Sheddy

thanks, hadn't realised that. better check some of my code as well!

Well I hope that's it. But I'm kinda surprised the original code has problems if there's only the one IRQ on the go. It shouldn't do an IRQ until STIMER has been written and then counted down? Given the few cycles in between the address setting, I would have thought it would be a pretty rare event for that to screw up.

another thing to check would be the code for determining whether the VBI has interrupted an IRQ in progress. bad things can happen if that IRQ is not allowed to finish before a CLI in the VBI allows the next IRQ to happen.

the game looks awesome, great work! how does the digital voice sound without anything else going on (no screen, no VBI/DLI's)? it should be possible to get it to sound almost as good as that state, although it may be very hard if your immediate (stage 1) VBI is large. (Rybags ideas sound good in that case) probably didn't explain very well in my message that you should only check for outstanding IRQ's and clear interrupt flag in VBI if you've disabled the OS VBI and are using your own (which are you doing?). Otherwise sticking the stuff in the deferred (stage 2) allows the IRQ to work again and already does that for you (just as Rybags has already noted) how many scanlines is your immediate (stage 1) VBI taking now Pete? have you moved all you can into deferred (stage 2) VBI (including RMT)? If RMT sounds too slow in stage 2 (IE VBI interrupted IRQ too often) here's some things to try: - try and make the immediate VBI as short as possible and take less time than the time between samples - you can try and time the IRQ's so they'll always "miss" hitting the VBI, so the sample rate is very important - if you don't miss the immediate VBI nearly every time, the deferred VBI won't get called enough - sync the start of the sample playback to a non-DLI scanline so you consistently miss the VBI/DLI's - try it without RMT to see if that is messing with the AUDCTL clocking on your channel (and screwing with your timer) I use 16-bit timers and 3.9KHz sample to miss the VBI and DLI's, but I don't think that should be necessary, although the 16-bit timer did improve quality a little (but that was more down to missing DLI's not VBI) keep up the good work

I've had some reasonable results using VirtualDub with the Frame Merger Plugin Filter, although it can sometimes look overly blurry

sox raygun-01.wav -r 4000 -2 -c 1 raygun48.wav -2 makes Sox output 2 byte sample - 16 bit -c 1 will make sure it is converted to mono if it isn't already

I haven't tried much, but I always thought they could be made to work. There are big problems to work through though. Unless you do something in a VBI to continue to play samples during vertical blank, you get a nasty sounding 50Hz/60Hz gap of no sample being played during vertical blank. Your display list would probably need to be longer than the recommended length too, because for just a 4KHz sample you need to output a sample every other VCOUNT (every 4 scanlines at 192 resolution). If you miss some samples, or play them at the wrong time, it soon becomes obvious to the ear. Maybe slower samples would work with DLI's though? NRV said he'd got some sample playing going with DLI's in the past.

