keropi wrote:

ah, now I get it. it's a shame.... I thought that sid files where something like midi files.

Don't get disappointed yet. I've been told that there is C64 emulator (including emulation of SID and graphics) that works on 386 processor.
That been said... If we remove from such programm heavy parts for drawing graphics and emulating SID and just stream commands for SID into replica - we must not get heavier on CPU. I think it must be less CPU-hungry because we do not need to draw anything and simulate SID (we have real hardware for it!). So it is THEORETICALLY possible to play *.SID with the replica on 386 CPU.
But we have to find some open source for such emulator that we can reuse. We didn't find such efficient solution yet.

bristlehog wrote:

What is HVSID by the way?

I think it is HVSC. I've already downloaded it.

I used to frequent #c-64 on ircnet ... maybe they know where to point me for obtaining source codes for the hardsid programs/drivers (which could then be adapted).

Scali wrote:

Yes, I suppose the cards are not that expensive anyway, so you could just buy two cards, one with 6581 and one with 8580, and swap the cards, not the chips 😀

This is exactly what I did. 😎

Fagear wrote:

And I tested some HardSID-compatible software and of course it doesn't recognize the replica and doesn't play.

It won't play, because HardSID only uses 2 ports to operate (vs. 29 ports of SSI-2001) and uses them differently.

HardSID write operation:

1_outp(port, value);
2_outp(port + 1, register);

SSI-2001 write operation:

1_outp(port + register, value);

I'm not very sure, but perhaps with port trapping it would be possible to make HardSID compatible with SSI-2001 to some extent.

If all else fails, wouldn't it be an option to create "sid dumps" (being just basic timing info and sid data) from .sid files, so you'd just have to send data to the ports on the card, rather than emulating a whole c64 enviroment at the same time.

Certainly not very elegant, but that way no one would get excluded because of a slow cpu ... the SIDStick (which is basically a pocket sized SID player takes that approach if I'm not mistaken).
http://gadgetgangster.com/find-a-project/56?projectnum=236

bristlehog wrote:

Great Hierophant wrote:

so we use this card as a playback device for HVSID.

What is HVSID by the way?

I meant to write the HVSC, the High Voltage SID Collection, a huge collection (over 46,000) of SID tunes from games and composers.

Fortunately, a SID player does not need to emulate a whole C64. It does not need to emulate the video output of the VIC-II, a 1541 disk drive or the input features of the CIA. But it has to have cycle accurate 6502 instruction support and proper timing from the CIA and the VIC-II.

shock__ wrote:

If all else fails, wouldn't it be an option to create "sid dumps" (being just basic timing info and sid data) from .sid files, so you'd just have to send data to the ports on the card, rather than emulating a whole c64 enviroment at the same time.

That's one way, but I think it would be difficult to support things like looping (you'd dump the data as the song is playing, but you don't know anything about the song position).
One way I'm thinking of is to take the 6502-code from the SID and recompile it for x86. Shouldn't be that difficult. In its most simple form you would just disassemble the code, and replace every instruction with a C-function that performs the operation. Then you compile the C-code and let the compiler inline the function calls etc, so it will become reasonably efficient code (even a 386 is WAY faster than a 6502, so the code doesn't have to be that tight).

Great Hierophant wrote:

so it utilizes 6502 based instructions and relies on either the VIC or the CIA for timing.

Normally a SID just contains an entry point that needs to be called at a given interval. The actual timing is not done by the code in the SID file, but by the program using the SID music.

This is what it looks like in my code (using KickAss):

1.var music = LoadSid("Nightshift.sid")
2
3			// Init music
4			lda #music.startSong-1
5			jsr music.init
6			
7			..
8			
9			// At a fixed position on every frame, eg using a raster interrupt at a given scanline
10			jsr music.play
11			
12.pc=music.location "Music"
13.fill music.size, music.getData(i)

So my guess is that SID replayers would simply set up a static timer at 50 or 60 Hz and call the code at that interval. I don't think any CIA or VIC emulation is required at all for basic SIDs. Sample playback tricks will probably be an exception to that rule though. But even then, you'd probably just have to emulate the most basic timing stuff... CIA timer and raster interrupt.

I did a little thinking, couldnt have capacitance choosing for either chip be done with a simple dipswitch, or jumper?
I tihnk that would be more modular than fixing the card to one chip.

Stojke wrote:

I did a little thinking, couldnt have capacitance choosing for either chip be done with a simple dipswitch, or jumper?
I tihnk that would be more modular than fixing the card to one chip.

