Frank Brickle wrote: > Many of the "Cognitive Radio" functions are likely to eat a lot of > cycles. This is particularly true for the ones that require > adaptive calculations, where there isn't a lot of offline work you > can do in advance, or those which need intricate probability > models with frequent updating. Similarly for blockwise algorithms > like EM. > I believe the difficulty in doing this with the Cell will be to schedule the 8 SPE's efficiently. Bayesian learning in EM will certainly fly at the (say) 50 GFlops we can expect to get upon optimization of the CDR code if we keep the SPE's flaming. I have no doubt in my mind that we can do blind learning about entire amateur bands and process them on the fly in the Cell on the PS3 with USB2 delivery of the data! The potential is just amazing now that it is here and upon us.
These multimedia engines, Cell, Nvidia CUDA GPU, and others will obsolete the standard Mobo processors for signal processing in the coming years technically speaking. I can't wait until I run DttSP, GnuRadio, HPSDR, "PowerSDR" in the idle cycles on my Nvidia Graphics card with the code written in Cg or doing all knew cognitive code running EM for blind learning and watching the entire world of radio change around us. The potential is huge if we don't run into serious pitfalls with these new toys. One of the most serious pitfalls as I see it is the built in stuff in the hardware or with the SDK's that seems to want to lock us out of doing open source code. On the Cell, I am quite worried about Hypervisor which seems to be HAL convolved with DRM on steroids, and Nvidia won't reveal a thing without signing an NDA. Of course this NDA means no code could be open source on the Nvidia GPU at this time. If we get beyond these kinds of issues successfully and soon we should be able to use a lot of the potential for these new toys. Bob > 73taxed > Frank > AB2KT > > -- Robert W. McGwier, Ph.D. Center for Communications Research 805 Bunn Drive Princeton, NJ 08540 (609)-924-4600 (sig required by employer) _______________________________________________ FlexRadio mailing list FlexRadio@flex-radio.biz http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz Archive Link: http://www.mail-archive.com/flexradio%40flex-radio.biz/ FlexRadio Homepage: http://www.flex-radio.com
