Hi Shilpa and Glen, The casper list is probably more appropriate (or at least more useful) for this discussion, since the Parkes spectrometer is a CASPER design. I have run into similar problems where a small design change makes the design not route properly or meet timing. I'm not sure what the solution is, other than just tweaking things until it works. One suggestion, have you tried building the original Parkes design on your build machine without changing anything? It's possible the libraries you are using are of a slightly different version and this can impact the resource usage. Glenn
On Mon, May 11, 2009 at 1:32 PM, Glen Langston <[email protected]> wrote: > > Hello, > > We (Shilpa Bollineni and Glen Langston) have been working with > the 512 channel Parkes spectrometer design, which uses I and Q > inputs with a ADC board plugged into ADC 0 slot on the IBOB. > (The design also uses 10GbE port 0 in the IBOB.) > > One IBOB’s we are working with has a bad "Q" data when connected to > the ADC 0 slot. Non-zero data appears only if we use the ADC 1 > slot on this IBOB. The I data for the design looks good. > The same design has been loaded into a different > IBOB and that device yields good results for both polarizations. > > The strange problem we are experiencing is that when we make only > one change, swapping ADC 1 ont ADC 0 in the design. We get a build > error. The error is: > > “There were not enough sites to place all selected components”. > > We have tried several builds by eliminating few components from the > design still it gives the same error. We would assume that the design > has problems in routing to all the components and hence giving us > this error. Does anyone have any suggestions or any experience on how > to fix this problem? > > Thank you, > > Shilpa Bollineni > [email protected] > > Electronics Intern > National Radio Astonomy Observatory > P.O. Box 2 > Rt. 28/92 > Green Bank, WV 24944-0002 > > > > > >
