Overview Features Coding ApolloOS Performance Forum Downloads Products Order Contact

Welcome to the Apollo Forum

This forum is for people interested in the APOLLO CPU.
Please read the forum usage manual.
Please visit our Apollo-Discord Server for support.



All TopicsNewsPerformanceGamesDemosApolloVampireAROSWorkbenchATARIReleases
Documentation about the Vampire hardware

16/24bit Audio?page  1 2 

Tim D

Posts 9
13 Sep 2017 12:53


As I understand it correctly current Vampire audio developments are the following:
  - Route Amiga Native audio via Vampire over DIGITAL-VIDEO
  - Implement Paula in vampire for standalone V4
 
  I was just wondering whether we can expect an AHI compatible 16bit or maybe even 24bit "sound card" being part of the Vampire core in the future?


Tim D

Posts 9
13 Sep 2017 12:55


Oops sorry, this thread can be removed. Pamela is offering HiFi, my bad.


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
13 Sep 2017 12:58


PAMELA is PAULA compatible

PAMELA offers 8 AMIGA style DMA driven hardware channels.
PAMELA support 8bit and 16bit sample data.
PAMELA can use all memory 128 MB on Vampire2, 512 MB for Vampire4 for sample storage.

PAMELA has 16bit samples and 24bit precision calculation /mixing


Wawa T

Posts 695
13 Sep 2017 14:17


btw, mazze has included aminet paula ahi driver in aros source tree, but so far as binary. it is 68k asm, phx-ass syntax afair. it could be built within aros build system and with aros toolchain from asm source i guess, since its only amiga-m68k relevant anyway. im trying to look at that, but as its publicly known, im a very lousy coder if any. i think this driver could actually be extended with pamela features, so maybe some asm guru could help with it?


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
13 Sep 2017 14:47


wawa t wrote:

btw, mazze has included aminet paula ahi driver in aros source tree..

One could rethink the AHI driver of it.
The old PAULA had the limitation that it could only play from chipmem.
This means if a sample to play was given to AHI, and the sample data was in fastmem then the driver needed of course to copy the data to chipmem. This copy step is not needed anymore.

As PAMELA can play from all Vampire fastmem with DMA.


Szyk Cech

Posts 191
13 Sep 2017 17:22


Gunnar von Boehn wrote:

  As PAMELA can play from all Vampire fastmem with DMA.

Is this possible in theory to properly handle this feature by AHI driver?!?


Fernando Pereira

Posts 68
13 Sep 2017 17:34


Gunnar von Boehn wrote:

  PAMELA is PAULA compatible
 
  PAMELA offers 8 AMIGA style DMA driven hardware channels.
  PAMELA support 8bit and 16bit sample data.
  PAMELA can use all memory 128 MB on Vampire2, 512 MB for Vampire4 for sample storage.
 
  PAMELA has 16bit samples and 24bit precision calculation /mixing
 

 
  Out of pure curiosity, what stops Vampire from having 16 channels (voices)? 
 


Philippe Flype
(Apollo Team Member)
Posts 299
13 Sep 2017 18:13


@Fernando
 
  Clearly, hardware choices. The more channels the more space used in the FPGA, so it's just a matter of compromise, considering that 8 channels offers yet a nice step forward to plays amiga standard modules on 4 channels + sfx ambiance on the 4 others channels (in addition to a higher resolution quality). Also, CPU might be fast enough for software mixing on the Vampire, depending how much resources the game would need.


Philippe Flype
(Apollo Team Member)
Posts 299
13 Sep 2017 18:19


@Szyk

In theory, yes, even if it would be to just use one single channel, but in 16 bits and more than 48 KHz.


Fernando Pereira

Posts 68
14 Sep 2017 08:08


Philippe Flype wrote:

  @Fernando
     
      Clearly, hardware choices. The more channels the more space used in the FPGA, so it's just a matter of compromise, considering that 8 channels offers yet a nice step forward to plays amiga standard modules on 4 channels + sfx ambiance on the 4 others channels (in addition to a higher resolution quality). Also, CPU might be fast enough for software mixing on the Vampire, depending how much resources the game would need.
 

 
  I see. Having 8 voices is already freaking awesome and the stereo separation option is something I've always wanted. Thanks for the explanation. Perhaps in future, with better FPGA boards, we can have more voices. I say this because I've used to use Octamed v5 for music development and even 8 voices was somewhat limiting.


Ian Parsons

Posts 230
14 Sep 2017 09:54


Just as Octomed allowed more than 4 channels using the CPU to generate them I imagine the Vampire will be capable of a similar trick to boost the number above 8.


Kef Emzy

Posts 50
14 Sep 2017 11:24


Fernando Pereira wrote:

Philippe Flype wrote:

    @Fernando
     
      Clearly, hardware choices. The more channels the more space used in the FPGA, so it's just a matter of compromise, considering that 8 channels offers yet a nice step forward to plays amiga standard modules on 4 channels + sfx ambiance on the 4 others channels (in addition to a higher resolution quality). Also, CPU might be fast enough for software mixing on the Vampire, depending how much resources the game would need.
   

   
    I see. Having 8 voices is already freaking awesome and the stereo separation option is something I've always wanted. Thanks for the explanation. Perhaps in future, with better FPGA boards, we can have more voices. I say this because I've used to use Octamed v5 for music development and even 8 voices was somewhat limiting.

Indeed, 16 voices would have been awesome. Nevertheless, I'll take those 16-bit 8 voices any day over the old chipset sound... :)


Fernando Pereira

Posts 68
14 Sep 2017 11:47


Ian Parsons wrote:

  Just as Octomed allowed more than 4 channels using the CPU to generate them I imagine the Vampire will be capable of a similar trick to boost the number above 8.
 

