Thanks, Its a nicely written document. Answers all the PFB questions I've never had time to look into.
Glen > Hi Simon and Terry, > > Concerning a gpu implementation of a PFB, the following/attached > is additional material. It is from Chris Harris at U. Western Aus. > > Best, > > Lincoln > > -------- Original Message -------- > Date: Thu, 2 Dec 2010 09:57:23 +0800 > From: Christopher Harris <[email protected]> > To: Randall Wayth <[email protected]>, > "[email protected]" > <[email protected]> > > Hi Lincoln and Randall, > > I did indeed. You can find the code in Appendix A.3 of my thesis attached. > Note > that it was only really written for the range of parameters described in > Section > 5.4, and that the polyphase filter function in the kernel code will likely > need > altering (there should be no performance impact). There is also some text > describing the code in section 4.3. > > If the circular buffer does not fit in shared memory, it will need to be > rewritten such that each block does a portion of the filter, split to > minimise > re-reading. > > Hope you find this useful and I'm happy to answer any questions about how > it > works. (Although I'll be on leave for all of next week) > > Chris > > >