nice! much easier than the GameBoy converter - no need to strip the headers and a lot more progs support 16-bit samples than 8-bit. you're right! I should have tried that first as it makes for a more straightforward example. here's the same code but using the OS. (Atari program is sampleOS.exe in the zip file). (remembered to put the Equates file in the zip as well this time) ATARI=0 .include "equates.asm" ; variables *=$f0 sample .ds 2 sample_nybble .ds 1 sample_value .ds 1 sample_size .ds 1 dli_colour .ds 1 *=$02e0 .byte <run,>run ; initial run address ; main code *=$2000 run lda #0 sta NMIEN ; turn off NMI's sei sta IRQEN ; disable IRQ's lda #<irq ; use immediate IRQ vector (no other OS IRQ's will happen) sta VIMIRQ ; 534 lda #>irq sta VIMIRQ+1 ; 535 lda #<dlist ; set up a display list sta SDLSTL ; 560 lda #>dlist sta SDLSTH ; 561 lda #<vbi ; set up an immediate VBI sta VVBLKI ; 546 lda #>vbi sta VVBLKI+1 ; 547 lda #<dli ; set up a DLI sta VDSLST ; 512 lda #>dli sta VDSLST+1 ; 513 lda #128+64 ; turn on NMI's - DLI's and VBI sta NMIEN ; 54286 cli ; enable IRQ's ; initialize sample play irq play jsr init ; initialize the sample to play wait1 lda TRIG0 ; any other stuff can go here, but... beq wait1 ; just wait until fire button is pressed wait2 lda TRIG0 ; bne wait2 beq play ; end of main code init lda #<sample_start sta sample lda #>sample_start sta sample+1 lda #[>sample_end->sample_start] sta sample_size ; sample size in 256 byte pages lda #15 ; 64KHz pokey clock divided by 16 = 4Khz timer sta AUDF1 ; in timer 1 lda #1 sta IRQEN ; enable timer 1 lda #0 sta sample_nybble ; initialize nybble ldx #$18 stx sample_value ; an initial sample value (with volume only bit) lda #115 init1 cmp VCOUNT bne init1 ; sync to a scanline for a precise start stx STIMER ; start timers (must be non-zero value) rts ; IRQ irq pha ; save a lda sample_value sta AUDC1 ; play sample ASAP to minimise DMA lag tya pha ; save y ldy #0 lda #1 sty IRQEN ; reset interrupt sta IRQEN ; re-enable only timer 1 eor sample_nybble ; switch between 0 and 1 (right and left nybble) sta sample_nybble beq irq1 lda (sample),y lsr a lsr a lsr a lsr a ; left nybble of sample bpl irq2 ; always branch irq1 lda (sample),y and #$0f ; right nybble of sample inc sample ; next lo byte of sample bne irq2 inc sample+1 ; next hi byte of sample dec sample_size ; check if at end of sample bne irq2 ; branch if not tya sta IRQEN ; end of sample - reset interrupt beq irq3 ; always branch irq2 ora #$10 ; turn on volume only bit irq3 sta sample_value ; save sample to play next irq pla tay pla ; restore y and a rti ; DLI dli pha lda dli_colour clc adc #2 sta dli_colour ; just rainbow colours down screen sta WSYNC sta COLBK pla rti ; VBI vbi lda #0 sta dli_colour ; reset dli colour jmp SYSVBV ; 58463 - do system vertical blank processes ; display list *=$2400 dlist .byte $70,$70,$70 .byte $8D+$40,<screen_start,>screen_start .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D .byte $41,<dlist,>dlist ; sample data *=$2800 ; NB start on a 256 byte page boundary ; sample will stop playing at nearest page boundary if not a multiple of 256 bytes sample_start .incbin "raygun44.raw" ; 4KHz, 4-bit raw sample data sample_end ; screen data *=$8000 screen_start NB if the DLI's cause sound quality problems you can "waste" another pokey voice to make a 16-bit timer running at 1.79MHz timer to more precisely "miss" the DLI's. [Edit] Unfortunately IRQ's don't seem to be much good if you want to use a lot of DLI's and 192 vertical resolution. Try changing the code to use Antic E instead of Antic D and the DLI's just mess up the timing of the sample playing and it sounds awful. [Another Edit!] Of course it will, because there are no "gaps" for the IRQ to fit into if there are DLI's every scanline. As long as the DLI's are spaced out and not too bunched together, it should be fine. samplesOS.zip

the only one I can think of is Dynakillers, but I don't know if it is using IRQ's. Brilliantly fun game BTW! Of course you do see it in some demos, although they usually don't have DLI's as well.

