Hi Achilleas, sorry for the belated reply but I was very busy working on extending MA over SR-DVB and with other academic stuff here.
MA actually stands for "Memory Acceleration" and is intended as an optimization technique aimed at signal-processing-generated computation over GPPs and DSPs (we haven't explored it yet, but I'm persuaded that it can be useful for FPGA implementations too). Such optimization technique belongs to the broader class known in computer science as space/time trade-off and basically consists of an "algorithmic toolbox" which takes a classical implementation of a typical communication signal processing algorithm (say for example an OFDM time and frequency offset estimator or a Viterbi Decoder, just to quote two highly heterogeneous algorithms we've been trying) and returns a "memory-accelerated" version capable of running much faster over the same HW. Just as an example: since my last email upon this topic we applied MA to another (very small) section of SR-DVB chain and managed to further reduce the computational cost of the entire system previously shown in http://www.youtube.com/watch?v=S5nGBDCxhmk by an additional 2%. This appears to confirm our idea that considerable margins do exist for optimization as most of the chain is currently still implemented in a traditional way. I hope I have answered your question, even if with some delay. I will present some more details about this next week in Karlsruhe at WSR10. my best regards vincenzo On Mon, 2010-02-01 at 14:30 -0500, Achilleas Anastasopoulos wrote: > Vincenzo, > > this seems very impressive! > > May I ask what the initials NA stand for? > > best, > Achilleas > > > _______________________________________________ > Discuss-gnuradio mailing list > [email protected] > http://lists.gnu.org/mailman/listinfo/discuss-gnuradio _______________________________________________ Discuss-gnuradio mailing list [email protected] http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