 
  The problem was that, when Paula was split into 8 channels using Octamed v5 (AHI), voices would share the same volume setting per hardware channel. In other words, changing the volume of channel 1 would affect voices 1&2; changing the volume of channel 2 would change affect voices 3&4 and so on.
 
  I believe this won't happen with PAMELA... Will it? (makes a panicked looking face)
 


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
14 Sep 2017 12:21


Some of the key advantages of PAMELA are

a) 100% backward compatible
b) HIFI quality 24bit AUDIO mixing
c) support for both 8bit and 16bit samples
d) own volume and separate sample rate support per channel
e) fully DMA driven (zero CPU use) channels
f) support for 128 MB sample memory




Fernando Pereira

Posts 68
14 Sep 2017 14:05


Gunnar von Boehn wrote:

Some of the key advantages of PAMELA are
 
  a) 100% backward compatible
  b) HIFI quality 24bit AUDIO mixing
  c) support for both 8bit and 16bit samples
  d) -----> own volume and separate sample rate support per channel
  e) fully DMA driven (zero CPU use) channels
  f) support for 128 MB sample memory
 

Frak yeah! :D



Gunnar von Boehn
(Apollo Team Member)
Posts 6207
14 Sep 2017 14:30


Kef Emzy wrote:

Indeed, 16 voices would have been awesome. Nevertheless, I'll take those 16-bit 8 voices any day over the old chipset sound... :)

Well on AMIGA you had 4 voices.
4 Voices were nice for doing MODs.
4 voices were more than enough for doing game SND-FX.

4 Voices for doing both together - game music and game SND FX - was not that much.

So the advantage of having 8 voices is very clear.
You can have e.g allocate 5-6 voices for game music and 2-3 voices for SND-Fx.
You can easily see the improvement over A500 times.

So question is how much more will 16 voices give in this scenario?


Kef Emzy

Posts 50
14 Sep 2017 16:22


Gunnar von Boehn wrote:

Kef Emzy wrote:

  Indeed, 16 voices would have been awesome. Nevertheless, I'll take those 16-bit 8 voices any day over the old chipset sound... :)
 

 
  Well on AMIGA you had 4 voices.
  4 Voices were nice for doing MODs.
  4 voices were more than enough for doing game SND-FX.
 
  4 Voices for doing both together - game music and game SND FX - was not that much.
 
  So the advantage of having 8 voices is very clear.
  You can have e.g allocate 5-6 voices for game music and 2-3 voices for SND-Fx.
  You can easily see the improvement over A500 times.
 
  So question is how much more will 16 voices give in this scenario?

In a game probably not so much. But 16 voices would be awesome when making music. I used to connect two A500's (sync them through midi) to get 8 channels for music "production". Believe me, 16 channels would have made quite a difference. A HexaMed with 16 voices and 16 bits would still make my head rotate in the right direction.

Nevertheless, I'm ok with "only" 8 voices. I am very grateful for what you guys are doing and as soon as the standalone Vampire is ready for sale, you will have a happy customer!


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
14 Sep 2017 16:52


Kef Emzy wrote:

  In a game probably not so much. But 16 voices would be awesome when making music.
 

 
Well you can have "voices" and "voices"
You can with the CPU mix voices too.
So mixing 64 soft voices with the CPU - using AMMX should be piece of cake.
 
For music making I think this should work very good.
When music making, the CPU is idle and can do this easily.
 
For games - I think its nice to have "some" free hardware voices for in game music, which cost no CPU time.
 


Fernando Pereira

Posts 68
14 Sep 2017 16:54


Gunnar von Boehn wrote:

   
Kef Emzy wrote:

    Indeed, 16 voices would have been awesome. Nevertheless, I'll take those 16-bit 8 voices any day over the old chipset sound... :)
   

   
    Well on AMIGA you had 4 voices.
    4 Voices were nice for doing MODs.
    4 voices were more than enough for doing game SND-FX.
   
    4 Voices for doing both together - game music and game SND FX - was not that much.
   
    So the advantage of having 8 voices is very clear.
    You can have e.g allocate 5-6 voices for game music and 2-3 voices for SND-Fx.
    You can easily see the improvement over A500 times.
   
    So question is how much more will 16 voices give in this scenario?
   

   
    If you want to create stereo sound, 8 voices for both music and SFX is definitely not enough. Having the kick coming out only from the left speaker and the bass from the right speaker is not how you create a stereo environment (at least a good one). Yeah you can cross them over in a single channel but you also take away sound sustain from samples, since they cut each other out.
   
    For a stereo sound you need 2 channels per sound, 2 channels for the kick (left + right) and 2 another for bass (left + right) and we have already 4 channels in use if we do this. Sure 8 channels will allow twice as much as PAULA did but it's clearly not suffice to recreate a real stereo ambient without heavy limitation.
   
    With 16 voices you could develop 6 instrument music in stereo (12 voices) and still have 4 channel to play 2 SFX at the same time in stereo mode.

I'm not saying I don't appreciate the PALMELA, heck 8 channels is a freaking dream on the Amiga! I just want to point out that 8 channels is far from being the suitable spot.
   


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
14 Sep 2017 16:57


Fernando Pereira wrote:

If you want to create stereo sound, 8 voices for both music and SFX is definitely not enough. Having the kick coming out only from the left speaker and the bass from the right speaker is not how you create a stereo environment (at least a good one). 

Actually PAMELA does support in HW placing the channels "in Stereo" room.


posts 35page  1 2