I've found the following free PC tools helpful to convert to low speed 4-bit samples. I don't know of anything that can do it all in one go though. SoX (Sound eXchange) This can do most anything with PC sound files! AFAIK PC's never used 4-bit sound samples so it doesn't do that. But it can do everything up to that last step of making into 4-bit. EG from the attached example, I created an 4KHz 8-bit PCM WAV without a header like this: c:\>sox raygun-01.wav -r 4000 -1 raygun48.raw now it just needs converting to 4-bit. It is not too hard to write a program to do that, but it would seem that GameBoy developers have exactly the same problem, and "MegaMan_X" has created a program to do it, which you can get from here: MegaMan_X's GameBoy Wave Converter It's not very sophisticated, but it does the job. NB only 8 character filenames can be used. The input file must be 8-bit mono raw (no header). EG in the attached example I created the 4-bit version of the sound like this: c:\>snd2gbw raygun48.raw raygun44.raw here's some example code that will use an IRQ to play a sample without the OS, showing some DLI's on a blank Antic D screen (ATasm assembler). sample44.exe in the attached zip is the assembled Atari program. [Edit: changed trigger button code to fix buzzing] ATARI=0 .include "equates.asm" ; variables *=$f0 sample .ds 2 sample_nybble .ds 1 sample_value .ds 1 sample_size .ds 1 dli_colour .ds 1 *=$02e0 .byte <run,>run ; initial run address ; main code *=$2000 run lda #0 sta NMIEN ; turn off NMI's sei sta IRQEN ; disable IRQ's lda #128+32+16+2 ; bit 7 on - ram at $5000, bits 4&5 on - 130XE compatible sta PORTB ; bit 1 on - disable os & rom, bit 0 off - disable basic lda #<nmi ; setup custom NMI handler sta $fffa lda #>nmi sta $fffa+1 lda #<irq ; setup custom IRQ handler sta $fffe lda #>irq sta $fffe+1 lda #<dlist ; set up a display list sta DLISTL lda #>dlist sta DLISTH lda #2+32 ; enable normal width screen+screen dma sta DMACTL lda #128+64 sta NMIEN ; turn on NMI's - DLI's and VBI cli ; enable IRQ's ; initialize sample play irq play jsr init ; initialize the sample to play wait1 lda TRIG0 ; any other stuff can go here, but... beq wait1 ; just wait until fire button is pressed wait2 lda TRIG0 bne wait2 beq play ; end of main code init lda #<sample_start sta sample lda #>sample_start sta sample+1 lda #[>sample_end->sample_start] sta sample_size ; sample size in 256 byte pages lda #15 ; 64KHz pokey clock divided by 16 = 4Khz timer sta AUDF1 ; in timer 1 lda #1 sta IRQEN ; enable timer 1 lda #0 sta sample_nybble ; initialize nybble ldx #$18 stx sample_value ; an initial sample value (with volume only bit) lda #115 is cmp VCOUNT bne is ; sync to a scanline for a precise start stx STIMER ; start timers (must be non-zero value) rts ; IRQ irq pha ; save a lda sample_value sta AUDC1 ; play sample ASAP to minimise DMA lag tya pha ; save y ldy #0 lda #1 sty IRQEN ; reset interrupt sta IRQEN ; re-enable only timer 1 eor sample_nybble ; switch between 0 and 1 (right and left nybble) sta sample_nybble beq irq1 lda (sample),y lsr a lsr a lsr a lsr a ; left nybble of sample bpl irq2 ; always branch irq1 lda (sample),y and #$0f ; right nybble of sample inc sample ; next lo byte of sample bne irq2 inc sample+1 ; next hi byte of sample dec sample_size ; check if at end of sample bne irq2 ; branch if not tya sta IRQEN ; end of sample - reset interrupt beq irq3 ; always branch irq2 ora #$10 ; turn on volume only bit irq3 sta sample_value ; save sample to play next irq pla tay pla ; restore y and a rti ; NMI nmi pha bit NMIST ; check for cause of interrupt bvs vbi ; branch if a VBI ; DLI dli lda dli_colour clc adc #2 sta dli_colour ; just rainbow colours down screen sta WSYNC sta COLBK pla rti ; VBI ; critical vbi vbi sta NMIRES ; reset interrupt txa pha ; save x lda #0 sta dli_colour ; reset dli colour tsx lda $104,x and #$04 bne vbiexit ; exit VBI if VBI interrupted an IRQ cli ; allow IRQ to interrupt from now on ; deferred VBI here vbiexit pla tax pla ; restore x and a rti ; display list *=$2400 dlist .byte $70,$70,$70 .byte $8D+$40,<screen_start,>screen_start .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D,$8D .byte $8D,$8D,$8D,$8D,$8D .byte $41,<dlist,>dlist ; sample data *=$2800 ; NB start on a 256 byte page boundary ; sample will stop playing at nearest page boundary if not a multiple of 256 bytes sample_start .incbin "raygun44.raw" ; 4KHz, 4-bit raw sample data sample_end ; screen data *=$8000 screen_start samples.zip

I wouldn't look at the early demo code, because as you rightly say, there are no timers used. the samples in there are good for ripping of course, but the code is messy and limited to drawing PM graphics and sample playing in a loop. It can't do anything else at the same time which is why I moved on to IRQ's. try the example code which sets up an IRQ first! just think of them as DLI's and you won't go far wrong - except you must reset the timer to stop it being called again as soon as you return, and also set it again to run more than once.

