strange way to set dma audio

[meynaf]

Originally Posted by leonard

But something is still confusing to me: if you said PT2.3F replay my stress-test mod correctly, it means player should also write sample address using non atomic write during the play, right? We know it's not "legit". Maybe there is no glitch in this specific case

There can not be a glitch in your stress-test mod, as the value is written long before AUDxDAT has a chance to expire.

[leonard]

Originally Posted by meynaf

There can not be a glitch in your stress-test mod, as the value is written long before AUDxDAT has a chance to expire.

I agree but my mod isn't a stress test, it's just a test to see if player supports or not the feature.
The code snippet I posted before really is a stress test, and produces audio glitch (at least in winuae)

My feeling is that you *could* have some .mods where the repeat ad is written at the exact bad time. I guess probability in normal music ( reasonable period, quite long sample etc) is quite low, but in theory existing players have this potential bug when dealing with "sample without a note"

[meynaf]

Originally Posted by leonard

My feeling is that you *could* have some .mods where the repeat ad is written at the exact bad time. I guess probability in normal music ( reasonable period, quite long sample etc) is quite low, but in theory existing players have this potential bug when dealing with "sample without a note"

That's very possible. IIRC i've seen Delitracker2's Paula-Noteplayer code waiting for some horizontal scanline position before touching some audio hardware regs. I wondered why, but it might be linked to this.

[Toni Wilen]

AUDxLEN is counted down 1-2 cycles after AUDxDAT DMA write so it always happens near the beginning of scan line.

[Thomas Richter]

Originally Posted by leonard

Hi,

The "official" safe way to start a new sampleA on voice A is:

1) switch off dma voice A ( move.w #$0001,$dff096 )
2) write sample address ( move.l #sampleA,$dff0a0 )
3) wait the right amount of time ( say X rasterlines )
4) enable the dma voice A ( move.w #$8001,$dff096 )

Actually, this is not "official". Busy waiting for a scan line is never necessary, and neither sufficient as you do not know what X is. You need to wait until Paula generates its interrupt, and then can write new data.


Thus, if audio DMA is currently off, write data and length into the registers, then write first two samples into the output register (as 16 bit), potentially set the period to 8 to lower the wait time for the next interrupt. This will initiate the audio state machine. Paula will generate an interrupt as soon as the first two samples are played (the same as those in your data).



Then wait for the interrupt to happen that you can write more data, which will happen as soon as Paula recognizes the write. At that point, you can enable the DMA. "Wait X Rasterlines" is not necessarily going to work, even if not knowing what "X" is going to be. Depending on the Paula release, X may depend on the current playback rate, and some ancient mod players failed exactly for this reason - not waiting "properly".



If audio DMA is already running, there is no need to stop it. Just wait for the next "audio done interrupt", which is an indication that Paula can take a new sample, and at this point, write the new length and address.



But there is also the audio.device to do that for you.

[Toni Wilen]

Waiting for interrupt after disabling DMA is not that simple.

Interrupt activates before channel state goes to idle. There is no safe way to do it without waiting (polling or timers or whatever) a bit. I think I explained this long time ago and AFAIK audio.device is not safe if CPU is "too fast". As in interrupt processing is really fast.

Also, there is no need to do any AUDxDAT writes. Most likely it was done that way to simplify driver (start and continue with new sample becomes identical). No one else does that except audio.device.

EDIT: Also you can't know if interrupt was due to sample ending or DMA stopping (Sample might end just after interrupt is cleared but before DMA is disabled). The shorter the sample the more unreliable it gets.

EDIT2: As I already said, "dma wait" requirement is Paula design flaw.

[Antiriad_UK]

Originally Posted by leonard

for instance this really cool attached .mod use a lot of "sample without note".

Did I?!

If I was doing something unexpected it was almost certainly an accident. I wrote the version of that tune in FL Studio and then did the amiga mod in a few hours after not creating a mod since 1990 so just winged it - I definitely wasn't trying to use weird features

Edit: And it was p61 I used in that intro and it sounded like I expected. I wrote the mod in milkytracker and never played it in Protracker on the amiga. (The mp3 of the fl studio master is in the intro download )

Edit2: Just opened the mod again, I think it must be the ostinato part you mean where the same riff goes on forever and various "similar" samples are used to retrig the previous note but with a different sample at a lower volume. Yeah that's chip tune 101 from the mods I looked at. I've not listened to this mod in ptplayer, but tbh it's so subtle I wouldn't notice in that mod because it's so fast. Maybe it _didn't _ play correctly in p61!!! (OpenMPT does as I just single stepped it and the sample changed)

[leonard]

Originally Posted by Antiriad_UK

Did I?!

