On Sunday, January 06, 2013 06:49:46 AM Troy Benjegerdes wrote: > I think I got most of this, but I appear to have gone the wrong > direction with the multiple phases. I thought I could remove the > 'p1' signals, but this seems broken.
Since you are still multiplying the SDRAM clock for commands by 2, you still have the 2 phases p0 and p1. You only need to touch the data serialization ratio (e.g. replace 1:4 IOSERDES with IODDR). The command path stays the same. > with: > https://bitbucket.org/dahozer/xulamist/commits/b6343bdc8613f1ea1723e04c6bb7 > e17056db4ad8 Your tREFI is wrong - 64ms is for a single row, but there are lots of them in a SDRAM. Sebastien _______________________________________________ http://lists.milkymist.org/listinfo.cgi/devel-milkymist.org IRC: #milkymist@Freenode