Soundstorm 2 in PS3?

[loekf2]

Could Sony just use their existing sound chips from HiFi and AV products?

No, of course not. The DSP and/or ASIC used in their MD/MP3 range of players isn't equipped to handle the hundreds of voices we'd expect of a next-gen sound processing system. Heck, they couldn't handle the dozens of voices in the games hardware of today!

That stuff is low-power, low-performance crap that only needs to be able to do a few specific tasks, the major being decompress and play audio, then read the buttons and update the display of the device. That's it.

A small auxiliary cell processor for I/O and sound is an idea I've already presented in this forum btw, I think it would be a neat idea if it had like 1PU, 4 APUs and ran at perhaps 500-1000MHz or some such, but of course nobody listens to me... I still think it's more likely with a custom-designed, DSP-based solution rather than a cell device for the sound, that would be the traditional way of going about things rather than trying something new, unproven.

You confuse me a bit. What's the big deal of doing Dolby Digital encoding in software on a somewhat general purpose CPU (= Cell) instead of doing software (firmware) on a dedicated DSP ? It's both software. I don't think SoundStorm's encoding bit is done via hardware (hardwired). Must be some DSP in there. Same applies to the HRTF filters for the 3D part.

Wild guess is that 5.1 en/decode will take a relatively small chunk out of the total CPU load. We're talking multiple GHz's here (> 4 as seen in the presentation). The CPU has lots of DSP capabilities, so shouldn't be too expensive.

 

[Psikotiko]

We have reason to believe that the new Soundstorm will be delivered through the PS3 as a primary chip for audio - this could then migrate to an addin card on a PCI-E bus from NVIDIA. All of the rumours of the soundstorm team being disbanded may well be true, but the team has been rebuilt and they are working on it as we speak.

http://www.hexus.net/#Soundstorm2

 

[rabidrabbit]

Would a separate sound DSP be included to PS3 just to make programming the PS3 that much easier?
I mean if you'd use the "Cell" CPU as a sound processor, wouldn't that mean on more "thread" of program code to be juggled between all the parallel APU's PPU's DMA's memorys etc....... even if it wouldn't take much processing power, would it just make the programmer's life unnecessarily more complicated?

 

[London Geezer]

Would a separate sound DSP be included to PS3 just to make programming the PS3 that much easier?
I mean if you'd use the "Cell" CPU as a sound processor, wouldn't that mean on more "thread" of program code to be juggled between all the parallel APU's PPU's DMA's memorys etc....... even if it wouldn't take much processing power, would it just make the programmer's life unnecessarily more complicated?

I guess it wouldn't be more headachey than doing the same on PC games today. Other than the obvious How do we do this on Cell issue.

I'm pretty sure there will be a separate chip, if not Soundstorm solution (would be NICE), then a modified version of PS2's one, which was a modified version of PS1's one.
PS2 does DTS in game today with very little performance hit (4% or something according to Archie), and i don't expect new standards like DTS+ will add a lot to that. But hey who am i to talk...

 

[loekf2]

I'm pretty sure there will be a separate chip, if not Soundstorm solution (would be NICE), then a modified version of PS2's one, which was a modified version of PS1's one.
PS2 does DTS in game today with very little performance hit (4% or something according to Archie), and i don't expect new standards like DTS+ will add a lot to that. But hey who am i to talk...

True. The PS1 had a "SPU" doing 24-voices, reverb, delay I guess. DMA was included. I thought that the PS2's SPU was also external (talking to the I/O processor IC in the original 3-chip solution), but feature wise not much new.

Yamaha offered a SW wavetable synth running on the EE in SW, so more proof the SPU "2" was not that powerfull.