Oh didn't even knew you composed this tune, congrats! I love it, feel both oldskool feeling and "melancholic" melody. So you didn't really used "sample without a note" on purpose? interesting

I don't remind what was the intro using this tune, could you post a link? ( need to listen to the original audio stream )

[ross]

Originally Posted by leonard

Oh didn't even knew you composed this tune, congrats! I love it, feel both oldskool feeling and "melancholic" melody. So you didn't really used "sample without a note" on purpose? interesting

I don't remind what was the intro using this tune, could you post a link? ( need to listen to the original audio stream )

Just check the message I sent you 2 days ago

[leonard]

Originally Posted by ross

Just check the message I sent you 2 days ago

oh damn I should always spot the private message counter at top right of the screeen!

[ross]

I LOVE this intro by Antiriad, so Amigaish and nostalgic.

[Antiriad_UK]

Originally Posted by leonard

Oh didn't even knew you composed this tune, congrats! I love it, feel both oldskool feeling and "melancholic" melody. So you didn't really used "sample without a note" on purpose? interesting

I don't remind what was the intro using this tune, could you post a link? ( need to listen to the original audio stream )

Thanks . The mp3 is quite fun. Here’s the link, it should be in the zip if I remember.

http://www.pouet.net/prod.php?which=81350

Yes I used the sample without a note on purpose. Pretty much every chip tune does something like that. I got it from looking at 4mats stuff in the 90s and assumed everyone did that .

[leonard]

Originally Posted by Antiriad_UK

Yes I used the sample without a note on purpose. Pretty much every chip tune does something like that. I got it from looking at 4mats stuff in the 90s and assumed everyone did that .

I tried to remove the "sample without a note" in pattern 7 in OpenMPT, and play pattern 6 ( with "sample without a note" ) chaining on pattern 7. I have the feeling I didn't spot a difference.

So I'm really confused now: I have a working modified LSP local version supporting this feature. Generic player is slightly slower, and slighty bigger ( about 540 bytes, not under 512 bytes anymore ). Insane mode is running at same speed.

So my question: Is it worth to support this feature in LSP?

what do you think everyone as a user?

[Antiriad_UK]

Originally Posted by Antiriad_UK

Thanks . The mp3 is quite fun. Here’s the link, it should be in the zip if I remember.

http://www.pouet.net/prod.php?which=81350

Yes I used the sample without a note on purpose. Pretty much every chip tune does something like that. I got it from looking at 4mats stuff in the 90s and assumed everyone did that .

Actually it just occurred to me. Doesn’t cream of the earth do this sample without a note retrigger? I need to look at the mod but from memory it was a single note retriggered on each line with an A01 effect as well. That’s as a popular mod as you can get

[a/b]

https://www.stef.be/bassoontracker/?...uleid%3D120017

[Antiriad_UK]

Originally Posted by a/b

https://www.stef.be/bassoontracker/?...uleid%3D120017

Yes but sample number and A0F How the hell did that mod stick in my mind for 35 years - I don't even remember what I had to eat last night haha

[leonard]

Originally Posted by a/b

https://www.stef.be/bassoontracker/?...uleid%3D120017

Is "cream of the earth.mod" broken if played with anything else than PT2.3F?

[leonard]

Originally Posted by leonard

Is "cream of the earth.mod" broken if played with anything else than PT2.3F?

I just figured out why "cream of the earth.mod" will play correctly with mod player that doesn't support "instrument without a note". Look at first pattern, voice C. These players (as LSP 1.03) won't retrigger a note even if the column is full of "instrument 10". Sample is not restarted BUT the voice volume is reset to instrument volume each row!
This module doesn't really showcase the "instrument without a note" bug of many amiga players.

[8bitbubsy]

Originally Posted by leonard

I tried to remove the "sample without a note" in pattern 7 in OpenMPT, and play pattern 6 ( with "sample without a note" ) chaining on pattern 7. I have the feeling I didn't spot a difference.

So I'm really confused now: I have a working modified LSP local version supporting this feature. Generic player is slightly slower, and slighty bigger ( about 540 bytes, not under 512 bytes anymore ). Insane mode is running at same speed.

So my question: Is it worth to support this feature in LSP?

what do you think everyone as a user?

Yes it is, "sample swapping" was a common technique used in plenty of ProTracker MODs, especially chiptune-styled ones.

[leonard]

Originally Posted by 8bitbubsy

Yes it is, "sample swapping" was a common technique used in plenty of ProTracker MODs, especially chiptune-styled ones.

so all these chiptune mods aren't played correctly with all existing amiga players but PT2.3F? Anyone can confirm that?