank by touching $BFC0..$BFFD where $BFC0 selects the bottom 8K of the image, $BFC1 selects the next 8K etc... When a 8K write/8K read bank is selected the following mapping applies in the banked area: $4000..$5FFF provides write only access $6000..$7FFF provides read only (and execute) access There is a significant restriction when writing, in that the way the cartridge write latching works means that 6502 dummy cycles can cause spurious writes elsewhere in the bank. For example, I find that sta (zp),y works ok when y is 0, but not when other values of y are used - I see additional spurious writes elsewhere in the bank. The attached demo program has 10 high resolution images stored in the first 10 banks accessible via the 0..9 keys. You can move the cursor around with the joystick and draw over them in white by pressing fire, or black with alternate fire. Start will fast cycle through all the images, any key will stop it. Rather than being anything useful (or particularly enjoyable) this is just to demonstrate use of the cart SRAM. The attached source code is ca65 - the bin directory is in my path and I use build.bat to assemble. demo.zip source.zip

1) I doubt the static display will remain in it's current form. I am all for not giving the player any respite at all and just go straight into the next level. It takes a few frames to set up the next level and without any intermission at all it looks like a glitch - ditto if the static display is not held long enough. The first 5200 demo held the static display for half a second or so (nothing written in it) and that seemed about right. My inclination is to keep the static display short and perhaps insert a code or message blipvert style. A code could be used in-game perhaps, a message would be something funny (hopefully). 2) & 3) are more Harvey's area but it is difficult to maintain a good contrast between the two. There were some very nice ANTIC 2 optical illusion designs which we canned because we just couldn't get sprite visibility anywhere near acceptable. 4) I suppose it should be possible to add something in the status area to indicate where the next attack will come from. Alternatively play it like squash and try and keep to the centre of the court. 5) Yes that's a possibility. At the moment I'm concentrating on putting the main building blocks together and once I'm happy I have all of the main ones there, then I'll start trying to turn it into something playable. The main thing with any spare time project is for it to be fun to work on. The plan is to work on it for the rest of the year and hopefully end up with something reasonably playable.