Atari800 based emulators have supported MaxFlash for a couple of years now Not the flashing of the cartridge yet, but only a few things take advantage of that. I don't intend to sell anything (for various reasons), and anything other than a cart would need too much RAM, so MaxFlash seemed the best bet, as people can put it on a MaxFlash if they want to and use it on a normal XL/XE machine My current estimate is that it will use at least 3/4 of the 8Mbit!

some of the later levels will be 20fps, but it struggles to do even a few sprites at 25fps. would like to do all the 18 levels of course! yes, the sync is important with the IRQ's, as is the sample rate. I always thought DLI's might be good for samples too, although I never tried it!

you're missing out on all the fun if you don't do it yourself you know OK, this is an example that sets up timer 4 without the OS to run at 4KHz and is a good starting point to show basic sample playback ; disable IRQ sei ; point the irq vector to your routine that plays back 1 sample lda #<irq_play_sample sta 65534 lda #>irq_play_sample sta 65535 ; we'll assume Pokey clock is set to normal 64KHz and set timer 4 to count to 16 (4KHz sample) lda #15 sta AUDF4 ; enable timer 4 lda #4 sta IRQEN ; enable IRQ cli ; start timers with a non-zero value sta STIMER NB apart from playing the next sample, your "irq_play_sample" routine should put 0 and then 4 into IRQEN to reset and re-enable IRQ 4 until all the samples are played. hope that helps

I forgot to say - thanks for the kind comments everyone! I will hang on to them, as sometimes I can only see the flaws and limitations in the game now, as I've been too close to it for so long.

It will be an 8Mbit MaxFlash image. Anything other than a cart would require a massive RAM expansion, so I'm not looking at doing any other formats.

It might work for some, but I think the engine would be too slow for OutRun if the road is done properly - I couldn't draw the proper Space Harrier grid on the floor quick enough on the Atari for my liking, and that is a lot simpler than OutRun's road. So I think the character based approach being taken with the current OutRun conversion is probably quite sensible.

Yes, that's how it's being done now, so this is what you should experiment with, Heaven. Although I did find using 2 voices for a more accurate timer helped with unwanted sample noise - that way IRQ's can be timed to miss the critical VBI and DLIS exactly, which otherwise will screw with the timing. Oh, and another tip is not to use the OS IRQ vectors! They seem to add too much delay in determining the cause of the interrupt and again seem to add unwanted noise.

OK! I've posted an update on the site www.sheddyshack.co.uk and added some movies.

good idea - it's not a major problem, but there's no harm in asking him nicely! (although I can see there is already a lot of optimizing in there already if different features are not used)

you're absolutely right about wasted CPU time if there's no 2nd Pokey. That bothered me a lot, but I found that just patching a few of the loops in the stereo RMT player when there's only a single pokey makes it work at almost the same speed as the mono player. (a slight rewrite of the setpokey routine was also needed to easily patch writing 2nd pokey registers or not) Sal (kjmann) has done the songs as Stereo but they also were designed to sound good if there is only 1 pokey, so no need for different versions combined with the patching. There'll only be one PAL/NTSC version but they'll be pretty much identical due to some compensations I don't fully understand RMT runtime routines, but there were no obvious optimizations when I looked through it, apart from maybe unrolling some loops, which wouldn't make that much difference speed wise, but make it much bigger. Of course assembling with only required FEAT does help a little.

wow, what are you optimizing now!? .. you need to start writing tutorials of this stuff (like donlebeau) NRV only that 256 byte wide screen I got rather excited about when discussing your great soft sprites demo posted here last year. I had only just discovered the benefits it would have for space harrier back then, but it took till nearly the end of the year before tools and game code were rewritten completely. (Only about 4-5% frame rate improvement though, which RMT more than wipes out!)

Wish I had found this game back in the day - it's fun and really clever (tough though!) - fascinating info about how it works - thanks for sharing, Don. @happymonster Been familiar with Storm for a while, as you know, but hadn't realised 'till now it was inspired by Gauntletak. Sorry to hear it didn't work out as shareware - it's a really decent game.

Space Harrier is still alive. No updates on the site though as its pretty minimal news - sorry to disappoint, but (another) major graphics engine rewrite to try and minimize slow down with Stereo RMT music has taken a long while, and I haven't added any extra stages beyond stage 4 yet I don't intend to do that many (news) updates, but now the rewrite is done there should be more visible progress to share over the coming months. It will take as long as it takes, but I will lose my mind if I don't finish this soon - I just realised I've been 8 years doing this multi-colour version now!

I can confirm chained display lists work correctly in Atari800WinPlus 4 at least