On Tue, Mar 5, 2013 at 3:56 PM, Matt Turner <[email protected]> wrote: > Replaces (samples - 1) ADDs and a MUL with (samples - 1) LRPs. > ---
NAK this. LRP is a 3-src operand instruction, which means 16-wide needs to double the instructions. In the best case, this might be slightly better but we don't do the proper register bank allocation so we'd probably get the worst case, which would be worse than what we have now. Maybe revisit. _______________________________________________ mesa-dev mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/mesa-dev