Like I said ... just add sockets, that way switching caps is as easy as switching jumpers ... except that you'd have to keep the right caps around.
Switching the supply voltage for the SID should also be hardwired in my opinion ... eliminates using an 8580 with 12 volts by "accident". Certainly could have been done with a jumper, but actually soldering on the card puts you in a more thoughtful mindset (usually including the schematic and/or manual) than just putting on a jumper or not.

Someone called nop from Belarus joined the SID player development. He's been of great help, and here's the video of Acidjazzed Evening over my P4-3066.

Scali wrote:

shock__ wrote:

One way I'm thinking of is to take the 6502-code from the SID and recompile it for x86. Shouldn't be that difficult. In its most simple form you would just disassemble the code, and replace every instruction with a C-function that performs the operation. Then you compile the C-code and let the compiler inline the function calls etc, so it will become reasonably efficient code (even a 386 is WAY faster than a 6502, so the code doesn't have to be that tight).

So my guess is that SID replayers would simply set up a static timer at 50 or 60 Hz and call the code at that interval. I don't think any CIA or VIC emulation is required at all for basic SIDs. Sample playback tricks will probably be an exception to that rule though. But even then, you'd probably just have to emulate the most basic timing stuff... CIA timer and raster interrupt.

I would think the PC's PIT would take care of most basic CIA timing duties. However, if a SID tune relied on 6502 cycle exact code, then converting that code to 8086 machine instructions would prove very problematic. Digital sampling with the SID, like the stuff from Impossible Mission "Another visitor, stay awhile, stay FOREVER!" may not work in that way.

Great Hierophant wrote:

I would think the PC's PIT would take care of most basic CIA timing duties. However, if a SID tune relied on 6502 cycle exact code, then converting that code to 8086 machine instructions would prove very problematic. Digital sampling with the SID, like the stuff from Impossible Mission "Another visitor, stay awhile, stay FOREVER!" may not work in that way.

Yup, basic timing for CIA or raster interrupts could be handled by the PIT I suppose.
I don't think many SID tunes would rely on cycle-exact code. If only because it prevents the CPU from doing other stuff, so it won't be very efficient. So it would only be used for standalone tunes, not for games/demos generally (much like most multispeed SIDs).
I think sample playback is usually done with CIA or raster interrupts ("play one sample every X scanlines"), so I would think you'd be okay with PIT-timing.

bristlehog wrote:

Someone called nop from Belarus joined the SID player development. He's been of great help, and here's the video of Acidjazzed Evening over my P4-3066.

Good news! 😎

Great Hierophant wrote:

SID tune relied on 6502 cycle exact code, then converting that code to 8086 machine instructions would prove very problematic. Digital sampling with the SID, like the stuff from Impossible Mission "Another visitor, stay awhile, stay FOREVER!" may not work in that way.

Well... I think if someone can capture direct commands to the SID, sampled by time, than anything else would not be necessary. Just stream of data going into SID's ports,

By the way, an hour ago I've sent another 5 boards to 4 Vogons users. 😀

Candytime? Candytime! (warning: 500kb jpeg)
http://i.imgur.com/DIBja0I.jpg
Shipping off to Fagear in the next days.

^^ Yay, nice one shock__! These SIDs look new with late '92 datecodes. And I do prefer the classic 'Commodore' logo on them too. RIP Commodore...

shock__ wrote:

Candytime? Candytime! (warning: 500kb jpeg)
http://i.imgur.com/DIBja0I.jpg
Shipping off to Fagear in the next days.

Are these from SLY Electronics?

Yup, they're from SLY. I even picked up some more (11 in total - 8 going to Fagear) - all working perfectly (after I figured out the original sockets in my C64 are the worst, cheap crap ever) - they still have a considerable amount left. So there shouldn't be a shortpass anytime soon.

The datecode 5092 also makes me assume it's one of the last batches ever made - since commodore declared bankruptcy in early 1994.

shock__ wrote:

Yup, they're from SLY. I even picked up some more (11 in total - 8 going to Fagear) - all working perfectly (after I figured out the original sockets in my C64 are the worst, cheap crap ever) - they still have a considerable amount left. So there shouldn't be a shortpass anytime soon.

The datecode 5092 also makes me assume it's one of the last batches ever made - since commodore declared bankruptcy in early 1994.

Cool!

I tried to buy from them, but they don't ship outside Germany. I'm sending you a PM, maybe you can help me! 😀

Thanks in advance...

I decided to make the post instead of a PM to Fagear 😀

I got the package today, EVERYTHING was excellent: packing, components, price, everything. Thanks for doing this Fagear!!!

the bagged components on top are some sockets and filtering caps that I got in order to make some tests. I'll start building the card tomorrow and update, I hope my 8580 SID from another source is not DOA 😁