Without getting into a long discussion -- the situation isn't really that complex, nor is it worth spending more than a few further moments on -- the problem is all in the audio subsystem. Between PowerSDR, PortAudio, VAC, and realtime user monitoring requirements, the fact that anything approaching QSK is even possible is a testament to the programmers who've hammered on the Windows SDR code over time.
The problem isn't PowerSDR. It's the lack of integrated, rational support in Windows for the kinds of state transitions in the audio subsystem that a QSK application requires, especially when users demand that it also cooperate transparently with third-party applications over which PowerSDR has no control. That's really all I as an individual developer have to say on the subject. The current state of PowerSDR is a best-effort attempt to bring an existing application into line with a number of fundamentally incompatible operational requirements, given also the demand that the nest of applications run on a platform with ill-defined capabilities and competition for system resources, and with an uncertain roadmap for the dominant OS in the future. Speaking only for myself, I think this line of discussion has run its course of usefulness. Development effort is being focused on doing it right, not patching the current system, which is basically unfixable in a systematic and maintainable way. Batting numbers back and forth is amusing, but the problems aren't quantitative. The problems are all practical and empirical, and they're being addressed. Hic taceo. 73 Frank AB2KT On Sat, Apr 26, 2008 at 10:09 AM, Ed Russell <[EMAIL PROTECTED]> wrote: > Jim, > > When I'm not getting QSK and my CPU is at 8%, I'm wondering where the > bottleneck is. 250ms latency is a real long time in the cpu universe. > > 73 Ed W2RF > > On 26 Apr 2008 at 6:37, Jim Lux wrote: > > > Quoting Robert Dennison <[EMAIL PROTECTED]>, on Sat 26 Apr 2008 05:36:16 AM > PDT: > > > > > We now await the first release of the new SW architecture > > > implementation.. The application just screams for multiprocessor - > > > multi-threaded implementation and that's the new architecture. In so > > > many ways, FlexRadio is the solution who's time has arrived... > > > > indeed, a multi processor implementation might be nice, however, it > > should be necessary. > > > > The existing implementation IS multithreaded (and has been since the > > beginning) > > > > I suspect that there are some idiosyncracies in the implementation > > that have "bad interactions" with some of Windows. The existing > > PowerSDR uses the pthreads package to provide a POSIX threads > > interface to Windows's multithreading environment. There may be some > > issues there. > > > > After all, there are existence proofs for Windows software that has > > very good real time response characteristics, running on a single > > processor. Games are the most notable one, but there's also data > > acquisition software (e.g. LabView) that does quite well at a > > millisecond scale timing. I suspect, though, that the dollars/labor > > invested in making first person shooter games and LabView play nice in > > the Windows environment is substantial. > > > > bear in mind, too, that having the multiple cores available doesn't > > mean that you automatically get better performance. The OS has to > > support farming the work out to the cores in a useful way, and you're > > still faced with a bunch of other bandwidth limits (e.g. memory bus, > > i/o device access). This is true regardless of the OS. > > > > > > > > > > Jim, W6RMK > > > > > _______________________________________________ > FlexRadio Systems Mailing List > FlexRadio@flex-radio.biz > http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz > Archives: http://www.mail-archive.com/flexradio%40flex-radio.biz/ > Knowledge Base: http://kb.flex-radio.com/ Homepage: > http://www.flex-radio.com/ > > -- The only thing we have to fear is whatever comes along next. -- Austin Cline _______________________________________________ FlexRadio Systems Mailing List FlexRadio@flex-radio.biz http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz Archives: http://www.mail-archive.com/flexradio%40flex-radio.biz/ Knowledge Base: http://kb.flex-radio.com/ Homepage: http://www.flex-radio.com/
