On Mon, Mar 18, 2013 at 03:04:34PM -0400, Mark Hahn wrote: > I'm not disagreeing, but wonder if you'd mind talking about SP-friendly > algorithms a bit. I can see it in principle, but always come up short > when thinking about, eg simulation-type modeling with such low-precision
Neural stuff needs not to be high precision, and for spatial coordinates you can use relative addressing (offset to voxel origin). E.g. the current Parallella Epiphany only supports 32 bit floats, and it's just as well, since there are only 32 kByte embedded RAM in each DSP core, which must be shared with code. > numbers. does someone really comb through all the calculations, looking > for iffy stuff? (iffy stuff can be pretty subtle - ordering > and theoretically equivalent substitutions. maybe there are compiler-based > tools that perform this sort of analysis automatically?) > > your mention of precious bandwidth is actually relevant to one of the > earlier threads - that there is an upcoming opportunity for cpu vendors > to integrate decent (much bigger than cache) amounts of wide storage > within the package, using 2.5d integration. if there were enough fast > in-package ram, it would presumably not be worthwhile to drive any > off-package ram - any speculation on that threshold? _______________________________________________ Beowulf mailing list, Beowulf@beowulf.org sponsored by Penguin Computing To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
