Originariamente inviato da
bibo01
Giusto per approfondire la mia proposta iniziale mi sono messo in contatto con il progettista in questione.
The sound was VERY good, the best I had heard from a computer at that time. I personally think this is a very good way to go.
I did a test several years ago where I took a stripped down puppy linux with realtime kernal and just ethernet and USB running (no X windows, no display) and netjack2 and ran this from a USB stick with a USB DAC on a whole bunch of diferent computers. (the USB stick is only accessed at boot time, the OS and all programs are copied into ram). The music data came over the ethernet from another computer also running netjack2 and a simple wav player. Netjack2 is a very simple protocol which just uses UDP so it takes very little processing on the "player" computer.
Netjack2 has some wonderful properties for this purpose (very small, very simple protocol and just uses UDP so very little processing effort required) but has one big issue, the sample rate cannot be changed on the fly. When you start jack on the hardware with the audio connected (the low power stripped down system) you give it a sample rate and that stays fixed for the duration of that jack instance.There is no easy way to fix this, jack was just not designed to do this.
Ha fatto questi test su parecchie macchine - FitPCslim (Geode LX800), VIA EPIA mini-itx board at600MHz, FitPC2, laptops di Lenovo, Fujitsu e Dell, e svariati desktops. Per apparecchi di bassa potenza, i risultati migliori (anche a livello assoluto) li ha ottenuti con CPU meno potenti; tra quelli medio potenti, i desktop erano meglio dei laptop.
D'altro canto, la soluzione offerta da Signalyst per il NAA è simile ma utilizza per la trasmissione ethernet il protocollo TCP. Si ha una maggiore sicurezza dei dati ma un dispendio di risorse maggiore.